AI Companion App Downloads and Revenue: What We Can Actually Verify

Every so often someone asks me for download numbers or revenue figures for AI companion apps, expecting a clean, citable statistic. I don't have one I'd stand behind, and I want to explain exactly why, rather than just making one up or repeating a number I found somewhere without checking it. Then I want to show you what we can actually verify, because there's real, useful data here, it's just a different kind of data than "downloads" or "revenue."

What we genuinely can't verify, and why

Download counts and revenue figures for specific apps generally come from either the company itself, which has every incentive to round favorably or not disclose at all, or from third-party estimation methods that infer numbers from indirect signals rather than measuring them directly. Neither of those is something we can independently confirm for a private company that doesn't publish its own numbers. Most AI girlfriend platforms fall into exactly that category: privately held, no public financial disclosures, no confirmed download counts we could check against a primary source.

Rather than pass along a number from somewhere else and hope it's accurate, we've made a deliberate choice: we only publish numbers we've computed ourselves from data we can verify directly. That's a stricter bar than a lot of coverage of this space applies, but it's the only way we can stand fully behind everything in this article.

How app category rankings generally work, in general terms

It's worth understanding, at a general level, how app stores and app category rankings typically function, since that context explains why specific numbers are so hard to pin down for a category like this one. App stores rank apps within categories based on a mix of factors including recent download velocity, ratings, and engagement signals, and that ranking position is usually visible even when exact download counts aren't. A high category ranking tells you an app is performing relatively well against other apps in the same category at that moment, but it doesn't hand you an absolute number, and rankings shift constantly based on factors outside any single company's control.

For AI companion and AI girlfriend apps specifically, there's an additional wrinkle: a meaningful share of this category operates through web apps and browser-based platforms rather than traditional app store listings, which means even the general app-ranking signals that exist for app-store-native products don't apply evenly across the whole category. That's a big part of why we don't think a single download or ranking figure could honestly represent "the AI companion industry" as a whole.

What we can verify instead: our own testing data

Here's what we actually have confidence in, because we generated it ourselves through direct, hands-on testing rather than inferring it from indirect signals: pricing, feature availability, and quality scores across 129 platforms, verified by using each product directly rather than reading its marketing copy.

Pricing as the real signal for this category

Pricing structure is a genuinely useful proxy for scale and demand, and it's one we can back with actual verified data. Across the 85 platforms in our database with a clearly parseable price, the average starting price is $11.85 a month. Out of all 129 platforms, 16 are genuinely free, 56 sit in the budget tier, 55 sit in mid-range, and only 2 are priced as premium. That distribution tells a real story: this category has converged on a specific, sustainable price band through actual market pressure across well over a hundred competing products, and that convergence is itself a form of demand signal, just not one denominated in a specific dollar-revenue figure.

Platform count as a signal of category scale

The sheer number of platforms we track, 129 and almost certainly still a floor on the category's real size, is itself a form of scale evidence. A market can't sustain well over a hundred competing products, active pricing experimentation across dozens of them in a single year, and continuous new entrants replacing the roughly 18% that fail or exit annually, without a genuine, sizable base of actual usage behind it. It's an indirect argument, but it's one built entirely on things we've confirmed ourselves rather than borrowed from an unverifiable source.

Feature investment tells a similar story from a different angle. 22% of platforms now offer AI video generation, a feature category that requires meaningful compute cost to run at scale. Companies generally don't build and maintain expensive infrastructure for a feature nobody uses, so the fact that over a fifth of the industry has made that investment is itself indirect evidence of real, sustained demand, even without a download number attached to it.

How to evaluate download or revenue claims you see elsewhere

If you come across a specific download or revenue figure for an AI companion app or the category as a whole, a few useful questions to ask: does the source show its actual methodology, or just state a number? Is the figure attributed to a named, checkable source, or presented as a general fact? And does the number get repeated identically across multiple sites without any of them showing their work? A specific-sounding number isn't the same thing as a verified one, and this is a category where that distinction matters more than most, given how little of it operates through channels with transparent public reporting.

A related pattern worth watching for: a number that appears on many different sites in identical phrasing usually traces back to a single original, often unverified, source that everyone else copied rather than independently confirmed. That's a common way weak statistics spread and calcify into apparent consensus, in this category and plenty of others, and it's exactly why we've limited ourselves in this article to numbers we generated ourselves through direct testing rather than joining that chain of unverified repetition.

Why so many platforms in this category skip app stores entirely

It's worth explaining why so much of this category operates as web apps rather than app-store listings, since it's directly relevant to why download data is so hard to pin down. Adult and NSFW-adjacent content faces significant restrictions in major app store policies, which pushes a meaningful share of AI companion platforms toward browser-based products instead. That's a reasonable business decision for platforms serving the 104 of 129 platforms in our database that allow NSFW content, but it also means an entire, sizable segment of this category generates no app-store download signal at all, verified or otherwise, which is a big part of why we'd treat any download-based estimate for "the AI companion industry" with real skepticism.

Our standard for the numbers we publish

Every figure in this article, and across our entire data pillar, comes from our own testing of 129 platforms rather than from third-party estimates we can't verify. We maintain a paid subscription on each platform, test features directly, and record pricing and quality data ourselves. You can read our full testing methodology, and see the complete set of verified numbers we track on our data hub.

If pricing and verified feature data are the numbers you actually trust, as they should be given everything above, the most useful next step isn't another statistic, it's a direct, side-by-side comparison. Our best AI girlfriend rankings apply this same verified pricing and feature data to every platform in our 129-platform database at once.

Further reading

• AI Companion Market Size and Growth: What Our Own Data Suggests
• The State of the AI Girlfriend Industry in Numbers
• The best AI girlfriend apps, ranked