When you ask Facebook how well Facebook is working, Facebook tells you it's working great. When you ask Google how well Google is working, Google tells you it's working great. Both claim credit for many of the same conversions. This is not a bug in how these platforms are designed, it's a structural feature of how platform attribution works.
Understanding why this happens, and what to do about it, is one of the most practically valuable things a performance marketer can learn. It's also one of the core reasons marketing triangulation exists.
Every platform counts the same conversion: total claimed revenue exceeds actual revenue
What Platform Attribution Actually Measures
Each advertising platform observes only what happens within its own ecosystem. Google sees when someone clicks a Google Search ad, watches a YouTube video, or is served a Display ad through the Google Display Network. Meta sees activity on Facebook, Instagram, and WhatsApp. TikTok sees TikTok.
None of them see what happens in the other's ecosystem. None of them can observe offline behavior, word-of-mouth recommendations, email newsletter reads, organic brand searches triggered by TV spots, or the cumulative effect of years of brand building.
When Meta's reporting says a conversion was driven by a Facebook ad, it means a Facebook ad appeared in that customer's journey before the conversion, within Meta's attribution window. It does not mean the Facebook ad caused the purchase. It does not account for what Google observed. It does not tell you what would have happened if the ad hadn't run.
The Attribution Window Game
Platforms compete for conversion credit partly through their attribution window settings.
Meta's default attribution window is 7-day click and 1-day view. This means if someone saw your Facebook ad (even without clicking) and then made a purchase from any source within 24 hours, Meta claims credit for that conversion. If someone clicked your ad and then purchased within seven days, Meta claims it. A 1-day view window is particularly generous, someone scrolling past your ad without stopping is enough to claim a downstream purchase.
The result of these overlapping windows: when you add up all the revenue attributed across your ad platforms, the total typically exceeds your actual revenue, sometimes by 100–200%.
Consider a concrete scenario. A brand runs Facebook and Google campaigns simultaneously. Facebook's Ads Manager reports $50,000 in attributed revenue for the week. Google Ads reports $45,000 in attributed revenue. The brand's actual revenue that week was $60,000. Combined platform attribution is reporting $95,000, 58% more than what was actually generated.
Every dollar in that overlap is a dollar being counted twice (or three times, if TikTok is in the mix). Budget decisions made on this data are made on fiction.
Why This Is Structural, Not Malicious
It would be easy to frame this as the platforms deliberately deceiving advertisers. That's not quite right.
Each platform is reporting what it genuinely observes within its measurement methodology. Meta is correctly reporting that a Facebook ad appeared in the customer journey before the conversion, within Meta's attribution window. From Meta's perspective, that's an accurate statement of what Meta observed.
The problem is that "appeared in the journey" is not the same as "caused the conversion." And no platform has an incentive to build an attribution model that aggressively discounts its own contribution. The incentive structure is entirely the other direction: attribution models that credit the platform generously make the platform look more valuable, which encourages advertisers to spend more.
This is not a conspiracy. It's a predictable consequence of asking someone to grade their own homework.
What Platform Attribution Is Actually Useful For
Platform-native attribution is not worthless. It's just limited in what it can tell you.
Within a single platform, attribution data is comparatively reliable for optimization decisions. If Facebook shows that one creative drives 3x more conversions than another at the same spend level, that signal is probably directionally accurate, the same methodological biases apply to both, so the relative comparison holds even if the absolute numbers are inflated.
Use platform attribution for:
- Comparing creative performance within a channel
- Evaluating audience segments within a channel
- Understanding conversion timing and path within a channel
- Daily and weekly operational optimization within a platform
Do not use platform attribution for:
- Comparing the ROI of Facebook versus Google versus TikTok
- Making major cross-channel budget allocation decisions
- Determining whether to cut or scale a channel
- Understanding the true revenue impact of your marketing overall
What to Use Instead
For cross-channel decisions, you need measurement methods that are structurally independent of platform self-reporting.
Incrementality testing (holdout tests, geo experiments) compares business outcomes between groups that did and didn't see your advertising. The measurement happens in your own backend data, orders, revenue, not in platform-reported conversions. This gives you causal evidence of whether the advertising is driving incremental business results.
Media mix modeling analyzes historical relationships between your media spend and your actual business outcomes using econometric methods. It doesn't rely on platform-reported conversion data at all. It's inherently channel-agnostic.
Both methods have their own limitations and costs, but neither has a structural incentive to credit any particular channel generously. That independence is the point.
A Practical Starting Point
You don't have to abandon platform attribution entirely to improve your measurement. The most practical first step is to calibrate it.
Run one incrementality test on your highest-spend channel, probably Google Search or Facebook retargeting. Compare the incremental ROAS from the test to the attributed ROAS the platform reports. The ratio between the two is your calibration factor for that channel.
If Google Search shows attributed ROAS of 8x and your incrementality test shows iROAS of 3x, you now have evidence that Google Search attribution is inflating its contribution by roughly 2.7x. Apply that calibration skeptically to your other Google data. Repeat for other major channels over time.
It won't give you perfect numbers. But it will stop you from making million-dollar budget decisions based purely on what platforms tell you about themselves.