By Christina Bell | Head of Strategy, Webtopia

Meta Incremental Attribution (IA) is an attribution model that uses machine learning, trained on thousands of Meta Lift studies, to optimise ad delivery specifically toward conversions that were caused by your ads, rather than conversions that would have happened regardless. Advertisers using Incremental Attribution are seeing an average 25% increase in incremental conversions compared to business-as-usual attribution strategies. For DTC brands frustrated by the gap between what Meta reports and what they can actually trace to revenue, IA is the most substantive change to Meta's attribution framework in recent years.

The Attribution Problem Every DTC Brand Has

Meta's standard attribution reports conversions based on who saw or clicked an ad within a given window, typically a one-day click or a seven-day click. The problem is that some of those people would have purchased anyway. They were already searching for the product, already warm from a previous brand interaction, or simply in a buying mindset that had nothing to do with your ad. Including them in your conversion count makes your campaigns look more efficient than they actually are.

This is not a flaw in Meta's reporting. It is a fundamental limitation of observational attribution. When you see a conversion credited to an ad, all you know is that the person was exposed to the ad and then converted. You do not know whether the ad caused the conversion. That distinction is the entire premise of incremental attribution, and it is the one that matters most when you are making budget decisions.

For DTC founders, the practical consequence is this: if your Meta ROAS looks strong but your revenue does not grow proportionally when you increase budget, you may be running into what is sometimes called the attribution illusion. You are optimising toward people who would have bought regardless, and the reported efficiency is masking the fact that significant spend is going to zero-incremental-value conversions.

What Incremental Attribution Actually Measures

Meta's Incremental Attribution model asks a different question from standard last-click or view-through attribution. Instead of asking 'did this person convert after seeing the ad?', it asks 'did this ad cause this person to convert who would not have otherwise?'

The model is trained on data from thousands of Meta Lift studies, which are controlled experiments that compare the behaviour of people who saw ads against matched groups who did not. From these experiments, Meta has built a machine learning model that can predict, in real time, the likelihood that a given user would have converted without ad exposure. Delivery is then optimised to prioritise people for whom the ad is predicted to make the decisive difference.

The updated model, announced in April 2026, leverages new types of data including user engagement behaviour, which Meta states has driven significantly stronger results. The headline number from this update is a 25% average increase in incremental conversions for advertisers using IA compared to those using standard attribution strategies. That figure comes from Meta's own internal measurement and is a comparison against business-as-usual approaches, not a guarantee of performance for any individual account.

Why the 25% Lift Stat Needs Context

A 25% increase in incremental conversions sounds compelling, and the underlying mechanism is sound. But it is worth being precise about what this means in practice, because it can feel counterintuitive when you first switch.

When you activate Incremental Attribution, you may initially see fewer total reported conversions in Ads Manager. That is not a bad sign. It is the model doing its job: deprioritising spend on people who would have converted anyway, and redirecting it toward people for whom the ad is genuinely driving the decision. The net result is that each conversion in your count is more likely to represent a real causal outcome, and the business impact per pound of ad spend increases even if the raw conversion number looks lower.

This is why Incremental Attribution should be evaluated alongside your external measurement source of truth, whether that is a media mix model, a post-purchase survey, or a GA4 revenue report, rather than solely within Ads Manager. The in-platform numbers will look different. The downstream revenue impact is the signal that actually matters.

Meta's own guidance for advertisers already using IA and seeing results is telling: headroom analysis shows that most advertisers who increase their IA share of spend do not see diminishing returns. The recommendation is to expand IA coverage by approximately 20% over the following one to two months while monitoring results in your measurement source of truth. That is a confident signal about the model's stability at scale.

How to Activate Incremental Attribution in Ads Manager

Incremental Attribution is activated at the ad set level during campaign creation or editing. In Ads Manager, when you reach the ad set configuration, you will find the attribution model selection as part of the optimisation and delivery settings. Switching to Incremental Attribution from this point changes how Meta's delivery algorithm selects the audience for your ads.

The activation applies across a variety of campaign types, which makes it broadly accessible regardless of whether you are running standard conversion campaigns, lead generation, or catalogue sales. It does not require a separate campaign structure or special access, which is meaningfully different from some of Meta's more advanced tools that are restricted to certain account types or require beta enrollment.

If you are new to IA, Meta's guidance is to begin with a structured test: run Incremental Attribution against your existing approach on a portion of your budget, measure the results over four to six weeks, and compare the incremental impact using your external measurement tools before committing the majority of spend to the new model.

The Relationship Between Incremental Attribution and Creative

There is a less obvious connection between incremental attribution and creative strategy that is worth understanding. When the delivery algorithm is optimised toward incremental impact rather than raw conversions, the audience it targets shifts. It becomes more likely to reach people who are genuinely persuadable, rather than people who were already going to buy.

People who are genuinely persuadable are, by definition, less warm to the brand. They have not already decided. This means the creative needs to do more of the actual persuasion work. An ad that performs well when optimising for standard conversions (reaching warm audiences who are close to purchase anyway) may not perform as well when the audience shifts to include more genuinely new buyers who need a stronger reason to act.

In our experience at Webtopia, brands that see the best results from Incremental Attribution tend to be those with strong creative variety across the funnel. They have upper-funnel content that introduces the brand and builds desire alongside mid-funnel content that addresses objections and drives consideration. When the algorithm is looking for the highest incremental impact, having that breadth of creative available gives it more to work with.

This connects directly to the value of structured creative testing. Our guide to Meta Creative Testing explains how to run clean creative experiments within your existing campaigns while keeping delivery learnings intact, which becomes especially important when you are building out the creative depth that Incremental Attribution rewards.

Incremental Attribution and Meta's Broader AI Delivery System

Incremental Attribution does not operate in isolation. It is one component of Meta's broader shift toward AI-driven optimisation, where the delivery system is making increasingly granular decisions about who to show ads to, at what time, in what format, with what creative. The goal across all of these systems is the same: maximise the genuine impact of advertising spend, not just the appearance of impact in a reporting dashboard.

Understanding how these systems interact is valuable for any DTC founder who wants to make informed decisions about their Meta strategy. Our piece on Meta Andromeda covers how Meta's AI system reads and categorises creative to match ads to the right audiences, which operates alongside the attribution model to determine both who sees your ads and how efficiently that reach is attributed.

The practical takeaway from all of this is that the Meta of 2026 is not the Meta of 2022. The platform has moved from a model where founders could exercise detailed manual control over audiences and bidding toward one where the algorithms are making most of the consequential decisions. The founders who perform best on the platform are not the ones trying to wrestle back control. They are the ones who understand what signals to give the algorithms, what creative to provide, and how to measure the outcomes accurately enough to know whether the system is working for their business.

Incremental Attribution is part of that measurement layer. It does not change what you do in your campaigns so much as it changes what you are optimising toward, and over time, optimising toward the right thing compounds into a meaningfully better result.