Affiliate marketing is the business model of almost every site in this category - the economics are described in how AI companion apps make money, where affiliate commissions are often one of an app's largest costs. Knowing how it works lets you use reviews for what they are good for, and discount what they are not.
How the money works
The incentive is real. Disclosure and method are what keep it in check.
- A reader clicks a link on a review site and subscribes.
- The app pays the site a fixed fee or a share of the subscription, sometimes for as long as the reader stays.
- Apps that pay more, or convert better, earn more for the site.
The incentive is obvious: rank the best-paying app first. The protections against it are disclosure, a published testing method, and a willingness to rank paying apps low when they test badly.
What the rules say
In the US, the FTC's Endorsement Guides require clear disclosure of material connections, such as affiliate commissions, close to the recommendation - not buried in a footer. Since October 2024 the FTC's rule on fake reviews also bans buying or selling fake reviews and undisclosed insider reviews, with civil penalties for violations. The UK and EU have similar consumer protection rules.
A site that hides its commissions is not only being unhelpful; it may be breaking the law.
Signs of a review based on real testing
| Sign of real testing | Sign of a rewritten press release |
|---|---|
| Specific moments: "on day nine it recalled..." | Generic claims: "amazing memory" |
| Prices checked on a stated date | Prices copied from the app's homepage, often out of date |
| Screenshots of paid features in use | Only marketing images from the app |
| Criticism of the recommended app | Every app is "great for most users" |
| A method page explaining how scores are given | Scores with no explanation |
| Updated after the app changes | Unchanged for years |
| Mentions privacy incidents involving the app | Silence about anything negative |
Signs worth being wary of
- Every list has the same top three, in the same order, across many sites. That can mean they are genuinely best - or that they pay best.
- "Best" lists covering apps the site never reviewed. A ranking of twenty apps with full reviews of none.
- No negatives at all. No app in this category is without trade-offs.
- Countdowns and "exclusive" discounts that never expire.
- Reviews that sound identical across different sites, a sign of shared copy.
Reading our own reviews critically
The same questions apply to us. Our disclosure is on every page. Our method - paid accounts, two-week tests, five scores, hidden costs and privacy checks - is published in how we test. We rank only apps we have tested, and some apps we link to sit near the bottom of the ranking. When something important happens after a test, such as a privacy incident, the right thing is to say so; see AI companion data breaches.
You should still check. The best defence against any review, including ours, is a week of your own testing - the plan is in how to test an AI girlfriend app yourself.
A quick checklist
- Is the affiliate relationship disclosed near the recommendation?
- Is there a method page, and does the review follow it?
- Are there specific, dated details that only use could produce?
- Does it criticise the app it recommends?
- Does it mention privacy, not just features?
A review that passes all five is worth reading closely, whoever wrote it. One that passes none is advertising, however it is labelled.