Meta Ads attribution: what the ROAS in Ads Manager doesn't tell you (and how to fix it)

30-second summary

• Meta Ads counts conversions by click (1, 7, or 28 days) and by view (up to 1 day after seeing the ad).
• The default window is 7-day click + 1-day view — which inflates the number.
• View-through attribution counts people who saw the ad but would have converted anyway.
• Fix: compare Ads Manager data against your destination platform and adjust the window to your purchase cycle.
• Never swap creatives or pause campaigns based on Ads Manager ROAS alone without that cross-reference.

Meta Ads is one of the platforms that delivers the most real results. It's also one of the most confusing in its reporting — especially when the account manager shows a 4× ROAS and the business's bank account doesn't feel the difference.

What is attribution in Meta Ads?

Attribution is the rule that determines which campaign, ad, or ad set receives credit for a conversion. The problem: the same sale can be attributed to multiple channels at the same time, each calculating its own ROAS.

In Meta, attribution works across two dimensions:

• Click window: how long after a click a conversion still counts for that ad. Options: 1, 7, or 28 days.
• View window: how long after *seeing* the ad — without clicking — a conversion still counts. Default: 1 day.

Meta's current default is 7-day click + 1-day view. That means: if someone clicked your ad on Monday and bought on Sunday, Meta credits that ad with the sale. If someone only *saw* the ad and bought the next day through another channel, Meta also counts that sale.

What is view-through attribution and why does it inflate ROAS?

View-through attribution (VTA) counts conversions from people who saw the ad but didn't click. The model's logic is that exposure influenced the decision.

The problem: Meta serves ads to millions of people every day. Many of those people were already on their way to buying — the exposure didn't cause the conversion, it just coincided with it. When Meta records that conversion as generated by the ad, ROAS goes up without the campaign having contributed anything.

This isn't manipulation — it's a real limitation of the exposure-based attribution model. The impact is largest for:

• Brands with strong organic demand: businesses with direct search traffic or strong referrals see VTA inflate their numbers more.
• Top-of-funnel campaigns: reach and awareness campaigns serve large volumes of people, broadening the base of "views" that can be attributed.
• Short-cycle products: if the average decision time is under 24 hours, the VTA window captures many conversions that would have happened without the ad.

Why does Ads Manager ROAS diverge from real ROAS?

Beyond view-through, two more factors create divergence:

1. Double attribution across channels. Meta, Google, and email can all attribute the same sale to themselves simultaneously. The sum of all channel ROAS numbers exceeds the real business ROAS. A single $100 order can show as $100 revenue in Meta, $100 in Google, and $100 in email — totaling $300 of attributed revenue for a single real transaction.

2. Delay between conversion and cash. Installment payments, pending charges, orders canceled afterward — Meta records the conversion at the moment of purchase click, not when the money arrives. If your cancellation rate is high, real ROAS is lower than reported.

The practical result: a 4× ROAS in Ads Manager can be a 2–2.5× ROAS when you cross-reference with confirmed orders in your system.

How to calculate real ROAS for Meta Ads?

Three approaches, in order of reliability:

1. Direct comparison (simplest): Pull total confirmed orders from your CRM or e-commerce in the period → calculate revenue → divide by Meta spend. Doesn't separate what came from Meta vs. other channels, but it's the number that matches your bank account.

2. UTMs + GA4 (more precise for web traffic): Set up UTMs on all Meta ads. In GA4, filter sessions and conversions from utm_source=meta. Compare against Ads Manager. The gap reveals the size of the VTA distortion.

3. Incrementality testing (most rigorous): Pause Meta for 2 weeks in a control region while keeping it running in another. Compare the sales variation between regions. The delta is the real incremental contribution. It's the most reliable method — and the most expensive to run. Makes most sense for accounts spending above $6,000/month.

For most businesses, combining approaches 1 and 2 delivers enough clarity for decisions. The post on the full Meta funnel shows how to allocate budget by stage once you have the real ROAS for each.

What to check before pausing or scaling a campaign?

The most common mistake: pausing a campaign with "low" ROAS in Ads Manager without verifying whether the number is real — or scaling a campaign with "high" ROAS inflated by VTA.

Before any budget decision, three checks:

1. Adjust the attribution window to your purchase cycle. If the average time from first contact to purchase is 3 days, use the 7-day click window and turn off view-through. If the cycle is 1 day (low-ticket product, impulse purchase), the 1-day click window is more realistic.

1. Check your VTA rate. In Ads Manager, you can break down conversions by "click" and "view." If more than 30% of attributed conversions came from views, the account deserves scrutiny before any decision.

1. Compare against external data. How many confirmed orders came in during the period? If the Ads Manager number is more than 20–30% higher, there's material distortion.

These checks take 15 minutes. They don't replace real-time metrics tracking, but they prevent the wrong call at the critical moment.

What is Meta doing to improve attribution?

Meta is investing in incrementality-based attribution models — the Conversions API (CAPI) is the main advance. With CAPI, conversion data comes directly from the business's server, not the browser, improving signal quality and reducing (but not eliminating) undercounting and overcounting.

What CAPI doesn't resolve: the decision about which attribution window best represents your business is still yours. The data gets better; the judgment about what to count always belongs to the account manager.

If the account still relies only on the browser pixel, migrating to CAPI is the first technical improvement to make — immediate impact on data quality and therefore on reported ROAS reliability. The natural next step is understanding what changes with platform automation and how to give the algorithm more control without losing visibility into what's working.

Frequently asked questions