The channel you weren't tracking just became too big to ignore
While DTC marketers have spent 2026 managing Meta CPM increases and Google's AI takeover of search, AppLovin went from referral-only experiment to the third largest ad network in e-commerce budgets.
A Jefferies survey of 30 e-commerce and web advertisers in Q2 2026 found that AppLovin now accounts for 11.1% of advertiser budgets — up 169 basis points from Q4 2025. That puts AppLovin at #3 by both budget share and ROAS, ranking just ahead of TikTok. That's not a niche test anymore. That's a meaningful allocation from brands doing real money.
The timing matters for anyone who hasn't tested it yet: AppLovin's Axon Ads Manager opened to all ecommerce advertisers globally in June 2026, ending a year of referral-only access. Brands above $10M in US GMV get $10,000 in free spend on approval. The door is open. The question is whether the results hold at scale.
What Axon actually does
AppLovin is not a social platform. Axon runs your ads across 140,000+ mobile apps — games, utilities, content apps — and the AI engine controls targeting, bidding, and creative optimization across that inventory without you selecting interests or demographics.
The targeting is behavioral. Axon looks for purchase-intent signals in app behavior and bids on impressions likely to convert. You're not building audience segments. You're feeding assets and letting the model route them.
Creative format is short-form video. You supply the raw material; Axon handles placement and optimization. Higher creative volume drives 7-23% better performance on Axon, according to CommonThread Collective's June 2026 data. The more ad variants you can produce, the more optionality the model has to find converters.
The results, and what the independent data actually shows
AppLovin published case studies from early access DTC brands. MaryRuth's, a supplement brand, beat its incremental ROAS target by 49% (per AppLovin for Business). Portland Leather drove over 70,000 incremental purchases. PROOF reported significant incremental sales lift.
These are AppLovin's own numbers — directional, not definitive. The Jefferies survey data carries more weight: that's a third-party view of what 30 actual advertisers are doing with their real budgets.
The independent incrementality data from Haus is the most useful benchmark. Haus analyzed 15 months of test data (January 2025 through March 2026) across DTC and omnichannel brands — first-time tests beat the typical median result 53% of the time, per Haus. That's a favorable rate for a first test. But as brands push more budget and run repeat tests, results converge toward the median. AppLovin is no longer the near-automatic win it was for the small group that had early access.
Run the first test. Don't assume the result scales without measuring it.
Why the attribution looks clean even when the increment is unclear
Axon reports on a 7-day click window with no view-through attribution. That's more conservative than Meta's default 7-day-click-plus-1-day-view setup, and it means AppLovin's dashboard doesn't claim conversions from people who only saw an ad.
But conservative attribution is not the same as incremental attribution. The real question isn't "did they click?" — it's "would they have bought anyway?"
Mobile app inventory is a different audience pool than Meta or Google. That separation reduces the probability of double-counting the same customer, but it doesn't eliminate it. Someone in your Meta retargeting pool also plays mobile games. If Axon reaches them three days before a purchase they were already planning, the click doesn't make that incremental.
This is exactly what the Haus data is measuring. The 53% first-test win rate means that for most new advertisers, real incremental lift exists. For brands on their second or third tests with more budget committed, that margin compresses. AppLovin works best as a new-customer acquisition layer, not as a retargeting overlay on existing intent.
Four things to have ready before you allocate budget
First, creative production capacity. Running fewer than 10 short-form video variants per month limits Axon's optimization window before the algorithm has enough data to differentiate. The 7-23% performance range from creative volume isn't theoretical.
Second, MMP integration. Axon integrates with AppsFlyer and Adjust. Before launch, confirm your post-purchase conversion event is firing accurately — a broken attribution setup on a new channel takes weeks to diagnose and corrupts the test data you need to make a scaling decision.
Third, audience signal quality. AppLovin uses your first-party data — customer lists, lookalike seeds — as targeting inputs. Clean email and phone matching against AppLovin's network is the difference between a broad behavioral audience and one anchored in actual buyer behavior.
Fourth, a separate campaign structure for Q4. Axon doesn't have native seasonal bid modifiers. If you're launching in August to build the channel before holiday season, structure your peak-spend campaign separately from day one. A campaign optimized at summer CPM levels won't maintain efficiency when Q4 inventory costs climb.
If you want to see how your current channel spend compares to DTC benchmarks and where your marginal dollar is most likely to find returns, the free audit at Gromerce will give you that picture in three minutes.
AppLovin's referral-only era is over. The early-mover advantage gap is closing every quarter. Running the test now is better than explaining to your Q4 review why you didn't.
Sources: CommonThread Collective, Jefferies Q2 2026 ecommerce advertiser survey, Haus incrementality research (via CommonThread), AppLovin for Business, July 2026
