Full synthetic sample · not a customer result
Ad Spend Decision Memo
An independent verdict from your real orders.
The SpendProof framework
Every memo runs the same five stages, so you always know where a conclusion came from. The verdict leads (§2) so you get the answer first; the stages behind it follow.
1. The decision being evaluated
The brand is planning to increase Meta ad spend by 30% starting next week — from ~$45k/mo to ~$58.5k/mo (about +$13,500/mo in new spend) — expecting the added spend to return roughly what Meta currently reports (a 6.4× ROAS).
2. Verdict
The current evidence is not strong enough to confirm the scale-up clears its own margin-adjusted breakeven. The move is not clearly wrong either — it is unproven on the numbers as they stand. A cheap, two-week test (Section 8) would resolve it before committing the full 30%.
| Outcome | Meaning | This decision |
|---|---|---|
| Supported | Evidence is strong enough to justify the move. | — |
| Unsupported | Evidence argues against the move. | — |
| Unresolved | Evidence is too conflicted or incomplete to justify it yet. | ◄ this one |
3. Confidence
Confidence in this illustrative verdict: Medium. The supplied figures contain a material contradiction, but they do not identify overlap at order level or establish marginal return. The margin band is a synthetic customer-supplied assumption, not an audited number.
4. Evidence reviewed
- Shopify net revenue for the window (after refunds and discounts).
- Meta, Google, and Klaviyo reported/attributed revenue for the same window.
- Reported spend for Meta and Google.
- The promotional calendar (a 20%-off bundle ran on ~18 of the 30 days).
- Stated margin assumptions (COGS, fulfillment, payment fees) — see Section 6.
Everything below is marked observed (read directly from the figures) or inferred (reasoned from them). Nothing is asserted as certain that isn’t.
5. Where the numbers conflict (the discrepancy map)
Same stated window, each system’s synthetic claim side by side. Shopify net revenue is used as the store-recorded sales anchor; it is not treated as attribution truth.
| System | Reports it drove (30d) | Spend | Reported ROAS |
|---|---|---|---|
| Meta Ads | $290,000 | $45,000 | 6.4× |
| Google Ads | $95,000 | $18,000 | 5.3× |
| Email / SMS (Klaviyo) | $120,000 | — | — |
| Sum of channel claims | $505,000 | $63,000 | — |
| Shopify net revenue (sales anchor) | $360,000 | — | — |
Observed in the synthetic inputs: channel-reported revenue totals ~$505k while Shopify net revenue is $360k, or about 140% of the store-recorded anchor. The ~$145k gap shows the claims cannot be added as unique revenue; it does not prove which claims overlap or by how much.
Inference bounded by the evidence: Meta’s reported 6.4× ROAS cannot be treated as incremental or marginal return. Possible explanations include overlapping attribution, different revenue bases, and window differences; the supplied aggregates do not isolate one cause. Blended MER is~5.7× ($360k ÷ $63k), but it is still an average, not the return on the next dollar.
Other distortions checked
- Returning-customer mix — unresolved. No new-versus-returning breakdown was supplied, so the memo does not estimate its effect.
- Cross-device / window leakage — possible, not observed. The aggregate figures cannot measure it or determine its direction.
- Promo-window distortion — observed. A 20%-off bundle ran ~18 of 30 days, over-stating baseline volume and dragging contribution to the low end.
6. Profit / margin overlay
ROAS is a revenue number. The decision is a profit decision. Overlaying margin changes the picture.
- Synthetic customer-supplied contribution-margin assumption: ~53–57% of net revenue (COGS ~30%, fulfillment ~8%, payment fees ~3%), using the midpoint 55% below. This is a scenario input, not a measured finding.
- Contribution profit before ad spend ≈ 55% × $360k = ~$198k.
- Less ad spend ($63k) → contribution profit after ad ≈ ~$135k for the window.
7. Scenario view
Modeling the +$13,500/mo of new Meta spend at the 55% contribution-margin midpoint. Incremental contribution = incremental spend × (marginal ROAS × 0.55 − 1).
| Scenario | Marginal ROAS | Incremental revenue | Incremental contribution |
|---|---|---|---|
| Optimistic | 4.0× | ~$54,000 | +$16,200 |
| Base | 2.8× | ~$37,800 | +$7,300 |
| Downside | 1.8× | ~$24,300 | −$150 (≈ breakeven) |
Read: the move is clearly profitable only if the next dollar of Meta spend holds a marginal ROAS well above the ~1.8× breakeven. The current aggregates do not measure marginal ROAS, so none of the scenario rows is presented as a forecast. The reported 6.4× describes attributed revenue on prior spend, not the next dollar.
8. Recommendation
- Step, don’t leap. Raise Meta spend +10–15% for two weeks and watch contribution profit, not platform ROAS.
- Deduplicate before you trust the number. Treat blended MER (~5.7×) and Shopify net as the anchor; discount stacked per-channel ROAS until order-overlap is confirmed.
- Take the promo out of the baseline. Re-read a clean, post-promo 14-day window before setting the new spend level.
8a. What to watch after you act
- Marginal contribution, weekly — the +10–15% step should hold contribution profit positive on a clean window. If it dips toward breakeven, stop before the full 30%.
- Blended MER, not platform ROAS — a falling blended MER as you add spend is the early signal the next dollar is underperforming the average.
- New-vs-returning mix — if the added spend mostly harvests returning customers, acquisition CAC is worse than it looks.
Come back for a follow-up read if the test is ambiguous, the blended MER slips, or the promo calendar changes — a post-move check is faster than a first memo because the baseline is already set.
9. What would change this conclusion
The verdict could move to Supported if new evidence resolved the named uncertainties, such as:
- A clean, post-promo window showing Meta efficiency holding without the discount distortion.
- Contribution-margin-by-order confirming margin sits at the higher end.
- A stepped-spend or geo-holdout test showing marginal ROAS above breakeven.
- Confirmation that channel claims don’t overlap as much as the 140% figure implies.
It would move to Unsupported if a clean window showed marginal contribution at or below breakeven, or if margins are thinner than assumed.
10. Limits (what this memo does and doesn’t claim)
- It does not create perfect attribution or establish the “true” ROAS of any single channel.
- It does not replace MMM, multi-touch attribution, incrementality testing, or a finance close.
- Findings marked inferred are reasoned from available data, not controlled tests.
- Dollar figures are bands with stated assumptions, not forecasts.
- The review is based on reported figures and screenshots for one decision window, not full account access.
—. Inputs required (what a real memo would ask you for)
Low-effort, no account access. Screenshots or CSV exports are enough, and you can redact freely.
- The decision + deadline — one or two sentences.
- Shopify net revenue for the window — screenshot or CSV, after refunds & discounts.
- Each system’s reported revenue + spend — screenshots, same window.
- Approx margin inputs — a ballpark COGS / fulfillment / payment % unlocks the profit overlay.
- Promo / discount calendar — a sentence, so the baseline isn’t distorted.
What we never need: passwords, account or ad-manager access, or customer-level personal data. You don’t need to connect anything to start.
Have a live decision like this?
One accepted Decision Brief, no account access, and a $199 flat total. Delivery is within three business days after complete usable evidence is confirmed; the complete memo comes before payment is due.
Full synthetic sample · SpendProof · not a customer result or benchmark. The accepted Decision Brief fixes scope and delivery criteria before work starts; the complete memo receives a separate critical QA pass before the $199 flat payment is due.