Snap today opened Unified Attribution to every app advertiser running AppsFlyer or Adjust, moving a measurement setting out of beta and into general availability across all markets. The campaign-level control hands the definition of a conversion to the advertiser's mobile measurement partner, and it cannot be switched off once a campaign exists.

Snap Inc. confirmed the global rollout of Unified Attribution on August 31, 2026, describing it as available to app advertisers using AppsFlyer or Adjust, with support for further mobile measurement partners in progress. The setting sits at the campaign level inside Snap Ads Manager, selected during creation under the App Promotion objective, and it replaces the platform's own default conversion accounting with events, windows and settings drawn from the advertiser's measurement vendor.

What sounds like a reporting preference is closer to a change in what the bidding system chases. Snap's help documentation for the feature states that Unified Attribution uses MMP-modeled signals alongside Snapchat attribution, so that optimisation is powered by Snapchat's attribution while outcomes are brought closer to what the MMP reports. Whether that description survives contact with the company's own marketing language is a question the two documents leave open, and it is examined further below.

What the setting actually controls

Unified Attribution is not a toggle applied to an existing campaign. According to Snap's Business Help Center documentation, the attribution method is fixed at creation and cannot be changed afterwards; an advertiser who wants to move an existing campaign onto the new method has to build a replacement. That constraint matters for anyone running always-on user acquisition, because the migration cost is a full campaign rebuild rather than a settings edit.

The supported surface is narrow and specific. Unified Attribution currently covers four app event goals: App Install, App Purchase, App Add to Cart and App Sign Up. Creative formats are limited to Single Image or Video, Story Ads, Collection Ads and Playables. Bid types include Auto Bid, target cost per action and Max Bid. Anything outside that set stays on Snap's existing methods.

Snap's documentation sets Unified Attribution against two alternatives in a direct comparison. A SKAdNetwork campaign uses Apple's SKAN schema for shared conversion events, runs on a 30/1 attribution window as the SKAN default, and requires SKAN tracking. A Snapchat Outcomes campaign uses MMP events, complemented by the Conversions API where implemented, and offers 28/1 and 7/0 windows described as in beta. Unified Attribution takes MMP events, attribution windows and settings wholesale, and offers no window option of its own.

That last detail produces a visible artefact in the interface. Because a Unified campaign runs on the MMP default window, the attribution window control in Customize Columns is greyed out when such a campaign is selected on its own. Select a mix of Unified and non-Unified campaigns and the control defaults to the non-Unified setting, while the Unified campaigns continue to use the MMP window regardless of what the dropdown displays. Two rows in the same table can therefore be counted over different periods without the interface saying so.

Configuration prerequisites differ by vendor

The two supported partners carry different setup requirements, and both include changes that have to be made inside the MMP rather than inside Snap.

For Adjust, the documentation states that Extended Privacy Measurement must be disabled, and recommends enabling probabilistic modeling. For AppsFlyer, the advertiser must be using Advanced SRN, must disable the AppsFlyer AAP toggle at the Snap network level, and is advised to enable view-through probabilistic modeling. On the Snapchat side, both paths carry the same recommendations: an account that is Advanced Conversions ready, a Signals status of OK or Good, and a minimum of 10 daily Unified conversions from Snapchat campaigns.

Volume is treated as a gating condition rather than a nicety. Snap's best-practice guidance asks advertisers to hold every other variable constant when comparing test and control campaigns, to avoid touching attribution windows or MMP settings while a campaign is learning, and to allow at least four days before making changes, extending to seven days for target CPA campaigns. Low-volume campaigns, the documentation concedes, can still yield directional learnings but perform less stably.

One provision addresses a situation that is more common than it sounds. If an app is connected to more than one supported MMP, Unified Attribution selects whichever partner records the highest volume of App Install events and treats it as the primary source. As event volumes shift between vendors, that primary designation and the resulting campaign behaviour can change without an advertiser touching anything. Snap directs advertisers to Events Manager to inspect event volumes across connected MMPs.

The beta figures, and what they cover

Snap released four performance claims from the beta period, each carrying a footnote marker whose supporting note was not included in the material distributed with the announcement.

The headline number comes from an unnamed real money gaming advertiser that ran Unified Attribution across Casino and Sportsbook apps in spring 2026, with performance evaluated by AppsFlyer. Snap reports a 26% lower cost per acquisition compared with SKAN-optimised campaigns. The comparison is against SKAN rather than against Snap's own Snapchat Outcomes method, which is a meaningfully different baseline given that SKAN delivers delayed, aggregated postbacks subject to privacy thresholds.

Mohegan Sun Online Casino is the one named AppsFlyer case. By week four of a comparison between Unified Attribution and its business-as-usual approach, Snap reports the product drove an 89.8% increase in return on ad spend and a 77% reduction in cost per install.

Two Adjust cases follow. Deblock, a financial application operating in France, combined Adjust signals with Snap first-party data and recorded a 25% lower effective cost per install, a 28% lower cost per completed onboarding, 43% more app installs and 31% more users reached. Codeway, described as a mobile app publisher, reported an 18% lower cost per install and 1.8 times higher scale while holding cost per purchase steady.

Each figure originates with Snap and its measurement partners rather than with an independent evaluator, and each describes a single advertiser rather than an aggregate across the beta cohort. Read alongside the wider debate over incrementality, the numbers describe attributed outcomes under a changed counting rule, not a demonstrated lift in business volume. A campaign that switches from SKAN accounting to MMP accounting will report different conversion totals even if user behaviour is identical, and the direction of that difference is not neutral: MMP-modeled attribution generally credits more conversions than SKAN postbacks do. The endogeneity problem that separates observed return from causal return is untouched by any of this.

Two descriptions of the same mechanism

The announcement material and the product documentation do not describe Unified Attribution in the same terms, and the gap is worth stating plainly.

Snap's help documentation opens by saying that Unified Attribution uses MMP-modeled signals in combination with Snapchat attribution, and that this ensures optimisation is powered by Snapchat's attribution while outcomes are brought in line with advertisers' MMP reporting. Further down the same page, the FAQ states that metrics labelled Unified use MMP-modeled signals for optimisation and reporting inside Snapchat Ads Manager. One passage places Snapchat's attribution at the centre of optimisation; the other places MMP-modeled signals there.

AppsFlyer's characterisation leans firmly toward the second reading. "For the first time, Snap's optimization engine and advertiser measurement are working from the same data," said Bright Park, AppsFlyer's Principal Partner Development Manager. Park described the gap being closed as one where publishers optimise against their own attribution signal while advertisers make budget decisions from a fundamentally different view.

Snap does not claim the two systems will reconcile. The documentation is unambiguous that Unified Attribution is designed to align more closely with MMP reporting but will not match it one to one, and repeats the point in two separate FAQ entries. Conversion reporting is based on impression time by default, and modelling may be applied where data is limited by privacy requirements. Advertisers are told not to compare standard cost per action against Unified cost per action, because the two are computed differently.

Reporting arrives as a parallel column set

Nothing in the existing Ads Manager reporting changes. Unified figures appear as an additional family of metrics that has to be added manually through Customize Columns, retrieved by searching for the term unified. The set includes Unified App Installs, Cost per Unified Install, Unified Install Rate, Purchase ROAS unified app acquisition, Cost per Unified purchase, Cost per Unified add to basket and Cost per Unified sign-up. Conversion-specific reporting requires opening the matching conversion category and selecting Unified app acquisition for the relevant row, such as Purchases or Purchases Value. The performance chart carries its own metric dropdown with a parallel set.

SKAdNetwork does not disappear. A separate SKAN Background Reporting option can be switched on at campaign creation, after which SKAN appears as a supplemental signal in applicable reporting views. The result is an account that can hold three concurrent accounts of the same campaign: Snapchat's own attribution, the MMP's modelled view surfaced as Unified metrics, and SKAN postbacks alongside them.

The partner roadmap

Adjust and AppsFlyer are the only supported partners at launch. Snap states it is working with Singular, Kochava and Branch to extend Unified Attribution support, with availability and timing varying by partner, and commits to updating the documentation as integrations arrive. Those three, together with the two already live, make up the standard set of certified measurement vendors across the major platforms. Kochava's own certification programme, established in December 2025, named Snapchat among the platforms meeting its integration quality criteria.

"Unified Attribution brings performance data from Snap and Adjust into a single view, helping advertisers better understand results, optimize campaigns, improve ad spend efficiency and make informed decisions that drive growth," said Andrey Kazakov, chief executive at Adjust. Adjust is owned by AppLovin.

Adrian Mulryan, Snap's VP of Global Agency, framed the release around advertiser trust in existing signals, saying app advertisers now have a simpler way to bring together the cross-channel signals they trust with Snapchat's, and describing the early results as meaningful.

Why it matters for app marketers

The pattern Snap is following is by now well established across the platform market, and Snap arrives late rather than early. Meta aligned its value optimisation with MMP measurement in November 2025, applying advertiser-configured reattribution windows from AppsFlyer and a 180-day exclusion window for Adjust and Singular. TikTok routed real-time iOS conversions through Kochava in October 2025 to work around SKAN delays. Reddit placed first-party, MMP and SKAN figures side by side in a single Ads Manager panel in May 2026. OpenAI connected ChatGPT Ads to AppsFlyer and Adjust across seven markets in July 2026. Even Comcast's Universal Ads added five MMPs as a new partner categoryon July 30, 2026, extending app attribution into premium television inventory.

The direction of travel is consistent: platforms are surrendering the definition of a conversion to third parties they do not control. Why would a publisher do that? Because the alternative is a permanent argument with the buyer over whose numbers are correct, and because budget follows the numbers the buyer believes. The independence that makes an MMP worth paying for is the same independence that makes its verdict binding, a dynamic underlined when Google, Meta, Moloco and Unity invested $2.7 billion in AppsFlyer in June 2026 under governance terms explicitly prohibiting preferential treatment in attribution logic.

For Snap specifically, the commercial context is a business that has been repairing its performance advertising credibility for two years. The company's Q1 2026 results showed advertising revenue up just 3% year over year to $1.24 billion, with impression volume rising roughly 17% while effective CPMs fell about 12%. The picture improved in the following quarter, when advertising revenue rose 9% and ad prices gained 10%, though chief financial officer Doug Hott attributed part of that to World Cup demand that would not repeat. Unified Attribution was already listed among the infrastructure investments Snap cited when it published India advertiser momentum data on June 3, 2026, and the broader ads platform overhaul arrived on June 18, 2026 with the Human-First, AI-Enabled suite and a Model Context Protocol server.

The practical consequence for a media buyer is narrower than the framing suggests. Unified Attribution does not make Snap's reporting agree with an MMP dashboard; Snap says so four separate times in its own documentation. What it does is move the target the bidding system optimises toward, so that the events being chased resemble the events the advertiser will later count. Campaign comparisons made across the boundary become invalid, attribution windows stop being an advertiser-side variable, and any account connected to two MMPs carries a silent dependency on which vendor happens to be recording more installs this week. Those are configuration facts, not performance claims, and they are the part of the release that will still be true after the beta figures have aged.

Timeline

Summary

Who: Snap Inc., through its Snapchat advertising platform, with mobile measurement partners AppsFlyer and Adjust. Named commentary came from Bright Park, Principal Partner Development Manager at AppsFlyer; Andrey Kazakov, chief executive at Adjust; and Adrian Mulryan, VP of Global Agency at Snap. Named advertisers cited from the beta include Mohegan Sun Online Casino, Deblock and Codeway.

What: General availability of Unified Attribution, a campaign-level attribution method for App Promotion campaigns that adopts the advertiser's MMP events, attribution windows and settings in place of Snap's own defaults. It supports four app event goals, four creative formats and three bid types, cannot be changed after campaign creation, and is not designed to match MMP reporting exactly.

When: Confirmed on August 31, 2026, following a beta period that included tests run in spring 2026. Unified Attribution was already named in Snap's June 3, 2026 India advertiser disclosure.

Where: Globally, inside Snap Ads Manager, for advertisers connected to AppsFlyer or Adjust. Support for Singular, Kochava and Branch is described as in progress, with timing varying by partner.

Why: Platform optimisation systems and advertiser measurement systems have counted conversions differently since Apple's App Tracking Transparency framework degraded deterministic iOS attribution in 2021, leaving buyers to reconcile two sets of numbers before allocating budget. Snap is aligning the signal its bidding system pursues with the signal the buyer already trusts, following comparable moves by Meta, TikTok, Reddit and OpenAI, at a point when the company is working to rebuild advertiser confidence in its direct-response performance.