The misleading-ad arbitrage and why it still pays

Misleading game ads are not a failure of taste or ethics that better teams have grown out of. They are a rational response to a specific economic structure, and they will persist for as long as that structure holds. If you want fewer of them in your own account, or in your genre, the useful move is to understand exactly which conditions make the arbitrage pay and to watch for the ones that are changing.

The thesis is uncomfortable: in a game where a small minority of players generate most revenue, an ad only has to be honest for that minority. Everyone else it disappoints costs almost nothing. That is why the practice survives every wave of criticism, and why moral pressure alone has never moved it.

The mechanism

Start with payer concentration. In most free-to-play titles the top few percent of players produce the bulk of in-app revenue. The other ninety-plus percent install, play a little, and leave without spending. From the business's point of view, they were always going to be close to worthless. Disappointing them changes little on the ledger.

Now put a misleading hook in front of that funnel. Response rate rises, CPI falls, install volume climbs. The overwhelming majority of the new installs churn quickly, but they were going to churn anyway. The question that determines whether the arbitrage pays is narrow: does the misleading ad recruit future payers at a lower cost per payer than the honest one, once you account for the minority who stay?

Often it does, for a reason that feels perverse. The people who become payers in a base-builder are frequently the ones who would have installed under almost any hook and then discovered the actual game. The misleading ad does not repel them; it just brings a much larger crowd of tourists along with them at a lower unit price. Cost per payer falls even though cost per retained player rises.

That is the arbitrage. It is not a trick played on networks. It is a trick played on the ninety percent, and the ledger does not record their disappointment.

What is narrowing it

Two forces work against the arbitrage, and both are visible in this week's news.

The first is the shift of spend toward delivery models that optimise directly to revenue events. AppLovin's Q4 results on Wednesday showed revenue up 66 percent year on year, most of it from algorithmic buying that bids to return rather than to install. Unity's same-day results put Unity Vector at 56 percent of its Grow revenue. When the model is learning from purchases, an ad that recruits a huge non-paying crowd is scored against that crowd's near-zero value. The tourists start to cost something, because they dilute the signal the model is trying to learn from.

The second is that ROAS-optimised delivery makes the honest ad's disadvantage smaller. If the honest hook recruits fewer installs but a higher payer share, a purchase-optimised model will learn to serve it to the right people, and the CPI gap that made the misleading hook attractive matters less because nobody is buying on CPI anymore.

Neither force has closed the arbitrage. Both have narrowed it, and both will keep narrowing it as more of the market's spend runs through models that see revenue.

The costs booked elsewhere

The trade-off most coverage misses is that the misleading advertiser does not bear all of the cost. Some of it is booked to the genre.

Store ratings absorb part. A game whose review page is full of complaints that the ad was fake pays a conversion penalty on every impression, including the honest ones, for as long as those reviews are visible. Platform policy absorbs another part: enforcement against misleading creative tends to arrive as blunt rules that constrain everyone, not just the offenders, because reviewers cannot adjudicate intent at scale.

The largest cost is trust erosion at the genre level. When a category becomes known for ads that do not match the game, viewers discount every ad in that category. The honest advertiser's response rate falls because of the dishonest advertiser's behaviour. It is a commons problem, and like most commons problems it is stable right up until it is not.

There is also the audience-model problem, which I wrote about last month in "The hook is a promise; retention is the invoice": a network trained on a misleading hook builds a model of the wrong player, and every later creative inherits that model.

A break-even test you can run

For a team deciding whether a borderline concept is worth running, here is a decision rule that uses numbers you already have.

Take the concept's projected payer share and cost per payer from its test cohort, and compare them with your honest baseline. Then apply three adjustments:

• Multiply the misleading concept's cost per payer by one plus your estimate of the ratings penalty as a share of conversion rate, because those tourists will review the game.
• Add the expected cost of a creative rejection and re-submission on your largest network, weighted by how often that network has flagged similar creative.
• If more than half of your spend runs through revenue-optimised campaigns, discount the CPI advantage by at least half, because the model is not buying on CPI.

With purely illustrative inputs: an honest concept at a cost per payer of 60, a misleading one at 48, a 10 percent ratings penalty, a 5 percent rejection cost, and 70 percent of spend on ROAS-optimised campaigns. The adjusted cost for the misleading concept comes out around 55 before the CPI discount and closer to 58 after it. The arbitrage has gone from a 20 percent edge to a rounding error, with policy risk left over.

The three-minute test

The simplest screen, before any of the arithmetic, is to watch a stranger play the first three minutes after seeing the ad and ask whether they would feel cheated. If the answer is yes, the concept is in arbitrage territory, and the question is not whether it will pay this quarter but how much of the market's spend has to move to revenue-optimised delivery before it stops paying.

Judging by this week's earnings, that share is rising faster than most creative roadmaps assume.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• A field guide to misleading mechanics
• Sound Design for Mobile Video Ads Most Players Never Hear