Snap announced a bundle it calls the Commerce Power Pack on September 10. One item in it, Third-Party Web Optimization, is “currently in open beta,” and Snap’s post says it “combines first- and third-party signals together, helping optimize towards the third-party purchase outcomes” that web advertisers already use to evaluate cross-platform performance. Snap says “early results from partners like Google Analytics and Northbeam have been encouraging” and gives figures for both from its internal data.
What is Snapchat’s Third-Party Web Optimization?
Third-Party Web Optimization is an open-beta option for Snapchat web campaigns, announced September 10 in Snap’s Commerce Power Pack. It combines first- and third-party signals to help optimize toward the third-party purchase outcomes advertisers already use to evaluate cross-platform performance, and Snap names Google Analytics and Northbeam among the partners with early results.
What Shipped, and at What Stage
The Commerce Power Pack bundles four items at different stages. Dynamic Product Ads (DPA) for Sponsored Snaps is “available now for all global advertisers,” bringing catalog product ads directly into Chat, which Snap calls its “most-used surface.” In June, Snap introduced brand AI agents in Chat; the September 10 pack adds catalog product ads to that same surface, not the agent format itself.
Value Optimization is in beta for both web and app advertisers, letting campaigns “optimize for return on ad spend and focus on purchases that drive more value”; Snap says it “can also be used in tandem with Unified Attribution,” a term the announcement never defines. Omnichannel Optimization, pooling web and app purchases in one campaign, is in alpha testing only. Third-Party Web Optimization is “currently in open beta”; of the four items, only DPA for Sponsored Snaps is described as available now for all global advertisers.
An Advertiser’s Own Scoreboard Becomes the Target
An advertiser can use a third-party analytics platform such as Google Analytics to check Snap-reported results. Snap’s post describes the new capability this way: “For web advertisers, Third-Party Web Optimization combines first- and third-party signals together, helping optimize towards the third-party purchase outcomes they already use to evaluate cross-platform performance.” For a campaign using it, the outside check becomes the target the optimization steers toward.
That is not a claim that Snap games or inflates the third-party numbers; nothing in Snap’s post says how the GA or Northbeam data reaches the optimization. It is a claim about evaluation design. An advertiser who uses GA’s Snapchat numbers as an independent read on Snap-reported performance, and also runs a campaign on Third-Party Web Optimization, would be grading that campaign by the kind of third-party purchase outcome Snap says the capability helps optimize toward. A check meant to catch an underperforming campaign needs to sit somewhere the campaign isn’t optimizing toward.
What the Numbers Cover
Each figure in the table below carries its own numbered footnote in Snap’s post, and the footnotes do not describe one consistent test.
| Figure | What it measures | Source window / footnote |
|---|---|---|
| 23% lower third-party CPP, 37% higher third-party ROAS | With Google Analytics | Snap internal data, May 18 – June 1, 2026 (fn4) |
| 37% lower CPP, 64% higher ROAS | With Northbeam, “across first-time purchasers”; Snap does not label these figures third-party | Snap internal data, May 18 – June 7, 2026 (fn5) |
| 22% improvement in 7-day CPA | DPA campaigns optimized for Pixel Purchase, autobid, “7-0 attribution window” | Snap internal data, July 17 – 26, 2026 (fn2) |
| 85% of Snapchatters use Chat regularly | Stated survey response, not observed platform behavior | 2024 Alter Agents study commissioned by Snap, n=12,000 (fn3) |
The GA and Northbeam windows differ, two weeks against three. Snap describes the Northbeam figures as “across first-time purchasers” and gives no purchaser scope for the Google Analytics figures. Snap discloses neither the advertiser count behind either result nor how either test was designed, so the two sets of results should not be compared, added together or averaged.
Snap’s Post Doesn’t Say How the Data Gets In
Snap’s post names Google Analytics and Northbeam as partners and says the capability “combines first- and third-party signals,” but it does not say how Google Analytics or Northbeam data reaches Third-Party Web Optimization. A September 13 search of Snap’s Business Help Center for “third-party web optimization” returned no article describing the feature. Neither the post nor that search describes the data path.
Check Your Own GA Reporting Before Judging Snapchat by It
One thing is within an advertiser’s own control regardless of how Third-Party Web Optimization works: tagging Snapchat destination URLs so Google Analytics can identify Snapchat campaigns before GA is used to judge them. Snap’s own Google Analytics FAQ tells advertisers to “ensure that your UTM parameters are configured properly to attribute conversions to Snapchat,” recommending the parameter-value pair utm_source=snapchat. The same FAQ warns that redirects on landing pages “might prevent analytics code from executing and properly recording that the traffic came from a Snapchat campaign,” and gives an example in which “the session will be recorded as organic.” Keeping tagging consistent across every destination URL is easier with a UTM builder that lowercases source, medium and campaign values by default.
The same FAQ explains why Ads Manager and GA can disagree: third-party analytics platforms “are generally not designed to measure cross-device conversions, but rather focus on same-session conversions that occur immediately after clicking on an ad,” while the Snap Pixel “can capture cross-device conversions up to 28 days after a user has seen an ad,” and by default its post-click window is 28 days and its post-view window 1 day. The FAQ describes that gap without reference to Third-Party Web Optimization. For a campaign that does use the capability, the independent read (a holdout group, a marketing-mix model, or a tool outside Snap’s optimization loop) needs choosing before the campaign is graded, not after.
