How much should we trust each ad platform's own reported results?
Our answer
Use each platform's reported results only to compare campaigns inside it. Before you move budget between channels, discount or raise each channel's results by channel and tactic, using your own tests, or benchmarks until you have them.
- Confidence
- Medium
- Strongest proof
- 3 controlled tests
- 11 datasets, 5 practitioners, 3 opinions or claims
- Sources
- 15 sources
- 22 statements
New here? Take the 1-minute tour
Why this is the answer
3 reasons and our judgmentPlatforms over-credit some channels and under-credit others. Google branded search kept a factor of 0.30, Snapchat 0.30 and Meta retargeting 0.60, but Meta acquisition 1.13 (Common Thread Collective, Q1 2026, 299 brands). Most Meta factors predate Meta's March 2026 click change.
Tests find platforms both over-crediting and under-crediting themselves. Haus's YouTube geo tests found 2.3 times the sales the platform credited. BlueAlpha's tests for one advertiser found Meta over-crediting itself and TikTok under-crediting itself.
Added together, the platforms claim more sales than you made. Balistro finds claimed conversions add up to 1.4 to 2 times real orders for most consumer brands it audits. Visionary Marketing found claimed conversions 34% above actual ones across 240 audited accounts.
Our judgment
We correct each report by channel and tactic before any budget decision, because the test results from Common Thread Collective, Haus, Measured and BlueAlpha find gaps in both directions that differ by channel. Until you have your own tests, read the benchmarks for direction rather than as your factor, since single Facebook acquisition tests range from 0.5 to 2.4 times around a median of 1.14 (Common Thread Collective). We also make the cross-platform sum against store orders a standing check, drawing on Balistro's audit test and Visionary Marketing's reconciliation.
Our judgment, built from Common Thread Collective, Haus, Measured, BlueAlpha, Balistro and Visionary Marketing: our reading of these sources, not a finding from any one of them.
What the sources say
22 statements from 15 sourcesMarin Istvanic says one client added Pinterest, TikTok and Snapchat, each reporting 7 ROAS, but total revenue did not change. If MER falls with the new spend while revenue stays flat, the channel is just taking credit.
“if you add a new channel and MER drops proportionally to the new spend while total revenue stays flat, that channel is stealing attribution.”
Haus says the typical brand's YouTube geo test finds 2.3x the DTC sales the ad platform credited (80% interval 1.7x to 2.9x), and about three in four tests land above the platform's number.
“None of these numbers are yours, and the only YouTube result that should move your budget is the one measured on your own data.”
Measured, drawing on 25,000+ experiments across 160+ enterprise brands, says platforms heavily over-report retargeting, while social prospecting is reported accurately or under-reported. It gives no adjustment figures.
“within a single platform, prospecting and remarketing routinely sit far apart in measured incrementality while showing near-identical platform-reported ROAS.”
Flighted, a paid social agency, says Meta and TikTok both over-report conversions. Use MER or a third-party tool as the source of truth, and watch weekly MER as budget shifts between platforms.
“If MER drops when you increase TikTok spend, that tells you something.”
Measured's panel put commerce retargeting DSPs, including AppLovin at 2.0x, at 1.9x its benchmark, but results were bimodal: one DSP indexed above 4x at one brand and near zero at another.
“Retargeting-heavy programs need incrementality validation more than any other buy, because platform attribution over-credits most here.”
Ridge says a four-week TikTok GMV Max geo-lift study on one product line found 88% of the lift came outside TikTok Shop: 49% on Amazon and 39% on its own site.
“88% of lift from our GMV max ads was coming outside of TikTok”
Michael Kaminsky, whose firm Recast sells mix models, says most businesses do not need one and it can mislead them; he cites studies where last-touch was about 85 percent as good as experiments in simpler mixes.
“most businesses do not need an MMM. In fact, an MMM will be net negative for them because it will actively mislead them.”
Common Thread Collective's H1 2026 data puts Meta's average incrementality at 113.4%, but some brands are below 25% and others above 200%, the widest spread of any channel.
“Some brands are seeing Meta incrementality below 25%. Others are above 200%.”
Balistro, an agency, says to add up the conversions every platform claims and compare with real orders; for most DTC brands it audits, the sum is 1.4x to 2x real orders.
“For most D2C brands we audit, the platform sum is 1.4x to 2x real orders.”
CTC, which sells ad buying and measurement, says its geo holdout tests put Google branded search at a median 0.27x incremental ROAS, against 1.14x for Facebook acquisition and 0.67x for non-brand Google.
“That does not mean branded search is worthless, but it means it should not compete for budget against channels with meaningfully higher incrementality.”
CTC, which sells geo tests, says across hundreds of its tests Facebook acquisition's median incrementality factor is 1.14, with single tests ranging from 0.5x to 2.4x, so one test is not enough.
“single test results are insufficient”
Eightx, a DTC finance firm, says Postscript's attribution over-counts: in its client clean-ups, with no sample given, SMS-attributed revenue typically drops 20% to 40% once the click window is cut to 24 to 48 hours.
“set a conservative 24-to-48-hour click window and report off that.”
AppLovin says brands' holdout-measured iROAS relative to its click-only reported ROAS has more than doubled in a year, giving no test count. Its own holdouts cover 20 to 50% of geos at 90% power.
“MER, total revenue divided by total marketing spend, is the simplest of all of them”
WorkMagic, which sells geo tests and mix models, says platform ROAS includes baseline demand, misplaced halo and double counting, and that budgets should follow marginal ROAS, which can be below 1x at a 4x average.
“A channel with a healthy average ROAS of 4x can have a marginal ROAS below 1x if it's past the point of peak efficiency.”
BlueAlpha, which sells incrementality testing, ran geo tests for beehiiv: Meta's true cost per signup was about 345% above what Meta reported, while TikTok undercounted itself by about 10%.
“Measured against a counterfactual, reported numbers are wrong in both directions, and which direction depends on the channel.”
CTC's Q1 2026 benchmark of 299 brands, using its incrementality factors, puts Google brand search at 19.1x platform ROAS but 5.7x incremental, and Meta acquisition at 1.83x reported versus 2.07x.
“Google Brand search reports 19x ROAS but delivers just 5.7x after incrementality adjustment (0.30x IF).”
In CTC's Q1 2026 benchmark of 299 DTC brands, Snapchat's platform ROAS of 6.18x fell to 1.85x once adjusted for incrementality, and Pinterest's 6.57x to 2.76x. Each took under 0.5% of total spend.
Across 299 DTC brands in Q1 2026, CTC's test-based factors cut Google brand search from 19.07x reported ROAS to 5.72x incremental, while Meta acquisition rose from 1.83x to 2.07x.
“nearly 70% of reported value was just capturing existing demand.”
Common Thread Collective, across 50 high-spend Meta accounts, saw about 13% less 7-day click revenue reported after Meta's March 2026 click change, so it cuts a 2.0x ROAS target to about 1.74x. Before and after only.
“A pre-change target of 2.0x should be adjusted to approximately 1.74x under the new definition (a ~13% haircut).”
Across 2,226 randomized Meta experiments, 7-day last-click attribution explained little of true incremental conversions per dollar (R2 0.19), while a model trained on past experiments reached 0.88.
“PIE disagrees with RCT-based decisions in only 8-12% of campaigns, compared to 12-20% for last-click attribution.”
Visionary Marketing, a UK agency, checked 240 client accounts against backend orders: Google over-claimed by 18%, Meta 24%, LinkedIn 31%, TikTok 36%. That measures over-claiming, not incrementality.
“the sum of claimed conversions exceeds actual conversions by an average of 34%.”
Zettler Digital, a Klaviyo agency, says last click overstates email and SMS. It tests one program at a time with a random 90/10 or 80/20 holdout for at least 2 to 4 weeks, measured on Shopify revenue per profile.
“Last-click attribution will almost always overstate the impact of email and SMS, especially for brands with strong brand demand, repeat customers, and short consideration cycles.”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- Where each source standsA chart placing each source on the question
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer
Read next
More in Attribution- How should we attribute sales from creators and influencers?
- How should we attribute sales from podcasts, TV or other offline channels?
- How should we credit email and SMS, which often claim sales already driven by ads?
- How do we stop affiliates and coupon sites from claiming sales we would make anyway?