Who Is Actually Buying Your Rewarded Video Ad on Web?

Every rewarded video sells attention to someone. On mobile, that someone is usually a competitor. On the web, it usually is not — and that difference should change what you optimize for.

The 300 Milliseconds You Never See

A player finishes a run, comes up short on coins, and clicks “Watch to continue.” Between that click and the first frame of video, something fairly remarkable happens.

A bid request leaves your page carrying a small bundle of signals: the ad slot’s dimensions, the page context, a coarse geographic region, the consent string your CMP produced. Within a few hundred milliseconds, dozens of buyers evaluate that bundle, decide what the impression is worth to them, and respond. An auction clears. A winner is chosen. A price is set. The video buffers, and your player watches it.

Most web publishers know that process exists in outline. Very few could name who won, who funded the bid, or what that buyer was actually trying to accomplish. And that gap matters more than it sounds, because nearly every number publishers do obsess over — eCPM, fill rate, revenue per session — is downstream of a buyer decision they have never examined.

So it is worth asking the question directly: when someone pays for a rewarded impression in your web game, who is that, and what do they want?

The Three-Layer Buyer Stack

Most publishers collapse “who is buying” into a single answer — usually the name of whichever network shows up in their dashboard. That answer is not wrong, but it describes only the outermost layer of a stack with three distinct levels, each of which affects your revenue differently.

The important observation is this: your eCPM is set almost entirely at layers two and three, while publishers spend nearly all their attention on layer one. Switching networks changes who clears the auction. It does not change who is willing to fund a bid on your audience, or how much that audience is worth to them. If your rewarded revenue is underperforming, the diagnosis usually lives deeper in the stack than the dashboard you are staring at.

The layers also behave very differently across platforms — which is where mobile and web split apart.

How Mobile Answers the Question

The mobile answer is well documented, and the scale involved is genuinely large. Sensor Tower’s Gaming Deep Dive on ad monetization reports that mobile gaming ad monetization generated more than $12 billion in revenue in 2025 across 19 measured markets, against roughly 2.4 trillion ad impressions served. That is not a supplementary revenue line; it is a demand engine comparable in importance to in-app purchases.

At layer one, that market is strikingly concentrated. Sensor Tower’s tracked network set includes AppLovin, Google’s AdMob, Meta Audience Network, Mintegral, Unity Ads, InMobi, Supersonic and Digital Turbine — and of those, AppLovin and AdMob together account for roughly two-thirds of mobile gaming ad revenue, a combined share that has been climbing at the expense of smaller independent networks. For most mobile studios, two logos are quietly responsible for the majority of the checks.

But layer two is the more revealing one. A very large share of the money flowing through mobile rewarded auctions is user-acquisition spend from other app developers. A casual puzzle studio bidding into your rewarded slot is not building brand awareness. It is trying to acquire a player who resembles your best players, using your game as the delivery vehicle. Sensor Tower’s data on the reach of this channel is blunt: ad-supported titles drove 24.6 billion downloads across its tracked markets in 2025, roughly 83% of all mobile game downloads — meaning advertising is not a side effect of the mobile business model, it is the discovery funnel itself.

Brand advertisers do participate, and their share rises around cultural moments, but the day-to-day engine of mobile rewarded demand is developer-to-developer. Which means that on mobile, when you open a rewarded slot, you are also opening a recruiting channel into your own player base.

Why the Web Answer Is Structurally Different

It is tempting to assume the same picture holds in the browser, only smaller. It does not. Four structural differences change who shows up to bid on a web or HTML5 rewarded impression, and none of them are cosmetic.

1. There is no device graph

The mobile UA loop runs on deterministic identity. A device identifier links the ad impression to the install, an attribution provider closes the loop, and the buyer learns exactly what that impression produced. In the browser, that loop does not reconstruct cleanly. There is no equivalent persistent device ID, no install event to attribute, and increasingly no third-party cookie to fall back on. Buyers whose entire bidding model depends on deterministic install attribution simply do not have a mechanism to bid confidently on web inventory — so many of them do not bid on it at all.

2. The demand path is different

Mobile rewarded demand arrives through SDK-mediated networks whose core business is app install advertising. Web rewarded demand arrives through programmatic exchanges, SSPs and DSPs — the same plumbing that serves the rest of the open web. Those are different buyer populations with different mandates. The media buyer bidding through a DSP on your HTML5 game is frequently the same buyer placing video on publisher sites, working from a video and CTV budget rather than a UA budget.

3. Brand and mid-funnel budgets make up more of the mix

When install attribution is unavailable as the default KPI, buyers fall back on the measures that do work in a browser: completion rate, viewability, time in view, and brand study lift. Rewarded video is unusually strong on precisely those metrics — the format is opt-in, full-screen, sound-on by default, and completed at rates the rest of the open web cannot approach. That makes web rewarded inventory legible to brand and mid-funnel budgets in a way most web video inventory is not, and it explains why the buyer mix skews differently than it does in an app.

4. Context does the targeting work

Under TCF 2.3 and consent-based frameworks generally, the practical targeting layer on the web is contextual: what game this is, what genre, what audience it implies, what geography, what device class. Publishers often read this as a pure loss relative to mobile’s behavioral precision. It is more accurate to call it a different currency. Contextual signals are stable, do not degrade with consent rates, and are exactly what brand buyers have always purchased against. A well-defined game with a clear audience is a legible contextual buy — regardless of how many users granted consent.

The takeaway is not that web rewarded demand is better or worse than mobile. It is that the buyer on the other side of your auction wants something different, measures something different, and responds to different qualities in your inventory. Optimizing a web rewarded placement using mobile assumptions is a reliable way to leave money on the table.

The Long-Tail Finding — And Why It Is Stronger on the Web

One of the more encouraging findings in Sensor Tower’s report concerns concentration. Games ranked below the top 1,000 by revenue capture roughly 29% of total ad revenue, compared with only about 9% of total in-app purchase revenue at that same tail of the market.

That gap says something structural about the two models. In-app purchase revenue rewards deep monetization systems, live-ops tuning, brand recognition and years of accumulated scale — advantages that concentrate naturally among the largest publishers. Advertising does not discriminate the same way. An impression is worth approximately what the audience and the placement quality make it worth, whether it sits inside a global hit or a niche title with a few thousand daily players.

On the web, that equalizing effect is stronger still, for reasons that have nothing to do with ad tech:

• There is no app store ranking to gate your discovery, and no featuring decision to wait on.
• There is no install step, so the friction between “someone saw your game” and “someone is playing your game” is a page load.
• Distribution runs through portals, embeds, search and social — channels where a well-made small game can outperform its catalog position.
• Your inventory is judged on completion and context, not on the size of the account behind it.

The practical reading for a small or mid-size web studio: you are competing on inventory quality, not on leaderboard position. That is a materially more winnable contest, and it is the reason advertising monetization deserves to be treated as a primary strategy rather than a consolation prize for studios that could not crack in-app purchases.

The Competitor Exposure Audit

Lower competitor exposure on the web does not mean zero. Game portals, publishers and cross-promotional networks do buy web inventory, and some of those buyers are in your genre. The difference is that you have fewer levers than a mobile mediation stack provides — so the ones you do have are worth using deliberately. Five checks, in rough order of impact:

1. Advertiser category and domain blocks

Confirm your provider supports blocklists at both the category and advertiser-domain level, and that the list is actually enforced rather than best-effort. Blocking a category you compete in costs you some demand; it also stops you funding a rival’s acquisition campaign with your own attention.

2. Session-level frequency capping

Per-session caps matter more than daily caps for web games, where sessions are short and a single sitting can generate several reward moments. A player who sees the same advertiser three times in one session has been over-exposed to that buyer, not monetized more effectively.

3. Demand-source transparency in reporting

Ask whether you can see, at minimum, buyer category breakdowns in your reporting. If your provider cannot tell you what proportion of your revenue comes from brand versus performance versus competing-game demand, you cannot diagnose layers two and three of the buyer stack at all — which means every optimization you make is aimed at layer one by default. This is one of the concrete arguments for a provider built for web publishers specifically: AppLixir was designed around consent-first, contextual web delivery rather than retrofitted from a mobile SDK, and the reporting reflects a web demand path rather than an app one.

4. Consent-state fill behavior

Establish what happens to your fill rate when a user declines consent. Some stacks collapse; others fall back cleanly to contextual demand. This single behavior can account for a large share of the gap between your theoretical and actual revenue, and it is rarely surfaced unless you ask.

5. Reward-moment placement review

Audit where in the session your rewarded prompts appear. A reward offered at a genuine friction point — a failed run, a locked level, a depleted resource — produces a materially more attentive viewer than one offered on a loading screen. Attentive viewers complete, and completion is the metric your web buyers are actually paying for.

A Choosier Market Rewards Better Placement

There is one more trend worth folding into this, because it changes the direction of travel. Sensor Tower’s H1 2026 Digital Gaming Market Index describes a first half of 2026 in which mobile in-app purchase revenue and time spent were both roughly flat year over year, and mobile downloads fell around 12% — while gaming ad impressions rose 14% and ad spend grew 8%.

Read those numbers together and the picture is clear: more money is chasing the same amount of attention. That is the definition of a more contested media market, and it has a predictable consequence. Buyers under pressure to justify every bid become more selective about which placements, audiences and moments they will pay a premium for. Generic, low-attention inventory gets bid down. Well-integrated, high-intent inventory gets bid up.

For web publishers, that is an argument for a specific kind of investment — not more impressions, but better ones. Make the reward genuinely worth the thirty seconds. Place the prompt where the player already wants something. Cap frequency before the format wears out. Keep the pre-roll experience fast enough that the player does not abandon before the first frame. Each of those choices raises completion, and completion is the currency your web buyers are bidding in.

Frequently Asked Questions

Who actually pays me when a rewarded video completes?

The network, exchange or SSP that won the auction pays you, minus its take rate. That entity is not the advertiser — it is a marketplace operator clearing a bid funded by someone else. The advertiser whose budget set the price is one layer further back, and is usually invisible in standard publisher reporting unless your provider surfaces buyer categories.

Are my competitors buying ads inside my game?

On mobile, very likely — developer-to-developer user acquisition is the dominant demand type. On the web, less so, because the deterministic install-attribution loop that makes that strategy work does not function cleanly in a browser. Some cross-promotional and portal demand still appears, which is why advertiser category and domain blocks are worth configuring rather than leaving at defaults.

Why are web rewarded CPMs different from mobile ones?

Because a different buyer population is bidding, with different KPIs. Mobile CPMs are pushed up by UA buyers pricing against install lifetime value. Web CPMs are set more by brand and mid-funnel buyers pricing against completion and viewability. Comparing a web eCPM to a mobile benchmark is comparing two different auctions with different participants — which is why web and HTML5 rates should be judged against web and HTML5 norms.

Can I block specific advertisers from my rewarded inventory?

Generally yes, through advertiser-domain and category blocklists, though the granularity varies by provider. Confirm before you commit that blocks are enforced at the bid level rather than applied after the fact, and expect a small fill-rate cost in exchange for the control.

Does requiring GDPR consent reduce who can bid on my inventory?

It changes the mix more than it shrinks it. Behavioral demand does depend on consent, but contextual and brand demand largely does not — those buyers are purchasing the game, the genre and the audience it implies. A stack that falls back cleanly to contextual demand when consent is declined preserves most of the revenue; a stack that does not will show a steep drop. That fallback behavior is worth testing directly.

Do I need a large audience to attract quality demand?

Less than most publishers assume. Concentration data shows the long tail captures a far greater share of ad revenue than of in-app purchase revenue, and on the web that effect is amplified by the absence of store ranking and install friction. What buyers price is audience quality and placement context, not the size of the studio behind them.

Note on data: figures attributed above are drawn from Sensor Tower’s “Gaming Deep Dive: Ad Monetization” report and its related H1 2026 Digital Gaming Market Index coverage, and describe the mobile app ecosystem. They are used here as a comparison case; web and HTML5 performance should be benchmarked against web-specific data. Any directional figures should be replaced with AppLixir network data prior to publication.