steps
Every Learn guide is grounded in peer-reviewed research, drafted with AI assistance, and reviewed by a human before it publishes — and this page explains exactly what that means, because "research-backed" is the most abused phrase in wellness writing.
Every research claim in a Learn guide traces to a peer-reviewed study, named in the text and listed in the guide's "research and resources" section with a link to the primary source. Guides may also cite books, videos, or podcasts — always labeled for what they are, and never carrying evidence weight: a podcast can be worth your time without proving anything. Each cited source gets its own page here, summarized in plain language — including a section on what it does not show, because the gap between what a source says and what a product would like it to have said is where most wellness writing goes wrong. Numbers we can't verify are omitted, not estimated. We link primary sources with referral parameters so publishers can see the traffic came from us.
Drafted with AI assistance, reviewed by a human — specifically, drafted by Claude against the study summaries and this policy, then read, edited, and approved by Tiago (the founder, whose own journaling practice is on the about page) before anything is published. Nothing goes live unreviewed. We publish slowly and deliberately — a genuinely useful page or none.
No invented statistics, testimonials, or reviews — the same rule as the rest of steps.org, where we'd rather show an honest empty state than fabricated praise. No medical claims: gratitude journaling is a well-being practice with a real but small measured effect, not a treatment for anything. No hiding the mixed findings — where the evidence is thin (physical health) or inconvenient (gratitude fatigue), the guides say so, with the citation.
Steps is the product these pages support, and we say so rather than pretending otherwise — comparison content discloses the conflict of interest at the top, competitors are described fairly at their published prices, and the research pages make the same modest claim to you that they make for us: the practice helps, it doesn't transform.
If a study is retracted, a number is wrong, or a characterization is unfair, we fix the page and update its "last updated" date. Spotted something? Write to [email protected] — corrections get read first.