Meta Attribution Window Deprecated: What Advertisers Need to Know
Meta changed how attribution works. If you're still on default settings, your ROAS numbers are wrong and so are your budget decisions.
In late 2025, Meta made a series of changes to its attribution model that most advertisers missed or only partially understood. A lot of accounts are now running with attribution settings that give a misleading picture of what's working.
This is not a technical edge case. If you're making budget decisions based on current ROAS numbers, this affects you.
What actually changed
Meta deprecates and modifies attribution windows periodically, but the more significant 2025 change was not about window length. It was about what Meta counts as an "attributed conversion" and how it surfaces that data in Ads Manager.
Three things shifted:
1. View-through attribution became less prominent in default reporting
Previously, 1-day view-through conversions were included in your default reporting window, which meant purchases that happened after someone saw (but didn't click) your ad counted toward your reported ROAS.
Meta's adjustments pushed view-through attribution into a separate reporting column that most buyers don't look at. If you're using default reporting, you're seeing a narrower picture of your ad's actual impact. This can make campaigns look less effective than they are -- or more effective than they are, depending on how your products convert.
2. Andromeda's reach expansion created longer conversion lags
Andromeda expanded how broadly Meta delivers your ads, reaching people earlier in their consideration cycle. People who wouldn't have been targeted before are now seeing your ads.
Conversions are happening on longer lags as a result. Someone might see your ad today and convert 9-12 days later. Under a 7-day click attribution window (the standard default), that conversion doesn't get credited to your campaign.
Your ROAS is being understated. You may be pausing campaigns that are working.
3. Modeled conversions vs. observed conversions
Meta has leaned further into modeled (statistically estimated) conversion data to fill in gaps left by iOS privacy changes. The ratio of modeled to observed conversions has increased, which means more of your reported conversions are estimates rather than directly measured events.
The models are reasonably accurate in aggregate. But your reported numbers have a wider error margin than two years ago, and comparing this year's ROAS to last year's involves more noise than the interface suggests.
How to know if this is affecting your decisions
Check your view-through data. In Ads Manager, add "1-day view" and "7-day view" as columns on your campaigns. Compare those view-through conversion numbers to your click-based numbers. For awareness-oriented campaigns, view-throughs typically represent 20-40% of conversions. If you haven't been watching this column, you've been missing part of the picture.
Compare Ads Manager ROAS to backend ROAS. Pull your actual revenue from your Shopify or e-commerce backend for the same period you're looking at in Ads Manager. If Ads Manager is showing significantly higher attributed revenue than what actually came in, your attribution is inflated. If Ads Manager is showing significantly lower, you're likely missing conversions due to short attribution windows or privacy gaps.
The gap between these two numbers is your attribution noise. A gap within 20% is roughly normal. Above 30%, your Ads Manager numbers are not a reliable basis for decisions without adjustment.
Look at your lag report. Meta has a "Time Lag" breakdown that shows you how many days elapsed between ad impression and conversion. If a meaningful chunk of your conversions are happening at 4-7 days (or longer), you're potentially cutting campaigns too early -- pausing when a campaign looks dead but is actually still converting on a lag.
What attribution window should you actually use?
The right attribution window depends on your product's purchase cycle, which most advertisers have never explicitly thought about.
Short consideration purchases (under $50, impulse-oriented): 7-day click, 1-day view is probably appropriate. These purchases happen quickly, so short windows capture most of the signal.
Mid-consideration purchases ($50-$300, some research involved): 7-day click, 7-day view is worth testing. Extending the view window can surface conversions that the 1-day view misses.
High-consideration purchases ($300+, extended decision cycle): 28-day click is the most accurate, though Meta has limited this window in most placements. At minimum, use 7-day click + 7-day view and supplement with backend data.
Subscription products: The initial conversion is just the first event. What matters is lifetime value. Attribution windows for subscriptions should be evaluated on retention metrics, not just initial ROAS.
Most advertisers are running the default 7-day click / 1-day view regardless of product type. That's fine for impulse purchases and wrong for almost everything else.
The blended CAC approach
Given the unreliability of any single attribution window, blended CAC (customer acquisition cost) is a more useful primary decision metric than ROAS.
Blended CAC works like this:
This metric is attribution-independent. It doesn't matter how Meta credits conversions internally, because you're measuring against real backend customer data. The question you're answering is: "How much does it cost us to acquire a new customer, including all the ad spend that contributed to that acquisition?"
Blended CAC requires combining Ads Manager spend data with CRM or e-commerce backend data. It's more work to set up than per-campaign ROAS, and considerably more useful for decisions.
When your blended CAC is below your target, you scale. When it's above, you investigate which channels or campaigns are dragging it up.
Setting up better attribution tracking right now
Three changes to make this week:
Add view-through columns to your Ads Manager report. Create a custom column set that includes 1-day view and 7-day view conversions alongside your normal click-based columns. Spend 15 minutes understanding what the split looks like in your account. You'll have more context immediately.
Check your Pixel event quality. Go to Events Manager and check your Pixel's event match quality score. Below 6/10, you're losing conversion signal that no attribution window adjustment recovers. Passing email, phone, and other match parameters is the highest-leverage improvement most accounts can make.
Start tracking new customers in your backend. Pull the number of first-time buyers from your Shopify or WooCommerce backend each week. This is the denominator for your blended CAC calculation. If you're not tracking it, your CAC calculations are guesswork.
Why this matters for creative decisions
Attribution gaps affect your creative strategy, not just your budget allocation.
If you pause a campaign because ROAS looks bad, and ROAS looks bad because of a short attribution window missing your 7-10 day converters, you've just paused a working creative. You'll re-launch it or something similar later. The "learning" from that creative is lost. You spend budget re-entering the learning phase.
Pausing, restarting, pausing, restarting is one of the most common and expensive mistakes in Meta advertising. The cause is usually not bad creative strategy. It's trusting a ROAS number that doesn't reflect what's happening.
Better attribution means fewer reactive decisions.
Run a free account scan with Campaiyn to see your current attribution gaps: how view-through conversions compare to click-based, and where conversion lag is introducing noise into your ROAS.
The takeaway
Meta's attribution changes won't stop. Privacy restrictions keep tightening, and the 2026 measurement environment is materially more complex than 2021. Build for that reality now.
The accounts that perform well in this environment combine Ads Manager data with backend data, view-through analysis, and blended CAC. They don't rely on a single number.
Your ROAS figure is a useful estimate. Treat it like one when you make budget and creative decisions.