A score is only worth as much as the test behind it. This page describes, in full, how OurDream AI puts an AI companion platform through its paces — so you can weigh the method, not just the number.

Every platform goes through the same routine, on a fresh paid-tier account where possible, so nothing rides on a demo that has been polished for reviewers.

The four things we measure

  • Conversational quality. Does it hold a thread, stay in character, and respond like something worth talking to — or does it loop, contradict itself and forget who it is?
  • Memory and continuity. We return across days and weeks to see whether shared details survive. A companion that forgets last week is the single most common failure in this category.
  • Billing honesty. How clear is the platform about what is free, what is paid, and how to cancel? Dark patterns at the checkout cost real marks.
  • Privacy and data control. What is collected, and can you actually get it back or delete it when you ask? This ties directly to the standards in our own Privacy Policy.

How testing runs

We build a companion from scratch and live with it. Conversations happen at different times of day, across several weeks, and include deliberate edge cases — dull stretches, low moods, and pointed attempts to trip up the memory system. We log behaviour as we go rather than relying on a single first impression, because first impressions in this category are often the most flattering and the least honest.

How scores are reached

Each of the four areas is rated on its own, then weighted into an overall editor's score. The weighting reflects what actually matters to a daily user, with conversational quality and memory carrying the most. Scores are revisited when a platform ships a major update, and the date on a review tells you how current it is.

What never touches a score

Money. Whether a platform pays us a commission has no bearing on where it lands — the firewall is described in the Editorial Policy and the commercial side in the Affiliate Disclosure. The testing is run by a named editor, Maya Brennan, whose background you can check, and the wider operation is explained in About Us.

If a result ever looks off to you, challenge it — Contact is open. You can also review our Cookie Policy and Terms of Service, or go straight to the reviews this method produced on OurDream AI.