Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Meta Ads › Pixel, Conversions API & Measurement

Attribution Settings

How Meta attribution settings work: click and view windows, what the setting changes in optimisation and reporting, modelled conversions, and choosing one.
Guideintermediate11 min readUpdated

What an attribution setting controls

Meta's documentation describes the attribution setting as a finite period during which conversions can be credited to your ads and used to inform campaign optimisation. Three parts of that sentence carry the whole subject. Finite period: a conversion that happens after the window closes is not credited, however clearly the ad caused it. Credited to your ads: credit requires Meta to match the person who converted to a person who interacted with the ad, which is the job of the pixel and the Conversions API covered in the pixel setup tutorial and the Conversions API guide. Inform optimisation: the credited conversions are the examples delivery uses to find more converters, so the setting is a training decision, not only a reporting one.

The setting lives on the ad set, in the optimisation and delivery section, and applies to ad sets that optimise for a conversion-type event. Meta documents a default, which at the time of writing is a seven-day click combined with a one-day view window, and a set of alternatives drawn from click-only and click-plus-view combinations of one and seven days. Meta also reports conversions against the date of the ad interaction rather than the date of the conversion, which means the most recent days in any report are still filling in for as long as the window is open. The general question of how credit should be divided among touchpoints is the subject of the attribution models framework and the glossary entry on Attribution; Meta's setting is narrower, because within its window Meta uses its own rule for which ad gets the credit, and the advertiser chooses only the window.

Click-through and view-through windows

A click-through window counts a conversion if the person clicked the ad and converted within the stated number of days. A view-through window counts a conversion if the person saw the ad without clicking and converted within the stated number of days. The settings Meta offers combine the two: a click-only window, or a click window paired with a shorter view window. Meta's documentation describes clicks taking precedence over views when both occurred, so a person who viewed one ad and clicked another is credited to the click, and view credit applies only where no qualifying click exists.

Meta has also documented an engaged-view option for video ads, under which a conversion can be credited when a person watched a documented amount of the video without clicking and converted within a short window. Its availability depends on the objective and the ad format, and the Ads Manager dropdown for the ad set is the reliable list of what is currently offered; longer click windows have been available in reporting comparisons and, for some objectives, as settings at different times, and the Help Centre entry on attribution settings is the place to check the present state rather than this page.

The windows measure different things. Click-through conversions are the strongest evidence that an ad played a part, because the person took a deliberate action on it. View-through conversions are weaker evidence and more numerous: in a large retargeting audience, many people who were going to buy anyway will have scrolled past an ad in the day before they did. That does not make view-through worthless; awareness and reminder advertising works largely without clicks. It makes it a different kind of count, and one that should be read alongside Incrementality evidence before it is used to justify budget.

How much weight each kind of conversion deserves once it has been counted is the subject of the view-through versus click-through conversions comparison.

How the setting changes optimisation vs reporting

The reporting effect is the obvious one. Widen the window and more conversions are credited; the Results column rises, cost per result falls, and nothing about the ads or the audience has changed. Narrow it and the reverse happens. Because the figures move so easily, any comparison across ad sets, campaigns or time periods has to be made under one setting, which the next section covers.

The optimisation effect is less visible and more consequential. Meta documents that the attribution setting informs delivery. Conversions inside the window are the examples the system learns from, so the window shapes the kind of person it seeks. A one-day click window supplies examples of people who act fast after a click; the system learns to find them, which suits impulse purchases and short decision cycles. A seven-day click window adds people who take their time, which suits considered purchases and lead generation with a follow-up process. Including a one-day view adds people who did not click at all, which suits remarketing and reminder campaigns where the ad's job is to be seen. In practice, accounts that shorten the window to "be strict" sometimes find delivery narrowing towards a small, fast-acting group and volume falling; the setting was never only a reporting choice.

The two effects interact with the learning phase. Changing the attribution setting on a live ad set changes the optimisation event's definition, and it is prudent to treat it as a significant edit: expect a period of less stable delivery, and do not compare the periods before and after as if the ads were constant. Meta does not document every case here, so test on one ad set before changing an account-wide convention. Before that test, the conversion goal change checklist covers the timing, the record and the break in reporting that any change to what delivery learns from needs.

Comparing windows in reports

Ads Manager includes a compare attribution settings option in the column settings that shows results under several windows side by side, independent of the setting each ad set uses for delivery. It is the tool for three recurring questions. How much of this ad set's result is view-through? Compare the click-only column with the click-plus-view column; the gap is view credit. Do these two ad sets really differ? Put both under the same window and compare again. What would happen to reported results if we changed the setting? Read the alternative column before deciding, then remember that the optimisation effect will not show in the report. The same split applied to value is what the glossary entry on ROAS in Meta Ads uses to separate click-through from view-through return.

Conversion rate (CVR) and cost per acquisition (CPA) in any report inherit the window they were computed under. A target set under one window and judged under another is a common source of false alarms, and the CPA calculator is a quick way to recompute a target against a different conversion count once the compare columns have been read. Report headings should state the window; a figure without one is not a figure.

Illustrative example — not a real result

A mattress brand reads the same month three ways

A mattress brand spends £48,000 in a month on prospecting and retargeting. Under the default seven-day click, one-day view window, Ads Manager reports 1,210 purchases at £39.70 each. Switched in the compare columns to seven-day click only, the count falls to 820 at £58.50; under one-day click only it is 505 at £95. Almost all of the difference sits in the retargeting ad sets, where 310 of 390 reported purchases are view-through. The finance team had set a £45 target against the default window and was about to call the month a success; the marketing lead shows that prospecting alone is at £71 on seven-day click and that retargeting's view-through credit is mostly people who had already added to basket. They keep the default window for delivery, report both the default and click-only figures every month, and commission a lift study on retargeting before the next budget review.

Modelled and aggregated conversions

Meta can only count a conversion directly when it receives the event and can match it to a person who interacted with the ad. Browser restrictions, app privacy frameworks and consent choices remove some of those matches, and Meta documents that where this happens it uses statistical modelling to estimate conversions it could not observe, and aggregates some reporting so that individual-level data is not exposed. In Ads Manager this appears in several ways: results flagged as partially modelled, breakdowns that are unavailable for certain data, and reporting that arrives with a documented delay as modelled estimates settle.

The practical consequences are three. Recent days are provisional: a conversion count read the morning after can rise over the following days as delayed and modelled data lands, and comparisons of yesterday with last week flatter last week. Fine slices are unreliable: a breakdown by age, placement and day for a small ad set may be partly estimated or partly withheld, and the finer the slice the less the number means. And totals are estimates with a method behind them, not counts; the method is Meta's and is not fully published, so a reported figure is best treated as the platform's best view rather than as ground truth. None of this is a reason to distrust the reports for day-to-day optimisation, which is what they are designed for. It is a reason to keep a second measurement system for the decisions that matter.

Choosing a setting by objective

The default serves most accounts. The cases for changing it follow from the optimisation effect above.

Objective and situationSetting to considerReasoning
Online sales, short decision cycle, prospectingSeven-day click, one-day view (default)Enough time for a considered purchase; views add reach to learning
Online sales, impulse products, strict efficiency reportingSeven-day click only, with view shown in compare columnsKeeps optimisation on deliberate actions; view credit stays visible but is not counted
Lead generation with a sales follow-upSeven-day click, one-day viewForms are filled days after the click; views matter less than the click here
Retargeting warm audiencesDefault for delivery; click-only in reports; lift study when budget is materialView-through is large and mostly non-incremental; report both and test
Awareness or reach campaigns that also track conversionsInclude a view windowThe ad's job is to be seen; click-only would hide its contribution
Video-led campaigns where engaged-view is offeredConsider engaged-viewCredits watched-but-not-clicked video more conservatively than a plain view

Whatever the choice, apply it consistently across ad sets that will be compared, record changes in the account's change log, and state the window on every report that leaves the team.

Reconciling with analytics and other platforms

Meta's figures will not match a web analytics tool or another ad platform, and the causes are the ones this guide has described: a window after a click or a view, versus a last-source rule that cannot see views; reporting by interaction date, versus reporting by conversion date; identity matched across devices through accounts, versus identity confined to a browser; modelled fill-in, versus counting only what the tag saw. A further cause is overlap: every platform credits itself for a conversion that its own ad touched, so the sum of platform-reported conversions exceeds the real total, and the excess grows with the number of channels and the length of their windows.

Reconciliation is therefore a matter of lining up the rules, not of finding the "right" number. Put Meta under click-only in the compare columns and the analytics tool under a last-click view of paid social and the two will be far closer; the remaining gap is identity and consent. The comparison of Google Ads and Google Analytics figures walks through the same exercise for Google's platform, and the causes map almost one to one. For budget decisions across channels, the cleanest approach is a shared source of truth — analytics, a marketing mix model or a lift programme — with each platform's own attribution used for optimising inside that platform. The measurement hub sets out how the pieces fit, and the author's paid media analytics expertise page describes how the reconciliation is run in practice.

To line Meta's figures up against the business's own orders or leads rather than analytics, the tutorial on reconciling Meta results with your orders walks a closed month from events received to a ratio that targets can be set on.

Common mistakes

  • Comparing ad sets under different settings. The one with the wider window always looks better; use the compare columns.
  • Treating the setting as reporting only. It trains delivery; a change alters what the system looks for.
  • Reading yesterday's results as final. Reporting by interaction date with an open window means recent days fill in.
  • Taking view-through credit at face value in retargeting. Much of it is non-incremental; test before funding it.
  • Changing the window mid-flight without a note. Before-and-after comparisons become meaningless and the change log will not explain why.
  • Slicing modelled data finely. Small breakdowns of estimated figures are noise.
  • Expecting Meta and analytics to agree. Line up the rules, accept the identity gap, and keep a second source for budget decisions.

Frequently asked questions

Does the attribution setting change what Meta optimises for, or only what it reports?

Both. Meta documents that the attribution setting determines how results are measured and also informs delivery, because the conversions inside the window are the examples the system learns from. A shorter window teaches it to find people who act quickly; a window that includes views teaches it that an impression without a click can be worth paying for. Change the setting and you have changed the lesson as well as the number.

Why do two ad sets with identical ads show different cost per result?

If their attribution settings differ, they are counting different conversions, and the ad set with the longer or view-inclusive window will report more results and a lower cost for the same spend. Use the compare attribution settings option in Ads Manager to view both under one window before concluding that one ad set performs better. The same applies to comparisons over time if the setting was changed part-way through.

What does it mean when results are marked as modelled or when breakdowns are unavailable?

It means Meta could not observe every conversion directly, usually because of browser or app privacy restrictions, and has estimated part of the total statistically, which it documents alongside its privacy changes. Modelled results are an estimate of what happened rather than a count, and Meta withholds some breakdowns for data it has aggregated. Treat the totals as directionally useful and avoid slicing them finely.

Terms used on this page

Tools to try

Frameworks

Where this page is used

Part of these learning paths

Also related

Sources

  1. About attribution settings — Meta Business Help Centre
  2. Meta Business Help Centre — Meta Business Help Centre

Related