The Biggest Web Games Platform Runs on 1,500 Titles. Your Portal Does Not Need 10,000.

Poki reaches 100 million monthly players with about 1,500 curated games while aggregator catalogues pass 19,000. Catalogue size is not what makes a portal work.

Sit in on a white-label portal negotiation and count how long it takes before someone asks how many games are included. It is usually the first question and almost always the wrong one. The number gets written into the board deck, the press release and the homepage badge, and then eighteen months later the portal has 6,000 titles, forty of which anyone has ever played twice.

The uncomfortable part is that the public evidence points the other way, and it has for years. The web platform with the most players has the smallest catalogue of the major ones. That is not a coincidence, and it is not a quirk of one company's taste.

📊 Three Web Platforms, Three Very Different Catalogues

Start with what the operators themselves publish.

Poki is the largest destination for browser games by players. Its own 2025 year in review reports 625 million players across the year, 100 million monthly, and 11.1 billion gameplays. It also reports the number that matters here: 227 new games released in 2025. Poki's Google Play listing and a 2026 profile of the company both describe a library of roughly 1,500 curated titles built with 600-plus developers.

CrazyGames sits in the same category with a bigger shelf. Its own site says 50 million players come to play. Published counts of its library vary by source and by how you count — Wikipedia's article cites about 4,500 games, secondary comparison sites quote figures roughly double that. Either way it is several times Poki's catalogue.

GameDistribution is the contrast case. It is not a destination competing for players; it is a supply layer that syndicates games to other people's sites. Its own about page describes 20,000-plus games distributed across 4,000-plus platforms, with new titles added daily. A catalogue that size makes complete sense for a distributor. It makes very little sense as a target for a portal.

These three are not doing the same job, and I am not claiming a clean inverse relationship between catalogue size and traffic. The point is narrower and harder to argue with: the biggest audience in web gaming is being served by the smallest of the three catalogues, so catalogue size clearly is not the binding constraint on growth. Something else is.

🎯 Everywhere Anyone Has Measured It, Play Concentrates

Public play-level data for web portals is thin — none of them publish per-title distributions. The best measured evidence comes from PC and console, and it is brutal.

Newzoo's PC and console report covering 2025, summarised by GameDev Reports, found that 79 games on PC accounted for 80% of all user playtime. On PlayStation it was 74 titles, on Xbox 91. New releases have taken a steady 12-13% of total user time for three years running; Xbox skews highest at 18%. The earlier edition of the same report, presented at GDC 2025, found 67% of PC playtime going to games more than six years old, and the average PC player getting through ten titles in a year.

Valve's own GDC 2026 disclosure tells a similar story from the revenue side: 5,863 Steam games earned more than $100,000 in 2025, but only about 1,700 of those had launched that year. The rest was catalogue.

Two honest caveats. First, this is PC and console, where players commit to long sessions and big installs — a browser portal's session behaviour is genuinely different, and I have not seen anyone publish equivalent web numbers. Second, Newzoo notes concentration actually loosened slightly in 2025, with titles outside the top 20 taking a bit more time than before.

Neither caveat rescues the "more games" thesis. In every catalogue anyone has measured properly, play collapses onto a small head, a long tail earns close to nothing, and old titles hold their audience better than new ones do. There is no reason to expect the web to be the one flat distribution in the industry — and if anything, a portal with a single-screen homepage and infinite alternatives one tab away should concentrate harder, not softer.

🧮 So How Many Games Do You Actually Need?

Size the shelf, not the warehouse

The right way to size a catalogue is bottom-up from the surfaces a player actually sees. Count them honestly: your homepage rails, your category pages, the "more games" strip beside the player, your search results for the twenty queries people actually type, and whatever the app or portal shell surfaces on launch.

Multiply that out and most portals need a few hundred well-chosen games to fill every visible surface without a gap. Everything past that is inventory, not merchandising. It costs money to carry and it earns nothing unless something routes traffic to it.

Depth belongs where the traffic already is

Depth is worth buying in the two or three categories that actually pull for your audience — and those categories are not the same for a European web portal, a MENA carrier portal, and a kids' site. If puzzle and .io games are 60% of your plays, a hundred more dress-up titles will not move anything. Twenty more good puzzle games might.

This is also the argument for buying selectively from a large source pool rather than buying the pool. You want the ability to pick 300 titles that fit your traffic, then pick 300 more next year when the traffic has changed shape.

🔁 Games Added Per Week Beats Games Owned

Poki's 227 releases in 2025 work out to about four a week, onto a library of roughly 1,500. That is a refresh rate near 15% a year on a catalogue that is deliberately capped.

Worth flagging a discrepancy: some profiles describe Poki as publishing "one game a day", which does not reconcile with 227 releases in its own annual report. I would trust the company's own year-in-review number over a secondary description. Either way, the cadence is regular and visible, and that is the mechanism.

A returning player does not experience your catalogue. They experience the delta since their last visit. A portal that adds four decent games a week to 1,500 feels alive. A portal sitting on 8,000 games with nothing new since launch feels abandoned, and no amount of back-catalogue depth fixes that impression — the player never sees the depth, only the front page they have already exhausted.

So when you negotiate, negotiate the refresh. Ask what lands per month, whether new titles are included in the licence fee or priced separately, and whether you get them at the same time as every other licensee or on a delay. That clause is worth more than a thousand extra titles at signature.

📉 The Number a Bigger Catalogue Cannot Move

Portals buy games when what they actually have is a retention problem, and games do not fix retention on their own.

The published benchmarks disagree enough that you should treat any single figure with suspicion. Vendor benchmark sets circulating in 2025-2026 put hybrid-casual around 15-20% day-7 retention, hyper-casual nearer 33% day-1 falling to about 12% by day 7 and 4% by day 30, while broader market medians reported elsewhere for 2026 are considerably worse — roughly 22% day-1 and single-digit day-7. The methodologies are not comparable and none of these are portal numbers. What survives the disagreement is the shape: the drop between first session and second is where the business is won or lost, and it happens before a player could plausibly have exhausted even a 200-game catalogue.

If your day-1 is bad, the reason is load time, a confusing first screen, an ad before the first play, or a game that does not fit the audience you bought traffic from. None of those are inventory problems. We have written separately about sizing HTML5 games for low-end Android devices, which is usually a bigger lever than anything on the catalogue side.

💸 What Catalogue Depth Actually Costs to Carry

Extra titles are not free just because the licence was a flat fee.

• Thin pages at scale. Eight thousand game pages with auto-generated 40-word descriptions is a large surface of near-duplicate content competing with itself in search.
• QA and rot. Older titles break — ad SDKs deprecate, third-party endpoints disappear, portrait handling fails on new devices. Somebody has to test them, or players find them broken for you.
• Localisation. Every extra title multiplies metadata translation cost across every market you launch in.
• Licence administration. More titles means more expiry dates, more territory restrictions and more platform scope to track. Every one of them is a compliance exposure if you lose track of what you are entitled to run where.
• Storage and CDN. Marginal per title, real at volume, and paid monthly whether the game gets played or not.

None of this is catastrophic. It is just cost with no matching revenue line, sitting on titles that were bought to make a number look bigger.

🚫 Five Ways Operators Get Catalogue Sizing Wrong

1. Buying breadth to substitute for a traffic plan. A catalogue is inventory, not demand. If you do not know where the first 10,000 sessions come from, no catalogue size fixes it. This is the single most common failure in white-label portal launches.
2. Treating the licence-included count as the launch count. Getting 2,000 titles does not mean shipping 2,000. Ship the 300 that fit, hold the rest, and add on a schedule.
3. Never removing anything. Portals add and never cull. Games with no plays in six months are not a long tail, they are dead weight diluting every recommendation surface you run.
4. Buying depth in categories the audience does not play. Look at your own play distribution before the next purchase, not the licensor's category list.
5. Confusing catalogue size with catalogue fit. A carrier portal in a market with 4G-era devices and a kids' site in Western Europe need different games, not more games. Fit is a selection problem and selection is work.

🎮 Where a Licensed Catalogue Fits

There is an obvious tension in a licensor arguing that catalogue size is overrated, so let me be direct about it. Forestry Games licenses 1,049 HTML5 and Android titles, and the useful way to read that number is as a source pool, not a shipping manifest. The value of a catalogue that size is that you can select a few hundred titles that match your traffic, your devices and your market — and select differently next year.

That is the same logic behind our white-label portal work and the catalogue itself: pick against your audience, not against a total. If you want to see what concentration looks like in practice, the most-played titles are a reasonable proxy for where play actually lands.

🧭 Before Your Next Catalogue Purchase

Do three things before you sign anything.

Pull your own concentration curve. Rank every title you already run by plays over the last 90 days and find how many account for 80%. If that number is small — and it will be — you have just proved the case on your own data, and you know exactly which categories deserve depth.

Count your visible slots. Add up every game position a player can actually reach in three clicks. That number, plus a working reserve, is your real catalogue requirement. Compare it to what you are about to buy.

Negotiate the cadence, not the total. Get new-title access, timing and pricing written into the agreement explicitly. A portal that adds four good games a week will beat one that bought ten thousand and stopped, every time.

The badge on your homepage saying how many games you have is read by nobody except your competitors. The player only ever wants the next good one.