### 1. Meta Totals, Blended CPA, and ROAS Under Each Source #### Meta Ads Manager (Self-Reported) * **Total Spend:** $\$1,500.00 + \$900.00 = \$2,400.00$ * **Total Conversions:** $38 + 22 = 60$ * **Total Revenue:** $\$3,420.00 + \$1,980.00 = \$5,400.00$ * **Blended CPA:** $$\text{CPA} = \frac{\text{Total Spend}}{\text{Total Conversions}} = \frac{\$2,400.00}{60} = \$40.00$$ * **Blended ROAS:** $$\text{ROAS} = \frac{\text{Total Revenue}}{\text{Total Spend}} = \frac{\$5,400.00}{\$2,400.00} = 2.25\times \text{ (or } 225\%)$$ #### Google Analytics 4 (GA4 Paid Social) * **Total Spend:** $\$2,400.00$ (incurred across Meta campaigns) * **Attributed Purchases:** $36$ * **Attributed Revenue:** $\$3,240.00$ * **Blended CPA:** $$\text{CPA} = \frac{\text{Total Spend}}{\text{Attributed Purchases}} = \frac{\$2,400.00}{36} \approx \$66.67$$ * **Blended ROAS:** $$\text{ROAS} = \frac{\text{Attributed Revenue}}{\text{Total Spend}} = \frac{\$3,240.00}{\$2,400.00} = 1.35\times \text{ (or } 135\%)$$ --- ### 2. Documented Causes for the Disagreement Based strictly on the tracking notes, the two sources diverge due to four specific structural and operational factors: 1. **View-Through Conversion Tracking:** Meta Ads Manager includes view-through conversions via its "1-day view" window. GA4 never tracks or credits view-through conversions under any configuration. 2. **Attribution Logic and Lookback Windows:** Meta uses a siloed model claiming credit if a user interacted within a 7-day click or 1-day view window, evaluating conversions only against its own ads. GA4 uses a cross-channel "last non-direct click" model with a 30-day lookback window that awards 100% of the conversion to only one channel. If a buyer clicked a Meta ad but later clicked an Organic Search result, GA4 credits Organic Search, whereas Meta still claims the purchase. 3. **Broken Auto-Tagging Outage (Oct 12–13):** Auto-tagging failed for two days, causing an estimated 8 Meta-referred sessions to arrive without UTM parameters. These were misclassified in GA4 as **Direct** instead of **Paid Social (Meta)**. 4. **Platform Double-Counting vs. Site-Wide Measurement Gaps:** Ad platforms collectively claim overlapping orders: summing all platform reports (Meta, search, affiliate) totals 149 claimed conversions, even though the store platform recorded only 118 real orders. Furthermore, GA4 recorded 112 total site-wide purchases, meaning it missed 6 completed store orders entirely. --- ### 3. Defensible Range and Honest Conclusions on Meta Attribution * **Defensible Range of Attributed Purchases:** **36 to 60 purchases** (associated with **$\$3,240.00$ to $\$5,400.00$** in attributed revenue). * **Floor (36 purchases):** Documented by GA4 under strict last non-direct click attribution. If any of the 8 untagged Meta sessions on Oct 12–13 resulted in a purchase, the true last-click floor lies between **36 and 44 purchases**. * **Ceiling (60 purchases):** Documented by Meta Ads Manager under 7-day click / 1-day view attribution. * **Attribution vs. Causation:** Neither number represents *incrementality* (purchases that occurred strictly *because* of the ads): * Meta’s figure of 60 includes view-throughs and retargeting (22 conversions) that likely capture shoppers who were already intending to buy. * GA4’s figure of 36 strips credit from Meta whenever another non-direct channel was the final click, obscuring Meta’s role in opening or assisting the path to purchase. --- ### 4. Recommendations and Next Steps #### Robust Conclusions (True Regardless of Measurement Model) * Meta generated substantial genuine interest: 1,840 tracked sessions and at least 36 confirmed last-click purchases. * Meta generated more top-line revenue than the $\$2,400.00$ spent under both reporting lenses ($1.35\times$ in GA4; $2.25\times$ in Meta). * Platform self-reporting is inherently non-deduplicated across vendors: the 149 platform-claimed conversions exceed the store’s actual 118 orders by 31 orders. Meta’s figures cannot be evaluated in a silo without acknowledging shared touchpoints. * Site-level tracking leaks exist: GA4 dropped 6 actual store orders (112 vs. 118), and the Oct 12–13 outage artificially deflated GA4’s Paid Social session and conversion counts. #### Source-Dependent Conclusions (Belief Required) * **Whether the ads "printed money":** If you believe Meta ($2.25\times$ ROAS, $\$40.00$ CPA), the campaign was highly profitable. If you believe GA4 ($1.35\times$ ROAS, $\$66.67$ CPA), the ads produced modest returns that may be near or below break-even once product margins and overhead are factored in. * **The value of ad views:** Believing Meta assumes 1-day view impressions actively drove purchasing decisions; believing GA4 assumes ad impressions that did not receive a click had zero commercial impact. * **Retargeting efficiency:** Meta attributes 22 purchases to retargeting; GA4's cross-channel view likely reallocates a portion of those to Organic Search or Direct if those channels captured the final non-direct click. #### Next Measurement Step 1. **Infrastructure Remediation:** Reconcile the discrepancy between the store’s 118 orders and GA4’s 112 orders, and audit auto-tagging/UTM pipelines to ensure Meta traffic does not fall into Direct. 2. **Causal Incrementality Testing:** Because neither observational attribution model measures true cause-and-effect, run a **controlled incrementality test** (such as a geo-lift test or conversion holdout test on Meta). This will measure whether turning Meta ads on or off causes a net change in total store orders (out of the 118 baseline), rather than debating which tracking tool gets the credit.