What 567 Clicks Tell Us About the AI Companion Market
Our first readout of real first-party click data: 567 outbound clicks over 30 days, one site, English only. Directional and modest, and honest about what a small sample can and cannot say about how people shop for AI companions.
By Ash Kepler · Jul 21, 2026 · 6 min read
For a year we have written about the AI companion market using other people's numbers. This is the first time we are showing you ours. Over 30 days, people clicked an outbound link on Pocket Animus 567 times, and because those clicks pass through our own attribution layer, we can say a little about where they went and what they were reading first.
Cite this readout: Pocket Animus, "AI Companion Market Readout, July 2026," https://pocketanimus.com/insights/ai-companion-market-readout-july-2026/
Last updated: July 21, 2026.
A word on what this is before any numbers. This is a first readout, not a market report. The dataset is 30 days of outbound affiliate clicks (June 20 to July 19, 2026), measured through our own redirect layer: 567 clicks, across 71 referring pages, going to 26 destinations. That is enough to see direction and nothing more. It cannot tell you what the market does, only what one site's readers did in one month. The audience is self-selected, the sample is small, and it is English only so far. Every figure below should be read as a lean, not a law. The modesty is the point, because a number you can trust a little is worth more than a number you cannot trust at all.
Free is pulling the heaviest shift
In a 30-day sample of 567 outbound affiliate clicks tracked by Pocket Animus between June 20 and July 19, 2026, pages about free or unlimited access drove 201 clicks, which is 54.6% of the 368 clicks that came from a known page on the site.
The pattern is hard to miss. Once you set aside the clicks that arrived with no referring page, the majority of the rest came from readers who were on a page about getting something without paying for it. Counted against all 567 clicks the share is 35.4%, still the single largest content theme, and neck and neck with no-referrer traffic. Either way you slice it, free is the word doing the work.
Read it as intent, not as a verdict on the apps. People who land on a free guide are shopping with their wallet closed, and they click through to try the thing that costs nothing first. It is worth saying plainly that this site publishes a lot of free and uncensored access content, so some of this is us finding our own audience in a mirror. That is exactly the kind of thing a first readout is for: noticing the shape, then watching whether it holds.
The crowd has its favorites
In that same 30-day sample of 567 outbound clicks tracked by Pocket Animus in July 2026, the three most-clicked destinations were SpicyChat with 171 clicks (30.2%), CrushOn with 86 (15.2%), and Candy with 65 (11.5%), together 56.8% of every outbound click.
The traffic is concentrated. The top five destinations took 70.4% of all clicks, and everything from the ninth destination down is a long tail of ones and twos. No quality judgment lives in these bars. A destination gets clicked because of which pages point at it and how much traffic those pages get, not because it is better or worse than the next one. What the chart shows is distribution, and the distribution is lopsided.
A third of you never say where you came from
In the 30-day sample of 567 outbound clicks tracked by Pocket Animus between June 20 and July 19, 2026, 199 clicks (35.1%) arrived with no referring page recorded, meaning they came directly or with the referrer stripped by the browser.
A third of the clicks show up without a coat check. Some of that is people who bookmarked a link or came back to a page they trusted. Some of it is browsers and privacy tools that strip the referrer on purpose, which is more common in exactly the kind of privacy-minded audience this category attracts. We do not get to know which, and we will not pretend to. It is a healthy reminder that a big slice of real behavior is invisible to any attribution layer, ours included.
The other two languages are still finding their seats
Pocket Animus publishes in English, Traditional Chinese, and Brazilian Portuguese, but only one of them shows up in this readout. During the same 30-day window, the Chinese-language section recorded 0 outbound clicks, which is a clean starting baseline rather than a problem, and the Portuguese section had launched too recently to register a single click in this sample. So the multilingual data starts now, at zero, which is the honest place for it to start. The next readout is where those lanes get their first real line on the chart.
How we run the tab (no cameras, no names)
Here is the whole mechanism, because a number is only as trustworthy as the way it was counted. When you click an outbound link on this site, it does not go straight to the destination. It passes through a small redirect running on our own server. That redirect does two things and then gets out of the way: it adds one to an aggregate counter, and it forwards you to where you were going.
The counter records three things only, and all three are about the click, not about you: which page on our site sent it, which destination it pointed at, and which day it happened. There are no cookies set, no device fingerprints taken, and no personal data stored anywhere in the process. What we end up with is a pile of tallies, the digital equivalent of a bartender marking a chalkboard, that we add up once a month.
That design is also why this readout is as modest as it is. Aggregate counts cannot tell us who clicked, whether they signed up, or whether they spent a cent. They can only tell us how interest was distributed across pages and destinations in a window of time. That is a narrow instrument, and we would rather show you a narrow instrument read honestly than a wide one read with a thumb on the scale. Next month we do it again, and for the first time we get to measure something better than a snapshot: change.
questions