Word and trivia games: the quiet UA machine

Word and trivia games are the most consistently profitable acquisition category in mobile, and the reason is not that they are cheap to make or easy to advertise. It is that the auction prices their audience wrongly. Most bidding systems predict value from in-app purchase behaviour, and the people who play word games rarely buy anything. To an IAP-optimised competitor they look like low-value traffic. To a word game publisher earning from impressions, the same person is a reliable, long-tenured, ad-tolerant customer. The gap between those two valuations is the whole business.

That is a claim a smart buyer could push back on. If the mispricing were that obvious, the argument goes, it would have closed years ago. It has not, and the mechanism explains why.

How the auction gets it wrong

Networks learn who to show an ad to from the conversion events advertisers send back. The heaviest spenders in casual, the strategy and puzzle titles that Sensor Tower's State of Gaming 2026 report credits with leading growth, send purchase events. Their models therefore learn that a fifty-eight-year-old on a mid-range Android device who has never bought a gem pack is not worth much, and bid accordingly.

A word game does not need that person to buy anything. It needs them to open the app most days for a year and sit through rewarded and interstitial placements. Those sessions are real revenue, but they are revenue paid by the ad networks themselves, and it arrives as eCPM rather than as a purchase event the network can see and learn from. The result is that word and trivia publishers are bidding for people the rest of the market has already decided are not worth bidding for.

Two data points from earlier this year sit behind this. The Axon by AppLovin and Kantar study published in March found that affluent, older households were unusually receptive to in-game advertising. And Adjust's March report on gaming sessions recorded casual sessions up substantially year on year, with retention rather than volume described as the core growth strategy. Retention is the entire word game economy.

The spread, not the CPI

Practitioners in this category do not really manage CPI. They manage a spread: what a user earns in ad revenue over their lifetime against what the user cost to acquire, with the awkward detail that both sides of the spread are set by the same handful of companies.

A workable decision rule, expressed as an illustrative example rather than a benchmark:

• Take day-30 ad ARPU for the cohort, say a hypothetical 40 cents.
• Divide by day-30 retention-weighted CPI, say a hypothetical 30 cents.
• If the ratio holds above roughly 1.2 for two consecutive weekly cohorts, the campaign is allowed to scale by a fixed increment.
• If it falls below 1.0 for two cohorts, spend is cut, not creative refreshed, because the problem is almost always the buy side of the spread and not the ad.

The numbers are invented for the sake of the rule. The rule itself is what matters: it treats acquisition as an arbitrage position and not as a marketing campaign, and it forces the team to look at both legs of the trade at once.

The second-order problem nobody prices

Here is the part most coverage misses. Because the customer and the supplier are the same firms, word and trivia publishers are structurally short the ad market. When eCPMs soften, their revenue falls and their acquisition costs do not fall at the same rate, because CPIs are set by IAP-driven bidders whose demand is unaffected. The spread compresses from both sides.

Unity's decision to shut the ironSource Ads network on April 30 is a live example. A network leaving the market removes demand from the monetisation side for publishers who were mediating it, while doing very little to lower CPIs for anyone buying on the remaining platforms. Publishers who had treated their ad revenue as a steady annuity found one of the payers had left the table.

This is why the largest word and trivia operators, the Tripledot Studios of the world with their Solitaire and Woodoku catalogues, spend so much effort on portfolio cross-promotion and on owning their user data rather than renting it. Adjust's announcement in March that it had migrated Tripledot's thirty-plus live games with their full install history intact was a reminder that a mature casual portfolio's most valuable asset is often the historical record of who its players are. Cross-promotion is the one acquisition channel whose cost is not set by someone else.

What this means for a team entering the category

The advice that follows is unfashionable, because the category rewards patience over cleverness.

Do not import a hybrid-casual creative playbook. Word game creative that mimics the failed-attempt hooks of puzzle titles usually attracts puzzle players, who churn when they discover there is no meta. The creative that works tends to be quiet: a partially filled grid, a hand hovering over a letter, no music. It looks unambitious and it filters for the right person.

Build the spread dashboard before the first campaign, with ad ARPU by cohort on one axis and effective CPI on the other, and make the weekly review about the ratio rather than either number in isolation. Teams that report CPI alone tend to cut spend at exactly the moment eCPMs recover.

Treat mediation waterfall changes as UA events. A new bidder entering or leaving mediation moves the revenue leg of the spread, and campaigns that were marginal can flip either way within a fortnight without anything changing in the ad account.

And accept that the category's ceiling is set by inventory demand from other people's games. Word and trivia publishers are, in effect, wholesaling attention to the rest of the industry. The business is very good when the rest of the industry is buying, and there is no creative genius that changes that when it is not.

Related archive reading

• Why simulation games over-index on TikTok
• Web3 games and UA: what remains