Meta Ads Attribution Settings: Windows, Reported ROAS and Which to Use

A Meta ads attribution setting is the rule that decides which conversions Ads Manager credits to your ads: how long after a link click, an engagement or a view a purchase still counts. At the time of writing the default for website sales is 7-day click-through, 1-day engage-through and 1-day view-through. Keep the default for most stores, but always compare clicks with views before you trust the ROAS it reports.

What is an attribution setting in Meta ads?

An attribution setting is a pair of choices: which kinds of interaction can earn credit for a conversion, and how long after that interaction the conversion still counts. That length of time is the attribution window.

It lives on the ad set, in the conversion section where you choose your performance goal. That placement matters more than most buyers realise. The attribution setting does two jobs at once:

1. It shapes reporting. The Results, Cost per result and ROAS columns in Ads Manager count conversions according to the setting.
2. It shapes optimisation. Meta's delivery system learns from the conversions the setting counts. An ad set on 1-day click learns to find people who buy quickly after clicking. An ad set that also counts views learns from people who bought after only seeing the ad.

So changing the setting is not just a reporting preference. It changes what the algorithm chases.

Which attribution windows can you choose in 2026?

Meta reworked attribution twice in early 2026, so older guides are out of date. In January 2026 Meta removed the 7-day view and 28-day view windows from Ads Manager reporting and from the Ads Insights API. Then, in an announcement on the Meta for Business blog on 3 March 2026, Meta narrowed click-through attribution to link clicks and introduced engage-through attribution. The changes rolled out in stages, so check what your own Ads Manager shows.

For campaigns optimising for website conversions, the default is 7-day click-through, 1-day engage-through and 1-day view-through. Each conversion is credited to one type only, and a link click takes priority when someone both clicked and saw the ad, so the three columns don't double count.

What changed with engage-through attribution?

Before March 2026, a "click" for attribution meant any click on the ad, including a like or a tap on the profile name. Now only link clicks count as click-through. The other interactions moved into engage-through, which replaced engaged-view attribution and widened it. Meta also lowered the video threshold for an engaged view from ten seconds to five.

The practical effect: in many accounts click-through conversions fell after the rollout while engage-through conversions appeared. Most of that is reclassification, not a drop in sales. If you compare click-only numbers from February with click-only numbers from April, you are not comparing like with like.

Attribution model and conversion count

Two more settings sit next to the windows in many accounts.

• Attribution model. Standard uses the windows above. Incremental, which Meta began rolling out in 2025, tries to count only conversions its models predict the ad caused. When you choose it, you can't edit the windows yourself.
• Conversion count. All conversions counts every purchase a person makes inside the window. First conversion counts only the first one, which is useful when you care about new customers rather than repeat orders.

How does the attribution setting change your reported ROAS?

Every extra kind of credit adds conversions to the numerator of ROAS while the spend stays the same. Here is a hypothetical week to show the size of the effect.

Say an ad set spends $2,000 and your store sells products at an average order value of $50. Ads Manager's comparison columns show:

The same ads, the same week, the same customers, and four honest answers from 1.30 to 2.50. If your break-even ROAS is 2.0, this ad set is either losing money or comfortably profitable depending on which row you read.

None of these rows is "the truth". View-through conversions include some people who would have bought anyway, and the click rows miss some people the ad genuinely persuaded. What matters is knowing which row you are looking at, and how much of your ROAS depends on the softer kinds of credit.

Delayed conversions and the reporting date

Ads Manager credits a conversion to the date of the click or impression that earned it, not the date of the sale. A purchase made on Friday from an ad clicked on Monday shows up under Monday. That means the last few days of any report are still filling in. Yesterday's ROAS on a 7-day click setting will usually rise over the next week.

This is the most common reason people panic and pause ads too early. Judge a day only once its window has mostly closed, or judge a full week at a time, as described in our guide to the weekly Meta ads report.

Why don't Meta, Shopify and GA4 numbers match?

Because they count different things with different rules. None of them is broken.

A few patterns follow from this table.

• Meta usually reports more conversions than GA4 credits to Meta. Views, engagements and cross-device journeys are invisible to GA4.
• Platforms overlap. A shopper who clicks a Meta ad on Monday and a Google ad on Wednesday can be claimed by both. Add up every platform's reported revenue and you'll often get more than the store took.
• Shopify is the closest to the bank. It doesn't know which ad caused a sale, but it knows the sale happened, at what value, and whether it was refunded.

Good UTM parameters on every ad help Shopify and GA4 credit the clicks they can see. They will still never see views.

Use MER as the referee

When the platforms disagree, look at MER, marketing efficiency ratio: total store revenue divided by total ad spend across all channels. MER ignores attribution entirely. If Meta's reported ROAS climbs while MER stays flat and you haven't changed other channels, Meta is probably claiming sales that would have happened anyway. If both climb together, the improvement is likely real. The MER calculator does the sum.

Which attribution setting should you use?

For most e-commerce stores, keep the default for optimisation and read the comparison columns for decisions. Change the setting only when you have a clear reason.

A few rules of thumb:

1. Keep one setting for the ad sets you compare. Comparing an ad set on 1-day click with one on the default is meaningless.
2. Don't switch settings to make numbers look better. Adding view-through to a struggling ad set changes the report, not the bank balance.
3. Expect a reset. In many accounts you can't edit the attribution setting once an ad set is published, so a change means a new ad set and a new learning phase.
4. Write your choice down in the campaign name or your report, so the team knows which numbers they're reading.

How do you know which conversions your ads really caused?

Attribution tells you which conversions followed an ad. Incrementality tells you which ones happened because of the ad. They are different questions.

The most reliable answer comes from a conversion lift study, where Meta holds back a random group of people from seeing your ads and compares their purchases with the group that could see them. Lift studies need enough spend and time to be conclusive, and availability varies by account.

If a full lift study is out of reach, use simpler checks:

• Geographic holdouts. Pause Meta in a few comparable cities or regions for two to four weeks and compare store sales with the rest.
• Spend changes. When you raise or cut Meta spend meaningfully, watch total store revenue, not just Meta's report.
• New-customer share. Check whether Meta-attributed orders come from new customers or from people who already buy from you. Retargeting your own buyers inflates attribution with little real gain.

Getting the inputs right

No attribution setting helps if the conversion data is wrong. Before you argue about windows, confirm that:

• Each order fires one Purchase event, with the correct value and currency.
• The Meta Pixel and Conversions API are both running and deduplicated, so server events fill gaps without double counting. Our Shopify pixel and Conversions API checklist covers each check.
• Event match quality is reasonable, because better matching lets Meta connect more purchases to the people who saw your ads. See event match quality.
• Refunds and cancellations are tracked in your store, because Ads Manager won't remove a purchase when the order is later returned.

Fly ROAS reads results from your own Meta ad account and judges tests on cost per purchase against the target CPA you set, so it inherits whatever attribution setting your ad sets use. Pick the setting deliberately, keep it consistent, and both your own reports and any automation will compare like with like.

The short version

Attribution settings decide which conversions count and what Meta learns from. Leave the default on for most stores, read the click, engagement and view columns side by side, judge recent days only once their windows have closed, and use store revenue and MER to keep Meta's numbers honest. When a large budget decision rests on the answer, measure incrementality instead of arguing about windows.