Evaluation Checklist

How to evaluate an AI mental health app

Most people should not start with “is this app good?” They should start with “what claims does it make, what happens to my data, and what support exists when things go wrong?”

01

Check the product claims first

Does the app describe itself as wellness support, coaching, journaling, or therapy? A trustworthy product is specific about what it is and what it is not.

02

Read the privacy policy like it matters

Look for clear language on what is collected, how it is stored, whether data is used for advertising, and how deletion works.

03

Look for a real crisis boundary

If a product is not emergency care, it should say so plainly and route people toward immediate support when needed.

04

Ask what evidence supports the habit

The app should explain why its format works: prompts, reminders, scheduling, or journaling design, not just "AI personalization."

05

Judge the retention design

The best apps reduce friction. They make it easy to start, return, and understand progress without depending on perfect motivation.

1. Separate wellness language from medical claims

Good products define their scope. They tell you whether they help with reflection, coaching, journaling, habit support, or clinical treatment. If that boundary stays fuzzy, you are already dealing with a trust problem.

2. Privacy is not a footnote

Mental wellness products collect deeply personal signals. The FTC's BetterHelp action is a reminder that “we care about privacy” is not enough. Look for specific language around ad targeting, deletion, retention, and third-party sharing.

3. Crisis handling should be explicit

If the product is not emergency support, it should say so clearly and route people toward licensed care or crisis resources. Silence on this point is not sophistication. It is a missing safety boundary.

4. Habit design matters more than clever AI language

Ask what the app does to reduce friction. Does it provide a schedule, a prompt structure, a review loop, or a clear next action? Consistency usually comes from good product design, not from branding something as intelligent.