View-Through Conversions Are Inflating Your Meta ROAS, and the Gap Just Got Wider

The ROAS number sitting in your Meta dashboard right now is not a measurement.

 Increasingly, it’s an estimate, and the gap between the two just got harder to ignore. Meta’s default attribution window for 2026 is 7-day click, 1-day view, a much tighter window than the 28-day click standard advertisers relied on before iOS 14.5. 

And for the iOS traffic that window can’t directly observe, Meta leans on probabilistic modeling to fill in conversions it never actually saw happen.

Translation: more of what’s on your screen is a statistical guess dressed up as a hard number. Here’s what changed, why the gap is structural rather than a rounding error, and what to actually do about it.

What Meta’s Attribution Window and Modeling Actually Look Like Right Now

Attribution windows determine how long after someone clicks or views your ad Meta will still credit a conversion to it. 

The 2026 default is 7-day click plus 1-day view, down from the 28-day click window that was standard before Apple’s App Tracking Transparency framework arrived in 2021. 

Meta also permanently removed the 7-day and 28-day view-through windows from its Ads Insights API on January 12, 2026, leaving 1-day view as the only view-through option left.

For the conversions Meta can no longer directly observe, mostly opted-out iOS traffic, it fills the gap with modeled conversions. 

As one measurement guide put it, modeled conversions are platform language for we’re estimating what happened because we can’t directly measure it. The models are trained on observed behavior from users who didn’t opt out, then applied to estimate what similar users who did opt out probably did. 

It’s a reasonable statistical approach. It is still a guess, not a receipt.

Skip Conversions API and the gap widens further. 

Pixel-only tracking now loses between 25 and 30 percent of an account’s conversion data before any modeling or analysis even happens. For a closer look at what the January window removal specifically cost accounts that didn’t catch it, see what the view-through window removal really cost you.

The Gap Is Not a Rounding Error, It’s Structural

One account we reviewed showed a 3.23x ROAS inside Meta’s dashboard. 

The real number, tracked from verified orders and tied to first-party data, was 0.93x. That’s not measurement noise. That’s the platform crediting itself for results it can’t actually prove happened, largely view-through and modeled conversions that never involved a verifiable click.

This isn’t one unlucky account. 

It’s the structural result of a platform that reports on its own performance using a mix of shrinking direct observation and expanding statistical estimation. 

Facebook typically over reports conversions by 15 to 30 percent due to iOS tracking restrictions, and modeling only partially compensates. 

Meta isn’t lying exactly, its documentation is fairly explicit about what’s modeled versus measured, but nothing in the dashboard forces that distinction on you by default. 

The number just shows up as ROAS, full stop, and most budget decisions get made off that single blended figure. 

That’s the core issue with letting any platform grade its own homework: it has every incentive to make its number look complete, and no obligation to flag what it filled in. Our deeper look at why that matters is in attribution accuracy: why platforms shouldn’t grade their own homework.

Why This Matters More at Today’s Ad Costs

A wider gap between reported and real ROAS was always a problem. 

It’s a more expensive one now that Meta CPMs are running roughly 20 percent higher year over year, averaging $14.19 industry-wide in 2026. 

When every impression costs more, a budget decision based on an inflated ROAS number doesn’t just look wrong on a dashboard. It compounds into real dollars misallocated toward campaigns that a platform’s own modeling made look better than they were.

This is playing out across every account running meaningful iOS traffic right now, not just the one we reviewed. If you’re setting budgets off the Meta dashboard alone, you’re setting them off a guess, and guesses get expensive fast at today’s auction prices.

What to Actually Do About It

• Implement Conversions API alongside your pixel, not instead of it. Meta recommends running both, and doing so typically raises match rate from the 60 to 70 percent range up to 85 to 95 percent, meaningfully shrinking the gap the platform has to model around.

• Separate modeled conversions from verified ones in your own reporting. Even a simple tag distinguishing “platform-modeled” from “confirmed order” changes how a budget conversation goes.

• Run a holdout test on your highest-spend campaigns. Comparing performance with and against a withheld audience segment tells you what’s incremental, independent of how Meta’s model chose to fill in the blanks. Our walkthrough on incrementality testing covers how to set one up.

• Track your real numbers separately, tied to your own store data. Click-based, first-party attribution can’t be inflated by a modeling assumption, because it only counts what actually happened.

If you want to see the real gap between what Meta reports and what your store actually sold, tracked independently and tied to your own order data, book a live AdBeacon demo.

—-

FAQ

What is Meta’s default attribution window in 2026?

The default is 7-day click plus 1-day view. Meta also permanently removed the longer 7-day and 28-day view-through windows from its Ads Insights API in January 2026, leaving 1-day view as the only view-through option.

What are modeled conversions in Meta Ads?

Modeled conversions are statistical estimates Meta generates for conversions it can’t directly observe, mostly from iOS users who opted out of tracking. The models are trained on behavior from users who didn’t opt out and applied to estimate what similar opted-out users likely did.

Why is my Meta-reported ROAS higher than my actual ROAS?

Reported ROAS blends verified, click-based conversions with view-through credit and modeled estimates for traffic Meta can’t directly track. That mix can significantly overstate results, since the platform has every incentive to fill gaps generously and no obligation to flag which numbers are estimated.

Does Conversions API fix the attribution gap?

It significantly narrows it. Pixel-only tracking can lose 25 to 30 percent of conversion data before analysis even starts, while running Conversions API alongside the pixel typically raises match rate into the 85 to 95 percent range.

How can I get an accurate ROAS number instead of relying on Meta’s dashboard?

Pair server-side conversion tracking with click-based, first-party attribution tied to your own order data, and run periodic holdout tests to confirm what’s actually incremental rather than trusting a single blended, platform-reported figure.

Sources

• Meta Ads Attribution Models: The Complete Guide 2026, Lionel Fenestraz
• How iOS 14.5 Changed Ad Attribution Forever, Hyros
• Meta Ads Benchmarks 2026: CPM, CPC, CPA & CTR by Industry, Ryze AI
• Meta Conversions API (CAPI): Setup, Deduplication & Best Practices (2026)
• Facebook Ads Attribution 2026: Measure True ROAS Post-iOS, Adligator