AppLovin Sees Early E-Commerce ROAS Ramp, Mid-Market Focus in Q2 2026

AppLovin (APP): Interpreting e-commerce ROAS feedback bias, and whether the “large advertisers hesitation” has changed

1) Are the e-commerce “we’re not at ROAS ceiling” comments biased, or representative?

What management says is actually observable

Management’s framing is that AppLovin is still in an early stage of the consumer/e-commerce categories—meaning advertisers may not have yet reached their “efficient frontier” of spend versus ROAS because the platform is still building data and model maturity. In consumer/e-commerce, they explicitly characterize the business as “still early in evolution,” with “less data penetration” and a “less sophisticated model” than gaming, which is “pretty mature” with “a lot of data” and a “more developed” architecture. ¹

They also connect the idea of ROAS “ceilings” to a gradual ramp rather than immediate saturation: advertisers “launched this product” are being described as running budgets 1–4 quarters ahead, with “most of their budget” going to “social and search,” implying a practical adoption ramp that can mask the true ceiling for AppLovin at any point in time. ¹

Evidence that their advertiser base is not necessarily hitting a ceiling (and why)

Management’s explanation is twofold:

1. Adoption/learning effects: They say AppLovin is a “new bucket” (a testing category) and “to graduate up takes time,” and that performance should improve faster than mature platforms because advertisers are “indexing lower” early on and ramp as they “invest more,” “get comfort,” and run “incrementality studies.” ²
2. Scaling constraints being internal to the advertiser/market testing stage: They state that even where advertisers see “really good ROAS,” “they’re not even spending at their ceiling yet,” and also claim they do not yet have “excess budgets” to remove. ³

These points imply that if you’re hearing “not at ROAS ceiling” feedback, management believes it is consistent with the stage of the category (testing → graduation) and with the fact that advertisers are still ramping budgets rather than having fully optimized and saturated spend. ²³

Potential for positive bias (and what management indicates about the range of advertisers)

There’s not a direct “survey of all advertisers” included in the excerpts, so the most intellectually honest conclusion is: the excerpts support the existence of an early-stage ramp dynamic, but they do not quantify whether feedback is skewed toward a subset of high-performing advertisers. ²¹

However, the excerpts do provide indirect context suggesting AppLovin tries to manage the composition of advertisers it works with—potentially affecting the “tone” of feedback:

• In e-commerce, they describe going through partnerships to get “very targeted customers onto our platform” rather than “just start by buying ads and bringing in long tail.” ⁴
• They also describe a strategic focus on mid-market first, describing it as a “sweet spot” given that small shops may not clear goals with limited spend, while mid-market brands are willing to absorb “learning cost” and “scale.” ¹
• They further say they are targeting “mid-market brands” and not the “very, very big” first or the “SMB or long tail” (which they call “harder for us to make work right now”). ⁵

Together, these suggest management may indeed have a sampling effect: conversations could be with advertisers they deem more likely to see success (mid-market) or that were onboarded via targeted partnerships—meaning feedback could skew positive relative to the broader universe of potential advertisers. ¹⁵⁴

But management’s “not at ceiling yet” narrative is presented as a general ramp thesis for the category (learning/graduation/time and faster improvement as data builds), not merely a story about a tiny winner set. ²³

Bottom line on the first question:

• The excerpts support management’s claim that e-commerce/consumer advertisers can be below a spend/ROAS ceiling because the category is early and advertisers are still ramping and learning. ²³
• The excerpts also indicate they deliberately focus on mid-market and targeted onboarding, which can create positive selection bias in what advertisers are in their conversation set, even if the underlying ramp logic still applies broadly. ¹⁵⁴

2) Has the “hesitation to work with larger advertisers” changed in the last 3–6 months, especially as some mobile game advertisers are large?

What they said about focus across advertiser tiers

In the excerpts, management articulates a clear tiering strategy:

• They say they are not focused first on the “very, very big” because they want to “get to the head of the market later,” while also not focused first on the “very, very small” SMB/long tail because it is “harder for us to make work right now.” ⁵
• They also tie the strategy to data/model maturity and reputation: mid-market execution builds “data coverage” and gives time for “the model is more sophisticated,” plus time for reputation to “seep into the bigger parts of the market.” ⁵

So, in the excerpts provided, the hesitation (or at least the phased strategy) toward “very, very big” customers is still present, not explicitly reversed. ⁵

Does this conflict with “working with larger advertisers” because some mobile gaming advertisers are large?

Management separately describes gaming as “pretty mature” with sophisticated architecture and strong performance even with new games launching on the platform (hitting return goals “really, really quickly” and “almost no learning budget”). ¹

That implies that in gaming—where the platform is more mature—they may face less friction with large spenders (because the model is already sophisticated and data-rich), even if in consumer/e-commerce they prefer to go mid-market first. ¹

In other words:

• Their “large advertiser” caution appears category-dependent in the excerpts: they emphasize consumer/e-commerce is early and thus mid-market is a sweet spot, while gaming is more mature and can support launches quickly. ¹⁵
• The excerpts do not state that “hesitation” has disappeared in the past 3–6 months; rather, management says their execution plan remains “mid-market first” to broaden later as the model matures and reputation expands. ⁵

Any signal of change in the direction of larger advertisers?

The closest supporting evidence for “less hesitation” is indirect:

• They say advertisers on the platform are “not even at a ceiling of what they can spend” given results they’re seeing, suggesting successful customers (potentially including larger ones) have room to scale. ⁵
• They also describe gaming’s maturity as enabling quick goal achievement even for new games, which could make working with large gaming advertisers more feasible than in consumer/e-commerce. ¹

But again, the excerpts do not explicitly confirm a changed stance toward large advertisers in the last 3–6 months; they primarily reaffirm the phased strategy for expanding upmarket. ¹⁵

Bottom line on the second question:

• In consumer/e-commerce, the excerpts continue to describe a deliberate phasing: prioritize “mid-market brands” now, and “head of the market later,” not the “very, very big” immediately. ⁵
• In gaming, however, management emphasizes maturity and low learning budget, which likely reduces friction with larger advertisers there; but the excerpted evidence does not equate that to a stated reversal of “hesitation” broadly. ¹

Overall synthesis (directly addressing both questions)

1. ROAS ceiling feedback in e-commerce: Management argues advertisers are not yet at spend/ROAS ceilings because the platform/category is early and advertisers need time to graduate/testing-to-optimization. ²³ At the same time, they also highlight targeted onboarding and a mid-market sweet spot strategy, which can create positive selection bias in the advertiser feedback you hear. ¹⁵⁴
2. Large advertiser hesitation and recent change: The provided excerpts reaffirm a phased go-to-market that deprioritizes “very, very big” customers in consumer/e-commerce for now, aiming to broaden after executing and maturing the model/reputation. ⁵ There’s no explicit evidence that this has changed in the last 3–6 months, though gaming’s maturity suggests that larger gaming advertisers may be more straightforward operationally than in consumer. ¹