verification/data-dictionary.md

GameTide: data dictionary

Every published file, what one row means, its key, its columns, how it links to other systems, and what it ties to. Published in full with every snapshot. Every table, its grain, keys, links and tie-outs are described below; the column catalog generated with each snapshot (verification/column_catalog.csv) lists every column of every file with its data type, an example value, how often it is empty and, where a column has 12 or fewer text values, the values seen across six months spread over the data.

Status: Phase A step 6, written 2026-10-17 from the research summaries in research/.

How to read this document

Each file has one block:

• One row: the grain (what a single row represents).
• Key: the column or columns that make a row unique.
• Columns: the column names exactly as published.
• Links: columns that point to other files, and the shared ID crosswalk entry they map to.
• Ties to: the GL accounts or other files this file must reconcile with (checked by the acceptance tests).
• Rows: approximate rows over three years of the scale-up (2024 to 2026). The ten-year totals (2020 to 2029) are about 3 times these, a little more for tables that grow with the customer base; each system starts in the month it is adopted (system-scope.md, When each system starts).
• Based on: the research summary the names come from.

Markers on column names:

• [u]: the research could not confirm this exact name from an official source (it came from search excerpts, third-party sites or memory). It is used as written until verified.
• [G]: GameTide-defined. There is no vendor standard for this column (for example custom fields, company-built worksheets, generic bank files), so GameTide chose the name in the replicated tool's style.

Data types follow the replicated tool: Salesforce and Intacct dates are ISO dates and timestamps; Salesforce currency is decimal dollars; Stripe API amounts and Kantata money fields are integer cents (_in_cents); Stripe report files, Avalara, Zylo and AWS amounts are decimal dollars; Google Ads cost is micros (dollars times 1,000,000). The engine stores every amount internally as integer cents and converts on export.

Naming rules

The full rules are in system-scope.md (File naming). In short:

• One folder per month-end date, then one folder per system named by function: 2024-01-31/crm/.
• File name = system prefix + the replicated tool's own table or object name + the month-end date: crm_opportunity_line_item_2024-01-31.csv. File names below omit the date for readability.
• Activity files hold that month's rows only; status files hold every record as of the month-end date and end with an as_of_date column. Each file entry below says which kind it is where it matters.
• Columns match the replicated tool exactly. Where the tool's on-screen export headings could not be verified, files use the tool's documented API field names (decision of 2026-10-17). Nested API fields are flattened with an underscore (payPeriod.startDate becomes payPeriod_startDate).
• Report-style exports sit in a reports/ subfolder, use the tool's field labels, and have report in the name. CRM reports keep Salesforce labels.
• SQLite table names equal file names without the date and .csv; one SQLite file holds all 120 months.
• Two published layouts (by month, and by table for Hugging Face) contain identical files.
• Names beside IDs. Wherever a file carries the ID of a customer, employee, user, account, department, product, vendor, opportunity, campaign, job, project or similar record, the record's name follows the ID in the next column, the way a report export from the real system shows it. The name column uses the system's own style: Account.Name and Owner.Name in Salesforce, CUSTOMERNAME and ACCOUNTTITLE in Sage Intacct, worker_name in Rippling, Company Name in Gainsight, customer_name in the warehouse. Names are taken from the system's own records, so they match the name shown everywhere else.

Shared IDs and the crosswalk

Each system issues its own IDs, in its own format, as the real tool would. The engine also keeps one master key per entity and publishes the mapping in the warehouse crosswalk files, so any two systems can be joined.

Entity	Master key [G]	IDs in each system
Customer	customer_key (for example GT-C-000123)	Salesforce Account Id (001..., 18 characters); Intacct CUSTOMERID (for example C-10123); QuickBooks customer Id and DisplayName (customers who signed by December 2022); Stripe customer (cus_...); Avalara customerCode (equal to the Intacct CUSTOMERID); Gainsight GS ID; Amplitude account_id group value
Employee	employee_key (for example GT-E-0042)	Rippling worker id and number; Intacct EMPLOYEEID; QuickBooks employee Id (hired by December 2022); Salesforce User Id (005...) for users of the CRM; CaptivateIQ employee_id; Greenhouse application_id of their hire; Carta Employee ID; Tempo author_accountId; Kantata user_id
Product	product_key	Salesforce Product2 Id (01t...) and ProductCode; Intacct ITEMID; Amplitude module event property
Department and cost center	department_code (16 codes)	Intacct DEPARTMENTID; QuickBooks class Id and Name; Rippling department reference_code; Anaplan Departments list; Greenhouse department external_id; Zylo cost_center
GL account	ACCOUNTNO (Intacct)	Anaplan Accounts list uses the same numbers; QuickBooks accounts map through warehouse_crosswalk_gl_account (one QuickBooks account, Legal and Professional Services, splits into three Intacct accounts)
Contract line	Salesforce ContractLineItem Id (811...)	Referenced by the warehouse ARR movements and by Intacct revenue recognition lines (DOCNUMBER)
Vendor	Intacct VENDORID	QuickBooks vendor Id and DisplayName; Zylo supplier_id; card and AP lines in the GL

1 and 2. CRM (crm)

Layout replicated: Salesforce Sales Cloud and Service Cloud, standard Quotes. Object exports use API names, 18-character IDs and ISO dates (Data Loader style). Every object also carries the system fields Id, IsDeleted, CreatedDate, CreatedById, LastModifiedDate, LastModifiedById, SystemModstamp, not repeated below; CreatedBy.Name and LastModifiedBy.Name follow their IDs. A related record's name follows its ID in the relationship style of a Data Loader export (Account.Name after AccountId, Owner.Name after OwnerId). Based on research/salesforce-crm.md and research/sales-marketing-exports.md.

crm/crm_account.csv

• One row: an account (customer, prospect, child school or department, partner). Key: Id. Rows: about 4,000.
• Columns: Name, AccountNumber, OwnerId, Owner.Name, RecordTypeId, ParentId, Parent.Name, Type, Industry, AccountSource, BillingStreet, BillingCity, BillingState, BillingPostalCode, BillingCountry, Phone, Website, LastActivityDate, ARR__c [G], Segment__c [G], Sport__c [G], Region__c [G], Customer_Since__c [G], Churn_Date__c [G], Next_Renewal_Date__c [G], Tax_Exempt__c [G], Payment_Terms__c [G].
• Links: ParentId to crm_account (district to school; university to department; conference to colleges); OwnerId to crm_user; crosswalk customer.
• Ties to: ARR__c equals the sum of active contract lines' annualized value; warehouse ARR by customer.
• Values: Type is Customer, Former Customer, Prospect, Partner (the state association, E21), or the child kinds School, University and Member school. Owners change when an owner leaves: accounts and open deals move to a colleague in the same role within two weeks (Q23).

crm/crm_contact.csv

• One row: a person at an account (status file). Key: Id. Rows: about 10,500 at the end of 2029.
• Columns: AccountId, Account.Name, FirstName, LastName, Title, Department, Email, Phone, MailingState, LeadSource, ReportsToId, ReportsTo.Name, OwnerId, Owner.Name.
• Values: a converted lead's person is the account's first contact (the champion, crm_lead.ConvertedContactId); customers then add an economic buyer, a billing contact and operations staff by segment; each child school or department has its head coach (Q31). Phone numbers are fictional (555-01xx).
• Links: AccountId to crm_account; ReportsToId to another contact (the champion reports to the economic buyer); ContactId on opportunities (the primary contact), service contracts (the billing contact), cases, contact roles and campaign influence (check KEY-08).

crm/crm_lead.csv

• One row: a lead. Key: Id. Rows: about 10,000.
• Columns: FirstName, LastName, Title, Company, Email, State, Status, LeadSource, Rating, OwnerId, Owner.Name, IsConverted, ConvertedDate, ConvertedAccountId, ConvertedAccount.Name, ConvertedContactId, ConvertedOpportunityId, ConvertedOpportunity.Name, MQL_Date__c [G], SQL_Date__c [G], GA_Client_ID__c [G]; marketing automation fields: pi__score__c, pi__grade__c, pi__first_touch_url__c, pi__utm_source__c, pi__utm_medium__c [u], pi__utm_campaign__c [u], pi__utm_content__c [u], pi__utm_term__c [u], pi__campaign__c [u], pi__first_activity__c [u], pi__last_activity__c [u], pi__conversion_date__c [u], pi__pardot_hard_bounced__c.
• Links: Converted*Id to account, contact and opportunity; pi__utm_* values join to web_analytics and ads source, medium and campaign.

crm/crm_opportunity.csv

• One row: an opportunity (New Business, Expansion, Renewal or Contraction). Key: Id. Rows: about 4,000.
• Columns: AccountId, Account.Name, Name, StageName, Amount, Probability, ExpectedRevenue, CloseDate, Type, LeadSource, NextStep, ForecastCategoryName, IsClosed, IsWon, CampaignId, Campaign.Name, Pricebook2Id, Pricebook2.Name, ContactId, FiscalYear, FiscalQuarter, AgeInDays, LastStageChangeDate, PushCount, OwnerId, Owner.Name, ARR__c [G], Net_New_ARR__c [G], Prior_ARR__c [G], Term_Months__c [G], Renewal_Of__c [G], Signed_Date__c [G], Loss_Reason__c [G], Primary_Competitor__c [G], Partner_Account__c [G].
• Values: Type is New Business, Expansion, Renewal or Contraction. StageName follows assumption Q1. Loss_Reason__c uses the shared reason list (assumption B19) plus the new-business reasons (Q9).
• Ties to: Amount is total contract value; ARR__c is the FP&A metric (policies doc CRM convention).

crm/crm_opportunity_history.csv

• One row: a stage, amount or close date change on an opportunity. Key: Id. Rows: about 20,000.
• Columns: OpportunityId, Opportunity.Name, StageName, Amount, PrevAmount, CloseDate, PrevCloseDate, Probability, ForecastCategory.

crm/crm_opportunity_line_item.csv

• One row: a product on an opportunity. Key: Id. Rows: about 10,000.
• Columns: OpportunityId, Opportunity.Name, PricebookEntryId, Product2Id, Product2.Name, ProductCode, Quantity, ListPrice, UnitPrice, Discount, TotalPrice, ServiceDate.

crm/crm_opportunity_contact_role.csv

• One row: a contact's role on an opportunity (status file). Key: Id. Columns: OpportunityId, Opportunity.Name, ContactId, Contact.Name, Role, IsPrimary [u].
• Values: every contact at the account by the close date, with a role by title: Decision Maker (the champion, primary), Economic Buyer, Influencer (billing), Technical Buyer, Evaluator (Q31).

crm/crm_product2.csv, crm/crm_pricebook2.csv, crm/crm_pricebook_entry.csv

• One row: a product; a price book (one per year from 2024); a product's price in a price book. Keys: Id.
• Columns: Product2: Name, ProductCode, Family, IsActive, Description, Tier__c [G], Module__c [G]. Pricebook2: Name, IsActive, IsStandard, Valid_From__c [G], Valid_To__c [G]. PricebookEntry: Pricebook2Id, Pricebook2.Name, Product2Id, Product2.Name, UnitPrice, IsActive, UseStandardPrice.
• Ties to: the price book in company-profile.md; Intacct items through the product crosswalk.

crm/crm_quote.csv, crm/crm_quote_line_item.csv

• One row: a quote; a quote line (status files). Keys: Id. Rows: about 3,200 quotes and 27,000 lines at the end of 2029.
• Columns: Quote: OpportunityId, Opportunity.Name, Name, Status, ExpirationDate, Subtotal, Discount (percent), TotalPrice, GrandTotal, IsSyncing [u]. QuoteLineItem: QuoteId, Quote.Name, PricebookEntryId, Product2Id, Product2.Name, Quantity, UnitPrice, Discount, TotalPrice.
• Values: new-business deals get a quote when they reach Proposal and, on 60% of deals, a second version at Negotiate with the final discount (Q32). Status is Presented while open; the final version becomes Accepted when the deal is won and Denied when lost; an earlier version is Denied once replaced. IsSyncing marks the final version.
• Ties to: an accepted quote's GrandTotal equals the opportunity's line items (check QUO-01).

crm/crm_service_contract.csv

• One row: a subscription contract (one per signed term; a renewal is a new row). Key: Id. Rows: about 1,500.
• Columns: Name, ContractNumber, OwnerId, Owner.Name, AccountId, Account.Name, ContactId, Pricebook2Id, Pricebook2.Name, ParentServiceContractId, RootServiceContractId, StartDate, EndDate, ActivationDate, Term, Subtotal, Discount, TotalPrice, Tax, GrandTotal, LineItemCount, Status, ApprovalStatus, SpecialTerms, ARR__c [G], Auto_Renew__c [G], Billing_Frequency__c [G], Payment_Terms__c [G], Signed_Date__c [G], Opportunity__c [G], End_Reason__c [G], Termination_Type__c [G], Termination_Notice_Date__c [G], Suspended_Date__c [G], Termination_for_Convenience__c [G].
• Values: Termination_Type__c is Customer cancellation, Termination for convenience, Non-appropriation or Non-payment (policy 25). End_Reason__c uses the shared reason list (assumption B19). Billing_Frequency__c is Monthly, Quarterly, Semiannual or Annual (C2). In the core engine, early ends are Non-payment for monthly card clubs after repeated declines (G4) or when a club folds (N21, reason Team or club folded). Suspended_Date__c is set on active contracts of a customer with an invoice open more than 90 days past due (C17), from the 91st day, and clears when the customer pays. Termination_for_Convenience__c is true on contracts with a convenience or non-appropriation clause (C8).
• Links: ParentServiceContractId to the contract it renews; RootServiceContractId to the first contract in the chain.

crm/crm_contract_line_item.csv

• One row: one team, module or seat block for one contract year. Key: Id. Rows: about 12,000.
• Columns: ServiceContractId, ServiceContract.Name, LineItemNumber, PricebookEntryId, Product2Id, Product2.Name, AssetId, Quantity, ListPrice, UnitPrice, Discount, Subtotal, TotalPrice, StartDate, EndDate, Status, ParentContractLineItemId, RootContractLineItemId, ARR__c [G], Is_Recurring__c [G], Team__c [G], Sport__c [G], Seasonal__c [G], Paused_From__c [G], Paused_To__c [G], Free_Months__c [G], End_Reason__c [G].
• Raw facts, not labels: escalators are a new line per contract year; expansions are new co-termed lines; decommissions end a line early or drop it at renewal; product swaps end one line and start a linked one on the same date. Categories live in warehouse_arr_movements.
• Ties to: Intacct invoices and revenue recognition lines reference the line Id.

crm/crm_campaign.csv, crm/crm_campaign_member.csv, crm/crm_campaign_influence.csv

• One row: a campaign; a lead or contact in a campaign; a campaign's credit on an opportunity under one attribution model. Keys: Id.
• Columns: Campaign: Name, ParentId, Parent.Name, Type, Status, StartDate, EndDate, BudgetedCost, ActualCost, NumberOfLeads, NumberOfConvertedLeads, NumberOfResponses, NumberOfOpportunities, NumberOfWonOpportunities, AmountAllOpportunities, AmountWonOpportunities, UTM_Campaign__c [G]. CampaignMember: CampaignId, Campaign.Name, LeadId, Lead.Name, ContactId, Status, HasResponded, FirstRespondedDate. CampaignInfluence: CampaignId, OpportunityId, Opportunity.Name, ContactId, ModelId, Influence, RevenueShare.
• Status files (built in step 2). Campaigns: program parents (Paid Media, Events and Clinics, Email Nurture, Outbound Prospecting, Referral Programs, Website), seven evergreen ad campaigns, yearly programs, quarterly SDR outbound, and dated events (coaching clinics January to April by region, the June athletic director conference, fall regional showcases, webinars). Every lead is a member of its first-touch campaign. Influence has three models (ModelId): first touch, last touch and even distribution, on every opportunity sourced from a lead.
• Ties to: UTM_Campaign__c matches the ad files' campaign names; ActualCost to date on ad campaigns equals ad spend to date in the ad platforms (check CMP-01), and a month's event campaigns equal that month's events bills (6220, check CMP-02); influence revenue shares add up to the opportunity Amount under each model (CMP-03). crm_opportunity.CampaignId is the first-touch campaign.

crm/crm_task.csv, crm/crm_event.csv

• One row: a task (call or email, including sales engagement sequence steps) or an event (meeting, demo, business review), in the month it happened (activity files). Keys: Id. Rows: about 307,000 tasks and 27,000 events over the ten years (Q14).
• Columns: Task: Subject, ActivityDate, Status, Priority, Type, TaskSubtype, WhoId, Who.Name, WhatId, What.Name, AccountId, Account.Name, OwnerId, Owner.Name, CompletedDateTime, CallDurationInSeconds, Sequence_Name__c [G] (sales engagement, system 6). Event: Subject, StartDateTime, EndDateTime, DurationInMinutes, Type (Meeting, Demo, Business Review), WhoId, Who.Name, WhatId, What.Name, AccountId, Account.Name, OwnerId, Owner.Name.
• Values: leads get a sequence (Outbound prospecting or Inbound follow-up) with WhoId the lead; deal touches have WhoId the primary contact and WhatId the opportunity; customer success check-ins and quarterly business reviews have WhatId the account. Won new business carries about twice the touches of lost; customer success touches halve in the 120 days before a churn (Q14).
• Links: WhoId to crm_contact or crm_lead; WhatId to crm_opportunity or crm_account (check KEY-09). crm_account.LastActivityDate is the latest activity on the account.

crm/crm_user.csv, crm/crm_user_role.csv, crm/crm_territory.csv, crm/crm_quota.csv, crm/crm_forecast_submission.csv

• One row: a CRM user; a role in the hierarchy (one per department, under Executive); a territory (segment band by region, 12); a quota for one rep for one quarter; a rep's weekly forecast call in one category.
• Columns: User: Name, Email, UserRoleId, ManagerId, Manager.Name, IsActive, Title, Department, Employee_Number__c [G]. UserRole: Name, ParentRoleId, ParentRole.Name. Territory [G]: Name, Segment__c, Region__c, OwnerId, Owner.Name. Quota [G]: OwnerId, Owner.Name, Period_Start__c, Period_End__c, Quota_Type__c (New ARR for account executives, Expansion ARR for customer success managers), Quota_Amount__c (the quarter, after ramp), Annual_Quota__c, Ramp_Percent__c (Q18). Forecast submission [G]: OwnerId, Owner.Name, Period_Start__c (the quarter), Forecast_Category__c (Closed, Commit, Best Case, Pipeline), Amount__c (ARR), Submitted_Date__c (each Monday).
• Kinds: users, roles, territories and quotas are status files; forecast submissions are an activity file (that month's Mondays).
• Values: the four forecast numbers nest: closed is won ARR in the quarter to date; commit adds the rep's judgment of open commit deals (85% to 110%, Q33); best case and pipeline add those categories in full. Comparing a quarter's Monday commits with its final closed ARR measures forecast accuracy.
• Note: Salesforce territory and forecasting objects vary by edition and setup, so territories, quotas and forecast submissions use GameTide-defined custom objects in Salesforce style.
• Ties to: forecast categories nest (FC-01); the last submission's closed amount equals won new business and expansion ARR in the quarter (FC-02).
• Links: Employee_Number__c to the employee crosswalk (Rippling number); UserRoleId to crm_user_role.

Reports (crm/reports/)

Report-style exports with Salesforce field labels, 15-character IDs and US date format:

• crm_report_accounts_with_arr.csv: Account Name, Account ID, Account Owner, Type, Segment, Region, Billing State, ARR, Customer Since, Next Renewal Date.

3. Marketing automation (marketing)

Layout replicated: Salesforce Account Engagement. Score, grade and UTM fields sit on the CRM Lead (above). Based on research/sales-marketing-exports.md.

marketing/marketing_list_email_stats.csv

• One row: one list email send (activity file). Key: list_email_id [G]. Rows: about 6 to 8 a month: the customer newsletter, three prospect nurture steps, event invitations in event months, and a fall readiness send from May to July.
• Columns: list_email_id [G], name [G], campaign_id, sent_at [u], sent, delivered, opens, unique_opens, total_clicks, unique_clicks, soft_bounced, hard_bounced, opt_outs, spam_complaints, delivery_rate, opens_rate, click_through_rate, unique_click_through_rate, click_open_ratio, opt_out_rate, spam_complaint_rate. API v4 names; only unique_clicks, soft_bounced and spam_complaints are confirmed from official docs, the rest from a connector mirror.

marketing/marketing_prospect.csv

• One row: a prospect (lifetime summary). Key: id. Rows: about 10,000.
• Columns: id, email, score, grade, campaign_id, campaign_name, first_activity_at, last_activity_at, source, created_at, crm_lead_fid [u], touch_count [G], visit_count [G].
• Status file. One prospect per CRM lead; campaign_id is the Salesforce campaign Id (connected campaigns). The lead's pi__utm_*, pi__first_touch_url__c, pi__campaign__c and activity dates come from the same record.
• Links: crm_lead_fid to crm_lead.Id (check KEY-04).

4. Website analytics (web_analytics)

Layout replicated: Google Analytics 4 Data API names. Based on research/sales-marketing-exports.md.

web_analytics/web_analytics_traffic_acquisition.csv

• One row: a month, channel group, source / medium and campaign. Key: yearMonth, sessionDefaultChannelGroup, sessionSourceMedium, sessionCampaignName. Rows: about 5,000.
• Columns: yearMonth, sessionDefaultChannelGroup, sessionSourceMedium, sessionSource, sessionMedium, sessionCampaignName, sessions, totalUsers, newUsers, engagedSessions, engagementRate, bounceRate, averageSessionDuration, eventCount, keyEvents, sessionKeyEventRate.
• Activity file, about 15 rows a month. Paid rows carry the ad campaign's UTM name; organic, direct, referral and social rows carry GA4's (organic), (direct) and (referral).
• Ties to: keyEvents (form fills and demo requests) roughly to CRM leads with the same source, medium and campaign in the month (about 1.1 to 1.3 times, since some form fills are existing prospects); paid sessions to ad clicks (check WEB-02); bounceRate + engagementRate = 1 on every row (WEB-01); user counts do not add across rows.

5. Ad platforms (ads)

Layout: two levels. Based on research/sales-marketing-exports.md.

Judgment call: the scope asked for one combined monthly table, but no platform shares column names with another (Google reports cost in micros, Meta has no single conversions column, LinkedIn returns campaign IDs only). So the combined file uses Google Analytics' documented cost data import layout, a real file analysts build, which joins to GA4 sessions and to the UTM fields on CRM leads. Each platform also gets a monthly native file in its own API names, because those carry conversions and are what tie to the platform's bills in the GL.

ads/ads_cost_data_import.csv

• One row: a campaign on one day. Key: utm_id, date. Rows: about 25,000.
• Columns: utm_id, utm_campaign, utm_source, utm_medium, date, impressions, clicks, cost.
• Ties to: utm_source, utm_medium and utm_campaign join to GA4 sessionSource, sessionMedium, sessionCampaignName and to pi__utm_* on CRM leads; the sum of cost by platform equals the three native files.

ads/ads_google_campaign_monthly.csv

• One row: a Google Ads campaign in a month. Key: campaign.id, segments.month.
• Columns: customer.id, campaign.id, campaign.name, campaign.advertising_channel_type, campaign.status, segments.month, metrics.impressions, metrics.clicks, metrics.cost_micros, metrics.conversions, metrics.conversions_value.

ads/ads_meta_campaign_monthly.csv

• One row: a Meta campaign in a month. Key: campaign_id, date_start.
• Columns: account_id, campaign_id, campaign_name [u], date_start, date_stop, impressions, clicks, inline_link_clicks [u], spend, account_currency [u], actions_lead [G] (the actions list flattened to the lead action type).

ads/ads_linkedin_campaign_monthly.csv and ads/ads_linkedin_campaigns.csv

• One row: a LinkedIn campaign in a month; a LinkedIn campaign. Keys: pivotValues, dateRange_start; id.
• Columns: monthly: pivotValues (campaign URN), dateRange_start, dateRange_end, impressions, clicks, landingPageClicks, costInUsd, externalWebsiteConversions, oneClickLeads. Campaigns: id [u], name [u] (the lookup the API requires for campaign names).
• Ties to (all native files): spend by platform to GL paid media (6210) with a one-month billing timing difference: a month's spend is the bill posted the next month (assumption F7, check ADS-02). The daily cost import sums to the native files (ADS-01). December 2028 spend is billed in January 2029, after the generated end, so it has no GL bill in the data.

6. Sales engagement

No files of its own. Sequence steps are CRM tasks with Sequence_Name__c and TaskSubtype.

7. Commissions (commissions)

Layout replicated: CaptivateIQ. Commission worksheets are company-built in CaptivateIQ, so the worksheet columns are GameTide-defined in CaptivateIQ's Title Case style. Based on research/sales-marketing-exports.md.

commissions/commissions_payout_summary.csv

• One row: a payee for one plan and payout date. Key: Employee ID, Plan, Payout Date. Rows: about 2,500.
• Columns: Employee ID [u], Employee Email, Employee Name [u], Plan [u], Period Group [u], Payout Date [u], Payout [u], Adjustments [u], Adjusted Payout, Currency [u].
• Activity file by payout date. Payout is the commission due on the prior month's deals; Adjustments are clawbacks recovered (negative).
• Ties to: Adjusted Payout equals commission earnings in the payroll run lines (check COM-01); accrued commissions (2050).

commissions/commissions_plan_worksheet.csv

• One row: a payee's credit on one deal. Key: Employee ID, Opportunity ID, Deal Type. Rows: about 5,000.
• Columns [G]: Employee ID, Employee Email, Employee Name, Plan, Period Group, Payout Date, Opportunity ID, Account Name, Opportunity Name, Close Date, Deal Type, Commissionable Amount, Quota, Attainment %, Rate, Accelerator, Commission Earned, Clawback, Capitalize, Amortization Months, Payout, Adjustments, Adjusted Payout.
• Activity file: deal rows in the month the deal closes; clawback rows (Deal Type = Clawback) in the month each amount is recovered from a payout. Quota is the payee's quarterly quota and Attainment % the payee's quarter-to-date credit after this deal; Accelerator is true when the deal takes the payee above 100% (Q26), and the part of the deal above quota earns 1.5 times the rate. Renewals earn 2% and do not count toward quota. A commission forfeited because the payee left before the payout shows Adjustments equal to minus the payout. Clawbacks (Q27) apply when the customer churns within 182 days of a new or expansion deal; amounts the payee's later payouts cannot cover are waived.
• Ties to: Opportunity ID to crm_opportunity.Id (KEY-05); Commission Earned to commissions accrued in 2050 (COM-02); clawbacks to 2050 against 1320 (COM-03); capitalized amounts to deferred commissions (1320) and amortization to commissions expense (6030).

8. E-signature

No files of its own. Signed_Date__c on CRM opportunities and service contracts.

9. Accounting (accounting, from January 2023; accounting_qbo, 2020 to 2022)

GameTide keeps its books in QuickBooks Online from the founding to December 31, 2022 and in Sage Intacct from January 1, 2023 (E38). Both are exports of the same ledger, so a QuickBooks line and the Intacct line it would have been carry the same amount, date, description, customer, vendor, employee and department; only the names and IDs differ, and the warehouse crosswalks join them.

Sage Intacct (accounting, from January 2023)

Layout replicated: Sage Intacct (GLDETAIL view and get_trialbalance). Based on research/intacct-gl.md.

accounting/accounting_gl_detail.csv

• One row: one GL line. Key: RECORDNO [u]. Rows: about 150,000 to 250,000.
• Columns: batch: BATCH_DATE, BATCH_NO, BATCH_TITLE, SYMBOL, BOOKID, REFERENCENO, ADJ, MODULEKEY; line: LINE_NO, ENTRY_DATE, TR_TYPE, DOCUMENT, ACCOUNTNO, ACCOUNTTITLE, DESCRIPTION, STATISTICAL; amounts: DEBITAMOUNT, CREDITAMOUNT, AMOUNT, CURRENCY; dimensions: DEPARTMENTID, LOCATIONID, CUSTOMERID, VENDORID, EMPLOYEEID, PROJECTID, ITEMID, CLASSID and their name columns; source document: DOCNUMBER, RECORDTYPE, RECORDID, CUSTENTITY, VENDENTITY, WHENDUE, WHENPAID, TOTALENTERED, TOTALPAID, TOTALDUE; status: STATE, CLEARED, CLRDATE, CREATEDBY, WHENCREATED; GameTide additions: CONTRACT_LINE_ID [G] (on revenue lines), TAX_JURISDICTION [G] (on tax lines), APPLIES_TO_DOCNUMBER [G] (on payments, credits and write-offs).
• Values: SYMBOL is GJ, ARJ, APJ, CMJ, PYRJ, EEJ, RRJ or STJ (statistical). Card transactions are EEJ lines; AP bills and payments APJ; invoices, payments and credit memos ARJ.
• When a line appears: WHENCREATED is when the entry became available in the real system: the posting date for daily transactions, the close date (business day 8 of the next month, business day 5 from the July 2027 close, E22) for month-end entries, and the posting date in March for audit adjustments. A month's folder holds its own entries available by its release date (about the 15th of the next month) plus any earlier-dated audit adjustments posted during the month (ADJ = true, BATCH_DATE = December 31 of the audited year, P9, E5).
• Credit memo reasons (DESCRIPTION): Price concession (4910), SLA credit (4920, E10) and Credit loss (a write-off against the allowance, 1110).
• Ties to: the trial balance; AR and AP aging rebuilt from documents; bank (1010, 1020); Stripe clearing (1030); sales tax (2100); payroll; every sub-ledger roll-forward.

accounting/accounting_trial_balance.csv

• One row: a GL account and department for one month. Key: REPORTINGPERIOD [u], GLACCOUNTNO [u], DEPARTMENTID [u]. Rows: about 30,000.
• Columns: REPORTINGPERIOD [u], GLACCOUNTNO, ACCOUNTTITLE [u], STARTBALANCE [u], DEBITS [u], CREDITS [u], ADJDEBITS [u], ADJCREDITS [u], ENDBALANCE [u], REPORTINGBOOK [u], CURRENCY [u], LOCATIONID, LOCATIONNAME [u], DEPARTMENTID, DEPARTMENTNAME [u]. The research confirms the content of these columns; the exact API names were not captured.
• Ties to: debits equal credits every month; end balances equal the sum of GL detail.
• Start balances: last month's end balances. January start balances close the prior year's income statement (accounts 4000 to 8999) into the accumulated deficit (3030). Audit adjustments to a closed year enter the start balances of the month they are posted (the income statement part through 3030), so the December trial balance exists before and after audit. In January 2023 the start balances are the conversion journal: one batch dated December 31, 2022 with REFERENCENO CONV-QBO-2022 and title "Conversion from QuickBooks Online: balances at December 31, 2022", carrying open invoices (with DOCNUMBER, WHENDUE and TOTALENTERED), deferred revenue (by invoice and contract line) and open bills document by document, and every other balance by account and department. It appears in the January 2023 GL detail like an earlier-dated batch.

accounting/accounting_periods.csv [G]

• One row: a month closed in Intacct, from January 2023 (status file). Key: PERIOD.
• Columns: PERIOD, START_DATE, END_DATE, CLOSE_DATE, RELEASE_DATE, STATUS, as_of_date.
• Values: CLOSE_DATE is business day 8 of the next month, business day 5 from the July 2027 close (A9, E22); RELEASE_DATE is the reporting date, the 15th of the next month (A3); STATUS is Closed once the month's close date has passed as of the folder's month.

accounting/accounting_chart_of_accounts.csv

• One row: a GL account (112, including 2085 Employee Reimbursements Payable for expense reports, J12). Key: ACCOUNTNO.
• Columns: ACCOUNTNO, TITLE, ACCOUNTTYPE, NORMALBALANCE, CLOSINGTYPE, CATEGORY, STATUS [u], REQUIREDEPT [u], REQUIRECUSTOMER [u], PL_GROUP [G].

accounting/accounting_departments.csv, accounting/accounting_customers.csv, accounting/accounting_vendors.csv, accounting/accounting_employees.csv

• One row: a department (16); a customer; a vendor; an employee. Projects appear on GL lines (PROJECTID, PROJECTNAME).
• Columns: departments: DEPARTMENTID, DEPARTMENTNAME, TITLE, PARENTID, PARENTNAME [u], STATUS [u], PL_GROUP [G]. customers: CUSTOMERID, NAME, TERMNAME [u], TAXABLE [u], STATUS [u], SEGMENT [G] (one of the five segments, or Reseller partner for the state association, E21, which is the customer of record for its schools). vendors: VENDORID, NAME, TERMNAME [u], FORM1099TYPE [u], STATUS [u], CATEGORY [G]. employees: EMPLOYEEID, NAME [u], DEPARTMENTID, STATUS [u].

accounting/accounting_ar_invoices.csv, accounting/accounting_ar_invoice_lines.csv

• One row: an AR invoice (status file: every invoice open at month-end, issued this month or paid this month); an invoice line (activity file: lines of this month's invoices, the sales tax line last). Keys: RECORDNO; RECORDKEY and LINE_NO.
• Columns: invoices: RECORDNO, RECORDID (the invoice number in the GL DOCNUMBER), CUSTOMERID, CUSTOMERNAME, WHENCREATED, WHENDUE, WHENPAID, TERMNAME, TOTALENTERED, TOTALPAID (payments and credits applied), TOTALDUE, STATE (Posted, Partially Paid, Paid), CURRENCY, DESCRIPTION (Subscription, Services, Hardware, Usage), CONTRACTID [G]. lines: RECORDKEY, LINE_NO, ACCOUNTNO, ACCOUNTTITLE (2200 for deferred subscription lines, the revenue account for lines recognized at invoice, 2100 for tax), ITEMID, ITEMNAME, AMOUNT, REVRECSTARTDATE, REVRECENDDATE, CONTRACT_LINE_ID [G].
• Ties to: open invoices add up to accounts receivable, 1100 (check AR-04); each month's invoices equal their lines (AR-05). Includes invoices issued in QuickBooks that were still open at the January 2023 conversion.

accounting/accounting_ap_bills.csv, accounting/accounting_ap_bill_lines.csv

• One row: a vendor bill (status file: open at month-end, entered this month or paid this month); a bill line (activity file). Keys: RECORDNO; RECORDKEY and LINE_NO.
• Columns: bills: RECORDNO, RECORDID (the GL DOCNUMBER), VENDORID, VENDORNAME, WHENCREATED, WHENDUE, WHENPAID, TOTALENTERED, TOTALPAID, TOTALDUE, STATE (Posted, Paid), CURRENCY, DESCRIPTION. lines: RECORDKEY, LINE_NO, ACCOUNTNO, ACCOUNTTITLE, DEPARTMENTID, DEPARTMENTNAME, PROJECTID, PROJECTNAME, AMOUNT.
• Ties to: unpaid bills add up to accounts payable, 2010 (AP-01); each month's bills equal their lines (AP-02). Bills are paid on the first Thursday after the due date (J13).

accounting/accounting_fixed_assets.csv

• One row: a fixed asset (status file). Key: ASSETID. Columns: ASSETID, ASSETNAME, CLASSIFICATION, GLACCOUNTNO, ACCUMDEPRACCOUNTNO, DEPARTMENTID, DEPARTMENTNAME, PROJECTID, PROJECTNAME, ACQUISITIONDATE, DEPRECIATIONSTARTDATE, COST, USEFULLIFEMONTHS, DEPRECIATIONMETHOD, MONTHLYDEPRECIATION, DEPRECIATIONTHISPERIOD, ACCUMULATEDDEPRECIATION, NETBOOKVALUE, STATUS (Acquired, In Service, Fully Depreciated).
• Values: laptops (36 months), furniture (60), leasehold improvements (96, or 72 for the 2028 refresh), demo and loaner hardware (36), and capitalized software placed in service (36, accumulated in 1690) (policies 15 and 16). Depreciation starts the month after acquisition; the last month takes the rounding remainder. Fully depreciated assets stay on the register (no disposals are modeled).
• Ties to: cost equals 1510 to 1540 and 1620 (FA-01); accumulated depreciation equals 1590 and 1690 (FA-02).

accounting/accounting_prepaid_schedule.csv, accounting/accounting_accrual_schedule.csv

• One row: a prepaid item with a balance or amortization this month (status file); a month-end accrual line (activity file). Keys: SCHEDULEID; ACCRUALID.
• Columns: prepaids: SCHEDULEID, VENDORID, VENDORNAME, DESCRIPTION, PREPAIDACCOUNTNO (1300), EXPENSEACCOUNTNO, EXPENSEACCOUNTTITLE, DEPARTMENTID, DEPARTMENTNAME, POSTINGDATE, AMORTIZATIONSTARTDATE, TERMMONTHS, ORIGINALAMOUNT, AMORTIZEDTHISPERIOD, AMORTIZEDTODATE, REMAININGBALANCE. accruals: ACCRUALID, DESCRIPTION, ACCOUNTNO, ACCOUNTTITLE, DEPARTMENTID, DEPARTMENTNAME, LIABILITYACCOUNTNO (2020), AMOUNT, ACCRUALDATE, REVERSALDATE.
• Values: annual software subscriptions (J5) and insurance (J8) amortize over 12 months; the 2025 missed prepaid appears once the audit restores it (H15). Accruals are booked at month-end and reversed on the first day of the next month (policy 20). Audit adjustments (H15) appear from the close they are booked in: the 2025 missed prepaid shows from March 2026; the 2024 audit accrual, booked and released in March 2025, is in the GL only.
• Ties to: remaining prepaid balances equal 1300 (PP-01); accruals open at month-end equal 2020 (ACR-01).

accounting/accounting_lease_schedule.csv

• One row: a month of the office lease (activity file). Key: PERIOD. Columns: PERIOD, VENDORID, VENDORNAME, LEASEPAYMENT, STRAIGHTLINECOST, INTERESTACCRETION, ROUAMORTIZATION, DISCOUNTRATE, LIABILITYENDBALANCE, CURRENTPORTION, NONCURRENTPORTION, ROUENDBALANCE.
• Values: an operating lease under ASC 842 (policy 23): straight-line cost, interest accreting on the liability, the right-of-use asset absorbing the rest; remeasured at the 2028 extension (E27).
• Ties to: liability equals 2400 plus 2410 and the right-of-use asset equals 1700 (LEASE-01).

accounting/accounting_revenue_schedule.csv

• One row: an invoice line's revenue schedule in a month: every deferred line recognized this month or still carrying a balance (activity file). Key: POSTINGDATE, DOCNUMBER, CONTRACT_LINE_ID.
• Columns: POSTINGDATE, DOCNUMBER, CUSTOMERID, CUSTOMERNAME, CONTRACT_LINE_ID, ITEMID, ITEMNAME, REVENUEACCOUNTNO, SCHEDULESTARTDATE, SCHEDULEENDDATE, RECOGNITIONMETHOD (Daily straight line, Percent complete by hours), SCHEDULEDAMOUNT, RECOGNIZEDTHISPERIOD, RECOGNIZEDTODATE, DEFERREDBALANCE.
• Ties to: recognized this period equals the month's revenue recognition journal (REV-01); deferred balances equal deferred revenue, 2200 (REV-02).

QuickBooks Online (accounting_qbo, January 2020 to December 2022)

Layout replicated: QuickBooks Online reports (General Ledger, Trial Balance, A/R Aging Detail, Invoices and Received Payments, with the columns QuickBooks lets a user add) and lists (the API entity fields of Account, Class, Customer, Vendor, Employee and Item). All columns are [G] in the sense that no single official export fixes them; the names are QuickBooks' own report headings and API fields.

accounting_qbo/accounting_qbo_general_ledger.csv (activity)

• One row: one line of a QuickBooks transaction. Key: Transaction ID and the line's position. Rows: about 63,000 over the three years.
• Columns: Transaction ID, Date, Transaction Type, Num, Adj (Yes or No), Name, Customer, Vendor, Employee, Class, Product/Service, Memo/Description, Account, Split (the other side's account, or -Split- when there are several), Debit, Credit, Due Date, Create Date.
• Values: Transaction Type is Invoice, Payment, Credit Memo, Bill, Bill Payment (Check), Check, Deposit, Transfer, Expense (card transactions and bank charges), Credit Card Payment, Sales Tax Payment or Journal Entry. Journal entries are numbered by their source: RR- revenue recognition, PR- the payroll register synced from Rippling, PT- payroll tax remittance, JE- other journal entries. Account is the QuickBooks account name (parent:child for sub-accounts, no account numbers); Class is the department name. Num is the invoice, payment reference, credit memo or bill number.
• Links: Account to accounting_qbo_accounts and warehouse_crosswalk_gl_account; Class to accounting_qbo_classes and warehouse_crosswalk_department; Customer, Vendor and Employee (display names) to the QuickBooks lists and the warehouse crosswalks.
• Ties to: the QuickBooks trial balance rolled forward (QBO-01); every ledger check of the Intacct years, run through the crosswalks (TB, AR, DR, CASH, PAY, KEY and the system checks).

accounting_qbo/accounting_qbo_trial_balance.csv (status)

• One row: a QuickBooks account with a balance at month-end. Columns: Account, Debit, Credit, as_of_date.
• Values: balance-sheet accounts to date; income and expense accounts year to date. QuickBooks closes the year automatically: prior years' profit and loss sits in Retained Earnings.

accounting_qbo/accounting_qbo_invoices_and_received_payments.csv (activity)

• One row: a payment or credit memo applied to an invoice. Columns: Date, Transaction Type, Num, Customer, Applied To Invoice, Amount Applied. Rebuilds receivables invoice by invoice with the general ledger (AR-02).

accounting_qbo/accounting_qbo_ar_aging_detail.csv (status)

• One row: an open invoice at month-end. Columns: Date, Transaction Type, Num, Customer, Due Date, Amount, Open Balance, Past Due (days). Ties to: Accounts Receivable (A/R) in the trial balance (QBO-03); the December 2022 file equals the open invoices in the Intacct conversion journal (MIG-02).

accounting_qbo/accounting_qbo_accounts.csv, _classes, _customers, _vendors, _employees, _items (status)

• Accounts: Id, Name, FullyQualifiedName, AccountType (Bank, Accounts receivable (A/R), Other Current Assets, Fixed Assets, Other Assets, Accounts payable (A/P), Credit Card, Other Current Liabilities, Long Term Liabilities, Equity, Income, Cost of Goods Sold, Expenses, Other Income, Other Expense), AccountSubType, SubAccount, ParentRef, Active, CurrentBalance.
• Classes: Id, Name (the 16 departments), Active.
• Customers: Id, DisplayName (unique across QuickBooks names; a repeat gets " (2)"), CompanyName, BillAddr_CountrySubDivisionCode (state), Taxable, SalesTermRef, Balance, Active, MetaData_CreateTime.
• Vendors: Id, DisplayName, Vendor1099, TermRef, Balance (open bills; MIG-03 in December 2022), Active.
• Employees: Id, DisplayName, HiredDate, ReleasedDate, Active.
• Items (products and services): Id, Name, Sku (the product code), Type (Service or Inventory), IncomeAccountRef, UnitPrice, QtyOnHand (cameras and accessories from March 2022), Active.

The move from QuickBooks to Intacct

• Read the ten years as one ledger by mapping each QuickBooks line through warehouse_crosswalk_gl_account (account name to intacct_accountno, rows marked split in Intacct are the Intacct accounts a single QuickBooks account divides into), warehouse_crosswalk_department (class to department), warehouse_crosswalk_customer (qbo_customer_id), warehouse_crosswalk_vendor and warehouse_crosswalk_employee (qbo_employee_id). The acceptance checks do exactly this (engine/checks_qbo.py).
• Checks: MIG-01 (Intacct start balances on January 1, 2023 equal the QuickBooks trial balance at December 31, 2022 through the crosswalk), MIG-02 (open invoices, one by one), MIG-03 (open bills), MIG-04 (every QuickBooks customer has an Intacct ID and every converted balance keys to one).
• QuickBooks has no statistical accounts; headcount, ARR and customer counts (9010 to 9040) start with Intacct in January 2023, and the warehouse metrics cover every month.

10. Billing, 12. AP automation, 13. Expenses and cards

No files of their own. Invoices, credit memos, vendor bills, AP payments and card transactions are GL lines in accounting_gl_detail with their document numbers, due dates and the document they apply to.

11. Payment processing (payments)

Layout replicated: Stripe. API objects use integer cents; report files use decimal dollars, as Stripe does. Based on research/finance-exports.md.

payments/payments_charges.csv

• One row: a charge attempt (card or ACH Direct Debit). Key: id. Rows: about 7,000 to 9,000.
• Columns (Charge object, API names): id, amount, amount_captured, amount_refunded, balance_transaction, captured, created, currency, customer, customer_name, description, disputed, failure_code, failure_message, invoice, outcome_seller_message, paid, payment_intent, payment_method_details_type, refunded, statement_descriptor, status, card_brand [G flattening]. All Charge field names are [u]: from memory of the API reference, since the Dashboard export headings are user-chosen and undocumented.
• Ties to: invoice and metadata to the Intacct invoice DOCNUMBER; succeeded charges to the payout reconciliation.

payments/payments_payout_reconciliation_itemized.csv

• One row: a balance transaction in a payout. Key: balance_transaction_id. Rows: about 9,000.
• Columns: automatic_payout_id, automatic_payout_effective_at, balance_transaction_id, created, available_on, currency, gross, fee, net, reporting_category, source_id, description, charge_id, customer_id, customer_name, invoice_id, payment_method_type.
• Ties to: the sum of net per payout equals the bank deposit; fee to payment processing fees (5040); clearing (1030) rolls to zero over the payout lag.

14. Sales tax (sales_tax)

Layout replicated: Avalara AvaTax and exemption certificate management (REST API model names). Based on research/finance-exports.md.

sales_tax/sales_tax_transaction_summary.csv

• One row: an invoice's tax by jurisdiction and tax type. Key: documentCode [u], jurisCode, taxType. Rows: about 30,000.
• Columns: documentCode [u], documentDate [u], customerCode, customerName [u], country, region, jurisType, jurisCode, jurisName, taxType, taxName, rateType, taxable, rate, tax, taxCalculated, nonTaxable, exemption, taxCategory [G] (Subscription or Hardware).
• Activity file: one row per invoice dated in the month, one combined state rate (jurisType STA). taxCategory is Subscription (subscriptions and usage), Hardware or Services (services are not taxed). Exempt buyers (schools, colleges, nonprofit clubs, the reseller) show the whole invoice as exemption.
• Ties to: documentCode to Intacct DOCNUMBER; tax to the GL tax lines and 2100 (TAX-01); tax = taxable x rate on every row (TAX-02); taxable + nonTaxable + exemption + tax = invoices posted to AR (TAX-04).

sales_tax/sales_tax_liability_monthly.csv

• One row: a state and month (about 1,800). Key: period [G], region.
• Columns: period [G], country, region, jurisName, taxable, nonTaxable, exemption, rate, tax. The AvaTax portal report headings could not be verified, so the file sums the TransactionSummary model.
• Ties to: the transaction summary, state by state (TAX-03); total tax to the credits to 2100 before remittances.

sales_tax/sales_tax_rates.csv, sales_tax/sales_tax_certificates.csv, sales_tax/sales_tax_customers.csv

• One row: a jurisdiction's rate for a tax category and period; an exemption certificate; a customer known to the tax system.
• Columns: rates: region, jurisName, taxCategory [G], totalRate, effectiveDate [G], endDate [G]. certificates: id, customerCode, customerName [G flattening of customers], signedDate, expirationDate, exemptionNumber, exemptionReason, exposureZoneName, status, valid. customers: customerCode, name, region, postalCode, country.
• Status files. Certificates run four years and are renewed 30 days before they expire.
• Ties to: every exempt sale has a certificate whose dates cover the invoice date (TAX-05); rates equal table G12 in assumptions.md.

15. Bank and treasury (bank)

There is no standard US bank CSV, so these files are GameTide-defined in common bank style, with BAI2 type codes. Based on research/finance-exports.md.

bank/bank_transactions.csv

• One row: a posted transaction in the operating or payroll account. Key: Bank Reference. Rows: about 15,000 to 24,000.
• Columns [G]: Account Number, Posted Date, Value Date, Description, Debit, Credit, BAI Type Code, Transaction Type, Check Number, Bank Reference, Customer Reference, Running Balance.
• Values: BAI Type Code uses the BAI2 codes in the research (for example 142 ACH Credit Received, 195 Incoming Money Transfer, 455 Preauthorized ACH Debit, 495 Outgoing Money Transfer, 698 Miscellaneous Fee).
• Activity file by bank posted date. BAI Type Code values used: 142 ACH Credit Received, 165 Preauthorized ACH Credit (Stripe), 195 Incoming Money Transfer, 251 and 581 book transfers between the two accounts, 301 Commercial Deposit (customer checks), 354 Interest Credit, 455 Preauthorized ACH Debit, 469 ACH Debit Originated, 475 Check Paid, 495 Outgoing Money Transfer, 698 Miscellaneous Fee.
• Ties to: GL cash (1010, 1020) after reconciling items (deposits in transit, outstanding checks); Stripe payouts; payroll funding.

bank/bank_reconciliation.csv [G]

• One row: an account at month-end (activity file). Columns: Account Number, Statement Date, Bank Balance, Deposits in Transit, Outstanding Checks, Adjusted Bank Balance, GL Balance, Difference.
• Ties to: Difference is zero (check BANK-01); GL Balance equals the trial balance (BANK-02); bank balances roll from month to month (BANK-03).

bank/bank_credit_facility_statement.csv and bank/bank_compliance_certificate.csv

• One row: the lender statement for a month; one covenant test for a month.
• Columns [G]: statement: Statement Date, Facility ID, Commitment Amount, Opening Principal Balance, Advances, Repayments, Closing Principal Balance, Available Amount, Interest Rate, Interest Charged, Interest Paid, Unused Line Fee, Next Payment Due Date. certificate: Reporting Period, Covenant, Required Level, Actual Level, Compliant.
• Activity files, one row a month from the facility's opening in March 2024. Interest Paid is the prior month's interest and fee, paid on the first business day.
• Ties to: credit facility (2500, FAC-01), accrued interest (2510), interest expense (7020: interest charged plus the unused line fee, FAC-02); the minimum cash covenant in assumption M2 (FAC-03).

16. Planning (planning)

Layout replicated: Anaplan multiple-column export, one column per list, then one column per line item. Based on research/finance-exports.md.

File	List columns	Line item columns [G]
planning/planning_pl_plan.csv	Versions, Time, Accounts, Departments	Amount
planning/planning_headcount_plan.csv	Versions, Time, Departments, Positions	Headcount, Planned Start Date, Base Salary, Loaded Cost
planning/planning_arr_bookings_plan.csv	Versions, Time, Segments	New ARR, Expansion ARR, Contraction ARR, Churned ARR, Ending ARR, Bookings, New Logos
planning/planning_drivers.csv	Versions, Time, Drivers	Value
planning/planning_cash_plan.csv	Versions, Time, Cash Lines	Amount

• Values: Versions items are GameTide-named: Budget 2025 (published in January), Forecast 3+9 2025, Forecast 6+6 2025 and Forecast 9+3 2025 (published in April, July and October), and LRP 2026-2028 (published in November): five a year, 25 in all. Actuals are in the GL, so there is no Actual version. Time uses month labels such as Jan 25 [u]; the LRP uses FY26 and summary lines (Revenue, Cost of revenue, Operating expenses, department All departments).
• Activity files: each month folder holds the versions published that month; a version is never changed afterwards. Amounts are positive for revenue and for costs.
• Cash lines: Opening cash, Collections, Payroll, Vendors and other, Credit facility, Ending cash (closed months) or Ending cash before facility draws (planned months), and Minimum cash covenant.
• Ties to: Accounts and Departments equal the GL codes (PLAN-01), so actual versus plan lines up; a reforecast's closed months equal the GL (PLAN-02); Positions link to Greenhouse jobs (custom_plan_position_id) and to HR (POS- plus the Rippling worker number for people on staff).

17. Equity (equity)

Layout replicated: Carta (grant fields from Carta's drafting glossary and ledger column names). Based on research/people-exports.md.

equity/equity_grants.csv

• One row: a grant (about 200). Key: Custom ID.
• Columns: Custom ID, Name, Employee ID, Relationship, Type, Original quantity, Exercise price, Board approval date, Grant date, Vesting start date, Vesting schedule, Expiration date, Cost center, Cumulative Vested, Vested Outstanding, Quantity Exercised, Exercised Unvested, Status, Canceled Quantity [u], Forfeited Quantity [u], Expired Quantity [u], Fair Value Per Share [u], then as_of_date (status file).
• Values: Status is Outstanding, Fully Vested, Canceled (left before the cliff) or Expired (vested options not exercised within 90 days of leaving). Grants run from the founding in 2020; no exercises are modeled.
• Links: Employee ID to the employee crosswalk (EQ-02).

equity/equity_sbc_expense.csv

• One row: a grant's expense for one month. Key: Custom ID, Period.
• Columns [u]: Custom ID, Stakeholder Name, Period, Cost center, Expense This Period, Expense To Date, Unrecognized Expense, Forfeiture Adjustment.
• Activity file: grants with expense or a forfeiture adjustment in the month.
• Ties to: stock-based compensation (6060) by department (EQ-01) and APIC (3020).

18. HR (hr)

Layout replicated: Rippling REST API (snake_case; nested objects flattened). Based on research/people-exports.md.

hr/hr_workers.csv

• One row: a person, including contractors (about 250 over the history). Key: id.
• Columns: id, number, user_given_name, user_family_name, user_preferred_given_name, work_email, start_date, end_date, original_start_date, status, title, level_id, department_id, department_name, manager_id, manager_name, is_manager, employment_type_type, employment_type_compensation_time_period, employment_type_amount_worked, overtime_exemption, location_type, location_work_location_id, termination_details_type, termination_details_reason. Names are fictional.
• Values: termination_details_type is VOLUNTARY or INVOLUNTARY (plus Rippling's other values where they apply); termination_details_reason is GameTide's list (for example Resignation, Performance, Reduction in force).

hr/hr_compensations.csv

• One row: a worker's current compensation record. Key: id.
• Columns: id, worker_id, worker_name, annual_compensation, annual_salary_equivalent, on_target_commission, target_annual_bonus, target_annual_bonus_percent, payment_type, salary_effective_date.
• Values: on_target_commission is the commission plan's rate times the year's annual quota for account executives (new ARR, 10%) and customer success managers (expansion ARR, 5%), $20,000 for SDRs, and 0 for everyone else (I13).

hr/hr_worker_changes.csv

• One row: one changed field in one worker change event (job and pay history). Key: id, field_key. Rows: about 3,000.
• Columns: id, worker_id, worker_name, applied_on, effective_on, event_type, reason_code [u values], field_key, field_display_name, from_value, to_value.
• Values: event_type is HIRE, CHANGE or OFFBOARD; reason_code is GameTide's list (Merit, Promotion, Transfer, Market adjustment, Reduction in force and others). Merit increases are CHANGE rows effective April 1 (P6, I15). A promotion to team lead (I25) shows as CHANGE rows with reason Promotion: one for annual_compensation and one event with title and is_manager rows; hr_workers shows the title, level and manager flag as of each month-end.

hr/hr_departments.csv

• One row: a department (16). Columns: id, name, parent_id, parent_name, reference_code (the shared department code).

hr/hr_engagement_survey_results.csv

• One row: a department's result in one engagement survey, plus an All employees row (activity file, April and October from 2022). Key: survey_name, department_code.
• Columns: survey_name, survey_close_date, department_code, group_name, invited, responded, participation_rate, engagement_favorable_pct, enps, manager_favorable_pct, growth_favorable_pct, wellbeing_favorable_pct, alignment_favorable_pct, suppressed [G layout, survey-tool style].
• Values: groups with fewer than 5 responses are suppressed (scores blank), as survey tools do to protect anonymity. Departments that go on to lose people voluntarily in the next six months score lower, and scores fall after the 2025 reduction in force (I24).

19. Payroll (payroll)

Layout replicated: Rippling Payroll API (camelCase). Per-employee arrays are published in long format. Based on research/people-exports.md.

payroll/payroll_runs.csv

• One row: a pay run (about 80 regular plus off-cycle). Key: id.
• Columns: id, name, runState, runType, checkDate, approvedAt, payPeriod_startDate, payPeriod_endDate, payPeriod_paySchedule.

payroll/payroll_run_employees.csv

• One row: an employee in a pay run. Key: run, role. Rows: about 12,000.
• Columns: run, role, employeeName, grossPay, netPay.

payroll/payroll_run_lines.csv

• One row: one earning, tax, deduction or credit line for an employee in a run. Key: run, role, lineType, name. Rows: about 150,000.
• Columns: run, role, employeeName, lineType [G] (earnings, taxes, debits, credits), name, amount, hours, earningTypeEnum [u], isERTax.
• Values: earnings name and earningTypeEnum: Regular Salary (REGULAR), Commission (COMMISSION, paid with the second payroll of the month after the deal closes, Q28), Bonus (BONUS, the off-cycle run nearest March 15, P6) and Severance (SEVERANCE, an off-cycle run after an involuntary exit, I23).
• Values: tax name strings are GameTide's [u]: Federal Income Tax, Social Security, Medicare, FUTA, NC SUI, NC Income Tax.
• Ties to: one PYRJ journal per run by department; salaries (6010), commissions paid (2050), bonus paid (2040), severance (6090), payroll taxes (6040), benefits (6050), withholdings (2060), payroll cash (1020); grossPay = sum of earnings; netPay = gross - employee taxes - deductions.

20. Recruiting (recruiting)

Layout replicated: Greenhouse Harvest v3 (v1 and v2 were retired on August 31, 2026). Based on research/people-exports.md.

recruiting/recruiting_jobs.csv

• One row: a job (requisition, about 200). Key: id.
• Columns: id, name, requisition_id, status, opened_at, closed_at, department_id, external_department_id, department_name, office_ids, custom_plan_position_id [G], custom_new_or_backfill [G], custom_backfill_for_worker_id, custom_backfill_for_worker_name [G], custom_agency_fee [G].
• Status file: one job per hire since the start of the data (opened about 60 days before the offer, I9) plus requisitions closed unfilled (the E13 hiring freeze; sales roles already at plan).
• Links: custom_plan_position_id to Anaplan Positions.

recruiting/recruiting_openings.csv, recruiting/recruiting_offers.csv, recruiting/recruiting_applications.csv, recruiting/recruiting_close_reasons.csv

• One row: a headcount slot on a job; an offer; an application; a close reason.
• Columns: openings: id, job_id, job_name, opening_id, opened_at, closed_at, open, application_id, close_reason_id, target_start_on. offers: id, application_id, job_id, candidate_id, opening_id, status, sent_on, resolved_at, starts_on, custom_base_salary [G]. applications: id, candidate_id, job_id, status, source_id, rejected_at, last_activity_at. close reasons: id, name.
• Links: the hired application_id to the employee crosswalk (Rippling worker).
• Status files. Close reasons: Hired, Hiring freeze, Not backfilled: team at plan, Position cancelled. About 15% of filled jobs had a declined offer first.
• Ties to: agency fees on jobs whose hires start in the month equal recruiting fees billed less job boards (6080, REC-01); openings and offers key to applications and jobs (REC-02).

21. Product usage (product_usage)

Layout replicated: Amplitude, an event segmentation chart grouped by the account_id group type, monthly. Amplitude has no fixed monthly-by-account file, so this reshaped layout is a design proposal using Amplitude's names. Based on research/product-ops-exports.md.

product_usage/product_usage_account_monthly.csv

• One row: an account, month, module and event. Key: month, account_id, module, event_type. Rows: about 40,000 to 50,000.
• Columns: month [G], account_id, account_name, gp:segment [G], gp:plan [G] (the account's highest tier), gp:licensed_seats [G], module [G], event_type, totals, uniques, sums [G] (sum of the event's numeric property: video hours, viewing hours, overage dollars or overage storage hours).
• Values: module is Platform, Video, an add-on module name, Hardware or Usage. event_type is Any Active Event, Session Start, Video Uploaded, Game Logged, Practice Logged, Athlete Tracked, Report Viewed, Scouting Viewed, Broadcast Started, App Error, Device Active, Broadcast Overage, Storage Overage [G names].
• Activity file: about 9 rows per active customer a month (about 3,500 a month). Active users come from licensed seats times the seat activity in assumption K1, which moves with the season of each team's sport and with the customer's engagement (the same engagement drives support cases, the health score and renewal churn, K4).
• Ties to: licensed seats from CRM contract lines, and active users never exceed them (USE-03); overage (broadcasts at $25, storage at $1.50 an hour) equals usage revenue (4020) for the month (USE-01); sessions and errors equal the version file (USE-02); hosting cost drivers.

product_usage/product_usage_account_version_monthly.csv

• One row: an account, month and app version. Columns: month [G], account_id, account_name, version_name, platform, totals_sessions [G], totals_errors [G].
• Ties to: the buggy release (event E10): version 5.0.0 (September 8, 2025) errors on about 2% of sessions until 5.0.3 (October 20, 2025); other versions about 0.5% (K9). Customers move to a new version within about three weeks.

product_usage/product_usage_active_users_daily.csv

• One row: a day and segment (activity file). Key: date, gp:segment. Columns: date, gp:segment, daily_active_users, weekly_active_users (7-day), monthly_active_users (30-day).
• Values: daily use peaks on Fridays and Saturdays (game days) and drops on Sundays; DAU is about 20% and WAU about 55% of MAU (K10).
• Ties to: on a month's last day, monthly_active_users equals the month's unique active users (uniques for Any Active Event) in product_usage_account_monthly for that segment; DAU never exceeds WAU, and WAU never exceeds MAU (check USE-04).

22. Customer success (customer_success)

Layout replicated: Gainsight CS (display names as in Gainsight reports). Based on research/product-ops-exports.md.

customer_success/customer_success_company.csv

• One row: a customer. Key: GS ID.
• Columns: GS ID, Name, ARR, CSM, Stage, Status, Renewal Date, Original Contract Date, Previous CSM, CSM Changed Date, Company Type, Parent Company, NPS, SFDC Account ID [G].
• Status file: every customer, including churned customers (Status Churned).
• Links: SFDC Account ID to crm_account.Id; CSM to the employee crosswalk. ARR equals the account's ARR__c (CS-03).

customer_success/customer_success_account_scorecard_history.csv

• One row: a company's monthly snapshot (about 16,000). Key: Company Id, Snapshot Date, Time Granularity.
• Columns: Company Id, Company Name, Snapshot Date, Time Granularity (Monthly), Overall Score (Company), Overall Label [G], Overall Color [G], Usage [u], Support [u], Engagement [u], Payment [u], Renewal Likelihood [G], Risk Flag [G], Risk Type [G].
• Activity file: one row per active customer per month. Usage = seat activity as a share of the segment's expected activity for the season (40% of expected scores 0, 120% or more scores 100); Support = 100 less points for cases per team, missed service levels and escalations in the last 90 days; Engagement = sessions per active user (2 or fewer scores 0, 10 or more scores 100); Payment = the oldest past-due invoice (current 100, 1 to 30 days 85, 31 to 60 days 60, 61 to 90 days 35, over 90 days 10).
• Ties to: the overall score equals the weighted components in assumption K7 (CS-01) and the color follows the scoring scheme (CS-02); each measure is computed from published data (product usage, support cases, AR from the GL).

customer_success/customer_success_call_to_action.csv and customer_success/customer_success_scoring_scheme.csv

• One row: a CTA; a score band.
• Columns: CTA: Id, Name, Company, Company Name, Type, Status, Priority, Reason, Due Date, Owner, Is Closed, Closed Date, Created Date. scheme: Score, Label, Color, Range From, Range To.
• A risk CTA opens when a company turns red and closes when its score is back to 50 or more.

customer_success/customer_success_survey_response.csv

• One row: an NPS response (activity file, March and October). Key: Response Id [G].
• Columns: Response Id, Company Id, Company Name, Survey Name, Response Date, NPS Score (0 to 10), NPS Category (Promoter 9 to 10, Passive 7 to 8, Detractor 0 to 6).
• Ties to: NPS per survey (promoters less detractors) lands on assumption K8; healthier accounts score higher. customer_success_company.NPS is the company's score at its latest survey.

23. Support (support)

Layout replicated: Salesforce Service Cloud Case, CaseMilestone and Entitlements. Based on research/sales-marketing-exports.md.

support/support_case.csv

• One row: a case (about 1,500). Key: Id.
• Columns: Id, CaseNumber, AccountId, Account.Name, ContactId, AssetId, OwnerId, Owner.Name, Status, Priority, Origin, Type, Reason, Subject, IsClosed, IsEscalated, CreatedDate, ClosedDate, ParentId, ProductId, Product.Name, ServiceContractId, ServiceContract.Name, EntitlementId, SlaStartDate, SlaExitDate, IsStopped, MilestoneStatus, CSAT_Score__c [G], CSAT_Response_Date__c [G], App_Version__c [G].
• Values: Type and Reason are GameTide's software support lists (the Salesforce sample values are for equipment). OwnerId is the support queue (00G...). ContactId and AssetId stay empty until contacts and assets are built.
• Status file: every case created to date, with its status at month-end. About 3 cases per 100 teams a month, more for less engaged customers and three times as many in the six weeks after the E10 release (K5).

support/support_case_milestone.csv

• One row: a milestone on a case (First Response, Resolution Time). Key: Id.
• Columns: Id, CaseId, MilestoneTypeId, StartDate, TargetDate, CompletionDate, IsCompleted, IsViolated, TargetResponseInMins, ElapsedTimeInMins, ActualElapsedTimeInMins.

support/reports/support_report_cases_by_account_month.csv

• One row: an account, case type and month [G layout, Salesforce labels]. Columns: Account ID, Month, Case Type, Cases Opened, Cases Closed, Open at Month End, High Priority, Average First Response (Hours), Average Resolution (Hours), SLA Breaches, Escalations, Average CSAT.
• Ties to: rebuilt exactly from the two files above (SUP-01); every customer with an SLA credit (4920) has a case with a missed service level (SUP-03).

24. Implementation (implementation)

Layout replicated: Kantata OX API (money in cents, time in minutes). Based on research/product-ops-exports.md.

File	One row	Columns
implementation/implementation_workspaces.csv	a project (about 600)	id, title, account_id, account_name, start_date, due_date, effective_due_date, stage, status_key, percentage_complete, budgeted, price_in_cents, budget_used_in_cents, estimated_minutes, minutes_logged, billable_minutes, revenue_recognition_method_name, crm_account_id [G], contract_line_id [G], project_type [G], go_live_date [G], delay_reason [G]
implementation/implementation_stories.csv	a task or milestone	id, workspace_id, workspace_title, parent_id, title, story_type, state, start_date, due_date, time_estimate_in_minutes, budget_estimate_in_cents, fixed_fee, billable
implementation/implementation_time_entries.csv	a time entry (about 20,000)	id, workspace_id, workspace_title, story_id, story_title, user_id, user_name, role_id, role_name, date_performed, time_in_minutes, billable, rate_in_cents, cost_rate_in_cents, currency, approved, is_invoiced
implementation/implementation_roles.csv	a role	id, name

• Workspaces (status file): implementation projects (fixed fee, revenue_recognition_method_name = Percent complete by hours), analyst services (one per customer and year, Time and materials) and hardware installations (Completed contract). estimated_minutes is the estimate at completion; the original estimate is on the stories. Time entries are an activity file, one a week per project.
• Ties to: fixed-fee implementation revenue (4110) = price x minutes logged to date / estimated minutes, less the same through last month (policy 18, IMP-01); analyst hours x rate = 4130 (IMP-02); hardware installation (4150) at completion (IMP-03); user_id to the employee crosswalk.

25. Engineering time (engineering_time)

Layout replicated: Jira Cloud REST API v3 field IDs for work items (Atlassian now says "work item" for issue and "space" for project) and Tempo API v4 for worklogs. Based on research/product-ops-exports.md.

engineering_time/engineering_time_work_items.csv

• One row: a work item (epic, story, task or bug). Key: id.
• Columns: id, key, project_key, summary, issuetype, parent_key, parent_summary, status, statusCategory, priority, labels, components, fixVersions, created [u], resolutiondate [u], product_area [G], phase [G], go_live_date [G].
• Note: parent_key replaces the retired Epic Link. Field IDs keep the old names (issuetype, project) even though the screens say "work type" and "space".

engineering_time/engineering_time_worklogs.csv

• One row: a worklog. Key: tempoWorklogId. Rows: about 40,000.
• Columns: tempoWorklogId, issue_id, issue_key, issue_summary, author_accountId, author_displayName, startDate, timeSpentSeconds, billableSeconds, description, attribute__Capitalizable_ [u], attribute__Phase_ [u], createdAt, updatedAt.
• Links: author_accountId to the employee crosswalk; issue_key to work items.
• Activity file: one worklog a week per person per work item. Work items (status file) are an epic per capitalizable project and, each month, a story under the project in development, a maintenance task and a bug item (the E10 regression fixes from September 2025 to January 2026).
• Ties to: each person's regular pay in the month (payroll) x capitalizable seconds / logged seconds = capitalized labor (6095 credit, 1610 debit) by project (policy 15, ENG-01).

engineering_time/engineering_time_accounts.csv

• One row: a Tempo account mapped to a capitalizable project or to maintenance. Columns: key, name, status, category, intacct_project_id [G].

26. Cloud hosting (cloud_costs)

Layout replicated: AWS Cost and Usage Report 2.0 column names, monthly granularity. Based on research/product-ops-exports.md.

cloud_costs/cloud_costs_cur_monthly.csv

• One row: a month, account, service, usage type, line item type and environment. Rows: about 8,000.
• Columns: bill_billing_period_start_date, line_item_usage_start_date, line_item_usage_end_date, line_item_usage_account_id, line_item_usage_account_name, line_item_product_code, product_servicecode, product_product_family, product_region_code, line_item_usage_type, line_item_operation, line_item_line_item_type, line_item_line_item_description, pricing_unit, pricing_term, line_item_usage_amount, line_item_unblended_rate, line_item_unblended_cost, savings_plan_savings_plan_effective_cost, reservation_effective_cost, pricing_public_on_demand_cost, environment (from resource_tags.user_environment [u prefix]; values prod, staging, dev, some blank).
• Activity file. line_item_line_item_type is Usage (at on-demand rates) or EdpDiscount (the commitment discount from 2025, negative); the net of the two is the cost.
• Ties to: prod and untagged cost to production hosting (5010); staging and dev to non-production hosting (6130) (CLD-01).

cloud_costs/reports/cloud_costs_report_cost_explorer_by_service.csv

• One row: a month, with one column per service (Cost Explorer's transposed layout). Header text [u]. Rebuilt from the CUR file for reshaping practice.

27. Hardware inventory (inventory)

Layout replicated: Sage Intacct inventory control objects. Returns and installed base have no Intacct standard, so they are GameTide-defined. Based on research/finance-exports.md.

File	One row	Columns
inventory/inventory_items.csv	an item (hardware SKUs plus non-inventory items for subscriptions and services)	ITEMID, NAME, STATUS, ITEMTYPE, PRODUCTLINEID, COST_METHOD, STANDARD_COST, BASEPRICE, TAXABLE, ENABLE_SERIALNO, REORDER_POINT, REORDER_QTY, INCOMEACCTKEY, INVACCTKEY, COGSACCTKEY, DEFERREDREVACCTKEY
inventory/inventory_documents.csv	an inventory transaction (receipt, shipment, adjustment, return)	RECORDNO, DOCID, DOCNO, STATE, TOTAL, datecreated, referenceno, customerid, customername [G], vendorid, vendorname [G]
inventory/inventory_document_entries.csv	a line on an inventory transaction	DOCHDRNO, LINE_NO, ITEMID, ITEMNAME, warehouseid, quantity, unit, cost, PRICE, departmentid, departmentname
inventory/inventory_serial_numbers.csv	one tracked unit on a line	DOCHDRNO, LINE_NO, serialno, lotno
inventory/inventory_valuation_monthly.csv	an item and warehouse for a month [G, headings u]	ITEMID, ITEMNAME, warehouseid, period, beginning_quantity, beginning_value, received_quantity, shipped_quantity, adjusted_quantity, ending_quantity, ending_value, average_unit_cost
inventory/inventory_installed_base.csv	a customer, SKU and month [G]	customerid, customername, ITEMID, ITEMNAME, period, units_installed, units_active, units_on_loan
inventory/inventory_returns.csv	a returned unit [G]	rma_number, serialno, lotno, ITEMID, ITEMNAME, customerid, customername, return_date, failure_code, outcome, in_warranty, cost

• Values: DOCID uses GameTide's transaction definitions: Purchase Receipt, Customer Shipment, Warranty Replacement and Demo Transfer [u names; company-configurable in Intacct]. Entry quantity is positive into stock and negative out. Accessories are not serialized. Returned units appear in the returns file only (they are scrapped, not restocked); the bad 2026 sensor lot is LOT-202601-SENS, the first lot of the E17 build (E17).
• Ties to: inventory valuation equals inventory (1200, INV-02); shipments, receipts and warranty replacements equal hardware COGS (5110), purchases into 1200 and warranty reserve use (2300) (INV-03); each replacement has a returned unit (INV-04); demo hardware (1540); the bad 2026 sensor lot (event E17) by lotno.

28. Warehouse (warehouse)

Built by the engine (no vendor). Snake_case names, all GameTide-defined. The warehouse is the answer key: acceptance tests check that every number here equals the number rebuilt from the raw system files.

File	One row	Main columns
warehouse/warehouse_crosswalk_customer.csv	a customer	customer_key, customer_name, salesforce_account_id, intacct_customer_id, stripe_customer_id, avalara_customer_code, gainsight_gs_id, amplitude_account_id, qbo_customer_id
warehouse/warehouse_crosswalk_employee.csv	an employee	employee_key, employee_name, rippling_worker_id, rippling_number, intacct_employee_id, salesforce_user_id, captivateiq_employee_id, greenhouse_application_id, carta_employee_id, tempo_account_id, kantata_user_id, qbo_employee_id
warehouse/warehouse_crosswalk_gl_account.csv	an Intacct account	intacct_accountno, intacct_title, qbo_account_id, qbo_account_name, qbo_account_type, mapping (one to one; primary, split in Intacct; split in Intacct)
warehouse/warehouse_crosswalk_department.csv	a department	department_code, name, intacct_department_id, qbo_class_id, qbo_class_name, rippling_department_id
warehouse/warehouse_crosswalk_vendor.csv	a vendor	intacct_vendor_id, vendor_name, qbo_vendor_id, qbo_display_name, category
warehouse/warehouse_crosswalk_product.csv	a product	product_key, product_name, salesforce_product2_id, product_code, intacct_item_id, amplitude_module
warehouse/warehouse_dim_customer.csv, _dim_employee, _dim_product, _dim_department, _dim_gl_account, _dim_calendar	one member of each shared dimension	keys plus segment, region, state, sport; employee title, band, department (code and name), manager (key and name), dates and status; product family, kind, tier and module; department and P&L group with headcount; account section and normal balance; one row per day with fiscal year and quarter, business day and month-end flags
warehouse/warehouse_arr_movements.csv	one ARR movement	movement_id, customer_key, customer_name, movement_date, month, category, subcategory, reason, arr_change, prior_contract_line_id, prior_product_name, new_contract_line_id, new_product_name, opportunity_id, opportunity_name, migration_program_id, is_correction
warehouse/warehouse_arr_by_customer_month.csv	a customer and month	customer_key, customer_name, month, opening_arr, one column per bridge category, closing_arr, carr, paused_arr (0 in v1.0), suspended_arr (closing ARR of a customer suspended at month-end, C17)
warehouse/warehouse_arr_by_customer_product_month.csv	a customer, recurring product and month	customer_key, customer_name, product_key, product_name, product_kind (tier, seats, module, bundle), month, opening_arr, the seven bridge categories, swap_in, swap_out, closing_arr. Each product's change takes the customer's own categories first; what is left is a shift between products with no ARR change, shown as swap_out on the products given up and swap_in on the products taken (policies.md, ARR by product). Summed over products, every column equals warehouse_arr_by_customer_month (ARR-07); swaps net to zero for each customer (ARR-08); closing ARR by product equals the active CRM contract lines (ARR-09)
warehouse/warehouse_pipeline_monthly.csv	a month-end, owner, opportunity type, stage and forecast category	month, owner_key, owner_name, opportunity_type, stage, forecast_category, open_opportunities, amount, arr, weighted_arr; equals the open CRM opportunities (check AK-05)
warehouse/warehouse_ar_aging_monthly.csv	a customer at month-end	month, customer_key, customer_name, intacct_customer_id, current, d1_30, d31_60, d61_90, over_90, total; rebuilt from GL documents and the payments and credits applied to them; adds up to 1100 (AK-01)
warehouse/warehouse_deferred_revenue_monthly.csv	a customer and month	month, customer_key, customer_name, intacct_customer_id, opening, billed, recognized, closing; closes to 2200 and rolls month to month (AK-02, AK-03)
warehouse/warehouse_headcount_monthly.csv	a department and month	month, department_code, department_name, opening_headcount, hires, exits, closing_headcount (AK-04)
warehouse/warehouse_metrics_monthly.csv	a metric, segment and month	month, metric, segment (All, or a segment for ARR, customers, the bridge and retention), value, unit (usd, ratio, count, number, multiple), definition_id: 63 metrics covering every metric in policies.md (AK-06 to AK-08)
warehouse/warehouse_metric_definitions.csv	a metric	definition_id, metric, unit, definition, source_tables; the same as config/metric_definitions.csv
warehouse/warehouse_events.csv	an event (headline, system adoption, pattern or minor)	event_id, event_date, layer (headline, system, pattern, minor), event_code (E1 to E41, S for a system adoption, P1 to P9, N1 to N21), assumption_id, cause, entity_type, entity_id, entity_name, expected_effect
warehouse/warehouse_cohort_retention_monthly.csv	a signing-year cohort and month	month, cohort_year (the year of the customer's first new-logo movement), cohort_customers, active_customers, starting_arr (new-logo ARR at signing), current_arr, logo_retention, net_arr_retention
warehouse/warehouse_logo_events.csv	a logo event (split or merge, E8, E9, N10)	logo_event_id, event_date, event_type, from_customer_key, from_customer_name, to_customer_key, to_customer_name, note; not ARR movements, so splits and merges are not counted as new or churned logos
warehouse/month_in_review/YYYY-MM.md	one month's written summary	assembled from the events table and the month's metrics
warehouse/warehouse_test_results.csv	an acceptance test for a snapshot	test_id, description, expected, actual, passed

• ARR movements carry the bridge category and reason (policies doc and assumption B19), each linked to the CRM contract lines that caused it, so users can check an ARR bridge built from the CRM line by line.

29. Software and AI spend (software_spend)

Layout replicated: Zylo data dictionary names for software; a generic usage layout for AI tools, with generic model tiers rather than vendor model names. Internal use only. Based on research/product-ops-exports.md.

File	One row	Columns
software_spend/software_spend_applications.csv	an internal tool (about 60)	id, app_name, category, subcategory, status, business_owner_user_email, business_owner_name, business_unit, supplier_id, supplier_name, pricing_model, activity_threshold
software_spend/software_spend_application_stats.csv	a tool and month	application_id, application_name [G], snapshot_month [G], purchased_license_count, user_count, active_user_app_count, percent_utilized_licenses, percent_active_licenses, cost_per_user, renewal_date, primary_cost_center
software_spend/software_spend_contracts.csv	a contract (about 50 to 70)	id, name, application_id, application_name, supplier_name, start_date, end_date, contract_term, billing_frequency, total_contract_value, current_year_contract_value, total_licenses, notification_deadline_date, status
software_spend/software_spend_contract_line_items.csv	a contract line	id, contract_id, contract_name, license_name, quantity, unit_price, total_price, start_date, end_date
software_spend/software_spend_charges.csv	a payment (AP or card)	id, transaction_id, application_id, application_name, payment_date, amount, payment_type, supplier_name, cost_center, email, user_name
software_spend/software_spend_ai_usage_monthly.csv	an employee, AI tool and month [G]	month, supplier_name, user_email, user_name, cost_center, model_tier (large, medium, small), requests, uncached_input_tokens, cached_input_tokens, output_tokens, cost_usd

• Files: applications, contracts and contract lines are status files; application stats, charges and AI usage are activity files. Contracts are one per tool per annual term.
• Ties to: transaction_id to the Intacct AP payment (PAY- plus the bill number); charges equal AP payments to those suppliers (SW-02); software subscriptions (6110) and prepaids (1300); AI usage cost equals AI tools (6120) by department (SW-01).

Published layouts and the verification package

tools/publish.py builds both layouts from one engine run: by_month/ (one zip per month folder and the SQLite file, for GitHub Releases and Kaggle) and by_table/ (each table's CSVs hard-linked from the same files, plus one Parquet file of all months with a leading file_date column, for Hugging Face), and verification/: the assumptions and the settings file the engine reads, the policies, story bible, company profile, system scope, this dictionary, the metric definitions, the answer key and results reports, every event (events_all.csv), every month in review and the acceptance test report (test_report.md, test_results_all.csv). checksums.csv lists a SHA-256 for every published zip and verification file.

Snapshot manifest

manifest.csv (top of each snapshot folder)

• One row: a published file. Columns [G]: path, rows, sha256, snapshot_month, engine_version, seed_id, data_version (clean or with issues).

Not yet set

• Full column descriptions, data types and allowed values: filled in during Phase B, table by table.
• The "with issues" version (deferred): same files under its own top-level folder; the kinds and rates of data problems are set when it is taken up.