Update app.py

app.py

@@ -3,41 +3,45 @@ import numpy as np
 import streamlit as st
 import requests
 
-# ▶️ Source CSV (with or without CPC column)
-SAMPLE_FILE_URL = "https://huggingface.co/spaces/Em4e/seo-b2b-saas-forecasting-tool/resolve/main/sample_gsc_data.csv"
+# ▶️ URL for your sample GSC CSV
+SAMPLE_FILE_URL = (
+    "https://huggingface.co/spaces/Em4e/seo-b2b-saas-forecasting-tool/"
+    "resolve/main/sample_gsc_data.csv"
+)
 
-st.set_page_config(page_title="SEO ROI + Savings Tool", layout="wide")
-st.title("📈 SEO ROI & Savings Forecasting for B2B SaaS")
+st.set_page_config(page_title="SEO ROI & Savings Forecasting", layout="wide")
+st.title("📈 SEO ROI & Savings Forecasting Tool for B2B SaaS")
 
-# — Info section on how it works
+# — Info section explaining the math
 with st.expander("ℹ️ How the app works", expanded=True):
     st.markdown("""
-1. **Load your GSC data** (must include `Impressions`, `Position`, and `CPC`; if CPC is missing we simulate \$0.50–\$3.00).  
-2. **CTR benchmarks** by position map average CTR for positions 1–20.  
+1. **Load your GSC data** (we lowercase all column names on load).  
+   If no `cpc` column is present, we simulate values between \$0.50–\$3.00.  
+2. **CTR benchmarks** by position map an expected click-through rate for positions 1–20.  
 3. **Incremental clicks** =  
      Projected_Clicks – Current_Clicks  
-     • Current_Clicks = Impressions×Current_CTR  
-     • Projected_Clicks = Impressions×Target_CTR  
+     • Current_Clicks = Impressions × Current_CTR  
+     • Projected_Clicks = Impressions × Target_CTR  
 4. **Financials**  
-     • Avoided Paid Spend = Incremental_Clicks×CPC  
+     • Avoided Paid Spend = Incremental_Clicks × CPC  
      • Net Savings vs Paid = Avoided Paid Spend – SEO Investment  
-     • Incremental MRR = Customers×MRR_per_Customer  
+     • Incremental MRR = Customers × MRR_per_Customer  
      • SEO ROI = (Incremental MRR – SEO Investment) ÷ SEO Investment  
 5. **Results**  
-   Top-line metrics + keyword-level table with Impact labels.
+   Top-line metrics + keyword-level table with impact labels.
     """, unsafe_allow_html=True)
 
-# — Sidebar Inputs
+# — Sidebar inputs
 with st.sidebar:
     st.header("🔧 Assumptions & Inputs")
     uploaded_file    = st.file_uploader("Upload GSC CSV", type="csv")
     target_position  = st.slider("Target SERP Position",        1.0, 10.0, 4.0, 0.5)
-    conversion_rate  = st.slider("Conversion Rate (Visitor→Signup %)", 0.1, 10.0, 2.0, 0.1)
-    close_rate       = st.slider("Close Rate (Signup→Customer %)",      1.0, 100.0, 20.0, 1.0)
-    mrr_per_customer = st.slider("MRR per Customer ($)",             10, 1000, 200, 10)
-    seo_cost         = st.slider("Total SEO Investment ($)",       1_000, 100_000, 10_000, 1_000)
+    conversion_rate  = st.slider("Conversion Rate (% → signup)", 0.1, 10.0, 2.0, 0.1)
+    close_rate       = st.slider("Close Rate (% → customer)",    1.0, 100.0, 20.0, 1.0)
+    mrr_per_customer = st.slider("MRR per Customer ($)",         10,    1000, 200, 10)
+    seo_cost         = st.slider("Total SEO Investment ($)",     1_000, 100_000, 10_000, 1_000)
 
-# — Download sample button
+# — Download sample CSV button
 sample_bytes = requests.get(SAMPLE_FILE_URL).content
 st.download_button(
     label="📥 Download sample CSV",
@@ -46,75 +50,132 @@ st.download_button(
     mime="text/csv",
 )
 
-# === Load & prep data ===
+# === Load & normalize CSV ===
 def load_csv():
     try:
-        df = pd.read_csv(uploaded_file) if uploaded_file else pd.read_csv(SAMPLE_FILE_URL)
+        if uploaded_file:
+            df = pd.read_csv(uploaded_file)
+        else:
+            df = pd.read_csv(SAMPLE_FILE_URL)
     except Exception as e:
         st.error(f"Error loading file: {e}")
         return None
 
-    # simulate CPC if missing
-    if 'CPC' not in df.columns:
-        st.warning("No `CPC` column found—simulating CPC values between $0.50–$3.00.")
-        df['CPC'] = np.round(np.random.uniform(0.5, 3.0, len(df)), 2)
+    # lowercase all column names
+    df.columns = [col.lower() for col in df.columns]
+
+    # ensure a 'cpc' column
+    if 'cpc' not in df.columns:
+        st.warning("No `cpc` column found—simulating CPC values between $0.50–$3.00.")
+        df['cpc'] = np.round(np.random.uniform(0.5, 3.0, size=len(df)), 2)
+
     return df
 
 # === Core calculation ===
 def calculate(df):
-    # map required cols (including CPC)
-    cols = {c.lower(): c for c in df.columns}
+    # required columns mapping
+    cols = {c: c for c in df.columns}
     required = {
-        'query':       ['query','keyword','queries'],
+        'query':       ['query', 'keyword', 'queries'],
         'impressions': ['impressions'],
-        'position':    ['position','avg. position'],
-        'cpc':         ['cpc']  # we’ll end up with a lowercase 'cpc' column
+        'position':    ['position', 'avg. position', 'average position'],
+        'cpc':         ['cpc']
     }
     found = {}
-    for k, opts in required.items():
-        for o in opts:
-            if o in cols:
-                found[k] = cols[o]
+    for key, opts in required.items():
+        for opt in opts:
+            if opt in df.columns:
+                found[key] = opt
                 break
-        if k not in found:
-            st.error(f"Missing column: {k}")
+        if key not in found:
+            st.error(f"Missing required column: {key}")
             return None, pd.DataFrame()
 
-    # rename to our standard lowercase names
+    # rename to our standard keys
     df = df.rename(columns={found[k]: k for k in found})
 
-    # CTR benchmarks & filtering omitted for brevity...
-
-    # after you’ve computed Incremental_Clicks:
-    # use direct key access to CPC:
-    df['Avoided_Paid_Spend'] = df['Incremental_Clicks'] * df['cpc']
-
-    total_avoided = df['Avoided_Paid_Spend'].sum()
-    net_savings   = total_avoided - seo_cost
-
-    # … build summary & table …
+    # CTR benchmarks
+    ctr = {i: v for i, v in zip(range(1, 11), [0.25,0.15,0.10,0.08,0.06,0.04,0.03,0.02,0.015,0.01])}
+    ctr.update({i: 0.005 for i in range(11,21)})
+    get_ctr = lambda p: ctr.get(int(round(p)), 0.005)
+
+    # filter positions 5–20
+    df = df[df['position'].between(5, 20)].copy()
+    if df.empty:
+        st.warning("No keywords in positions 5–20.")
+        return None, pd.DataFrame()
+
+    # clicks projections
+    df['current_ctr']     = df['position'].map(get_ctr)
+    df['target_ctr']      = get_ctr(target_position)
+    df['current_clicks']  = df['impressions'] * df['current_ctr']
+    df['projected_clicks'] = df['impressions'] * df['target_ctr']
+    df['incremental_clicks'] = df['projected_clicks'] - df['current_clicks']
+    df = df[df['incremental_clicks'] > 0]
+    if df.empty:
+        st.warning("No positive incremental clicks projected.")
+        return None, pd.DataFrame()
+
+    # conversions → MRR
+    conv  = conversion_rate / 100
+    close = close_rate      / 100
+    df['signups']    = df['incremental_clicks'] * conv
+    df['customers']  = df['signups']            * close
+    df['mrr']        = df['customers']          * mrr_per_customer
+
+    # financials: avoided spend & net savings
+    df['avoided_paid_spend'] = df['incremental_clicks'] * df['cpc']
+    total_avoided            = df['avoided_paid_spend'].sum()
+    net_savings              = total_avoided - seo_cost
+
+    # totals & ROI
+    tot_clicks    = df['incremental_clicks'].sum()
+    tot_signups   = df['signups'].sum()
+    tot_customers = df['customers'].sum()
+    tot_mrr       = df['mrr'].sum()
+    seo_roi_pct   = float('inf') if seo_cost == 0 else ((tot_mrr - seo_cost) / seo_cost) * 100
+
+    summary = {
+        "clicks":    f"{tot_clicks:,.0f}",
+        "signups":   f"{tot_signups:,.1f}",
+        "customers": f"{tot_customers:,.1f}",
+        "mrr":       f"${tot_mrr:,.2f}",
+        "roi":       f"{seo_roi_pct:,.2f}%",
+        "avoid":     f"${total_avoided:,.2f}",
+        "net":       f"${net_savings:,.2f}"
+    }
 
-    # when building your table, again reference the column by key:
-    out = df[['query', 'MRR', 'Avoided_Paid_Spend']].copy()
-    out.columns = ['Keyword', 'Projected Incremental MRR ($)', 'Avoided Paid Spend ($)']
+    # keyword-level table
+    out = df[['query', 'mrr', 'avoided_paid_spend']].copy()
+    out.columns = [
+        'Keyword',
+        'Projected Incremental MRR ($)',
+        'Avoided Paid Spend ($)'
+    ]
+    out['Impact'] = pd.cut(
+        out['Projected Incremental MRR ($)'],
+        bins=[-1, 500, 2000, float('inf')],
+        labels=['Low Priority','Moderate ROI','High ROI']
+    )
+    out = out.sort_values(['Impact','Projected Incremental MRR ($)'],
+                          ascending=[True, False])
 
     return summary, out
 
-
-# === Run & display ===
+# === Run forecast & display ===
 if st.button("Run Forecast"):
     df = load_csv()
     if df is not None:
         summary, table = calculate(df)
         if summary:
             c1,c2,c3,c4,c5,c6,c7 = st.columns(7)
-            c1.metric("Incremental Clicks", summary['clicks'])
-            c2.metric("Projected Signups",  summary['signups'])
-            c3.metric("New Customers",      summary['customers'])
-            c4.metric("Incremental MRR",    summary['mrr'])
-            c5.metric("SEO ROI",            summary['roi'])
-            c6.metric("Avoided Paid Spend", summary['avoid'])
-            c7.metric("Net Savings vs Paid",summary['net'])
+            c1.metric("Incremental Clicks",     summary['clicks'])
+            c2.metric("Projected Signups",      summary['signups'])
+            c3.metric("New Customers",          summary['customers'])
+            c4.metric("Incremental MRR",        summary['mrr'])
+            c5.metric("SEO ROI",                summary['roi'])
+            c6.metric("Avoided Paid Spend",     summary['avoid'])
+            c7.metric("Net Savings vs Paid",    summary['net'])
 
             st.subheader("📊 Opportunity Keywords")
             st.dataframe(table, use_container_width=True)