People make real decisions based on our reviews: which app to trust with their mental health, which platform to try therapy through, which software to run a practice on. So our reviews follow a written method, the same one every time, and it is published here in full. Our editorial policy governs everything on this site; this page covers the extra rules that apply when we evaluate a product, including reviews of AI mental health apps, guides to therapy software and coaching software, and comparisons across our AI therapy section.
What we review, and what we refuse to review
We cover products people are actually searching for and asking about: consumer mental health apps, AI chatbots and companion tools, online therapy platforms, and software for therapists and coaches. Coverage is an editorial decision. A company cannot pay to be reviewed, and it cannot pay to not be reviewed.
Before we write a word, we confirm the product actually exists and is what it appears to be. That sounds obvious, but in the AI mental health space it is not: many searched-for "apps" turn out to be defunct products, rebrands, or something else entirely wearing a similar name. When a product fails that check, we either say so plainly on the page or we do not publish a review at all. We do not review products that cannot be verified, and we do not invent impressions of tools we could not access.
The steps every review goes through
-
Verify the product and the company
We confirm the product is live, who operates it, and that the claims on its marketing pages match what the product actually is. Defunct or misrepresented products are flagged as such.
-
Use it hands-on where possible
Where a product can be signed up for and used, we use it: real accounts, real tasks, real conversations. Where hands-on access is not possible, the review says what our assessment is based on instead, such as published research, documentation, and verified user reports. We do not imply testing that did not happen.
-
Verify pricing and terms at the source
Pricing, free tiers, and trial terms are checked against the vendor's own site, not secondhand summaries, and reviews state when key facts were verified. If pricing changes after publication, we update it when the page is next revised.
-
Check claims against the evidence
Where a product claims research support, we look at the research itself: what was actually studied, on whom, funded by whom, and what it did and did not show. Vendor-funded studies are identified as such. Evidence standards follow our editorial policy.
-
Assess safety and privacy
For consumer mental health tools we look at how a product handles someone in crisis, what its privacy policy lets it do with sensitive data, and whether it is honest about what it is. An AI tool is not a therapist and not a crisis service, and we hold products to how clearly they say so themselves.
-
Publish under named oversight
Reviews are published under the oversight of our Editorial Board, following the same review and labeling rules as the rest of the site.
What decides the verdict
Different categories weigh differently, but the questions behind every verdict are the same:
- Does it work? Evidence of benefit, quality of the underlying approach, and whether the product delivers what it promises.
- Is it safe and honest? Crisis handling, appropriate boundaries, and truthful marketing, especially for AI tools used by vulnerable people.
- What happens to your data? What is collected, who it is shared with, and whether sensitive disclosures are protected.
- Is it worth the price? Cost against alternatives, the honesty of the free tier, and how easy it is to cancel.
- Can you live with it day to day? Usability, reliability, and support, based on our own use wherever possible.
Independence rules
- No pay for placement, ever. Rankings and scores cannot be bought, and inclusion in a "best of" list is never sold.
- No pre-publication approval. Companies do not see reviews before they publish and cannot veto criticism.
- Affiliate links never touch the verdict. Some product links earn us a commission, disclosed on the page and explained in how we make money. Products we have no commercial relationship with regularly outrank products we do, because the ranking is the editorial judgment, not the business model.
- Ownership is disclosed, prominently. Our founder co-founded Quenza, a coaching software product we cover. Every guide that includes it says so directly, and we would rather over-disclose that connection than have you wonder.
- Reviews get maintained. When a product shuts down, changes pricing, or materially changes what it is, we update the review and date the change rather than leaving a stale verdict standing.
How we vet therapists and make matches
Products are one half of what we evaluate. The other is the professionals in our therapist directory and matching service, where the method matters even more because the outcome is a human relationship.
- License verification is the backbone. Therapists can verify their professional license with us, and the Verified badge on a profile reflects that verification. It cannot be bought.
- Matches are ordered by fit, never by payment. When you tell us what you are going through and where you are, matching weighs relevance to your situation: specialty, location, and what the therapist actually offers. No therapist can pay to appear in your results or rank higher in them.
- The matching assistant is a matchmaker, not a therapist. Its job is to understand what you need and connect you with a licensed human professional. It does not provide therapy, counseling, diagnosis, or treatment.
- Listings are not endorsements. A profile in our directory means the therapist meets our listing requirements, not that we are recommending them for your specific situation. The fit decision is always yours.
Tell us when we miss
If a review is out of date, or your experience with a product sharply contradicts ours, email hello@psychology.com. Corrections follow the process in our editorial policy.