Nine gold mines of app research

Your competitors have already done your user research. They spent years and millions finding out what people want, and then they left the notes lying around in public. Read them before you write a line of code: a day of digging changes the feature list, the pitch and the growth plan.

The nine places below run from what users say, to what the market offers, to the people around you, to the science. Each comes with one thing we found there while researching RepGarden, the app for personal trainers and their clients that I am building in public.

1. Review sites

The App Store, Google Play, G2, Capterra and Trustpilot. Every app you compete with has hundreds of public reviews. The one-star reviews tell you the promise that got broken, in the user’s own words, and what they would switch for. The five-star reviews tell you what people actually pay for.

The counts tell you something too: how big a platform really is. One trainer platform claims more than 380,000 coaches on its homepage. Its client app has about 2,700 App Store ratings, which only happen when a real client uses it. ^[1]

2. Forums and comment sections

Reddit, Facebook groups, YouTube comments and the niche forums for your users’ trade. Here people talk about the problem when nobody is selling to them, and you hear the workarounds they have built, the stage they are really at and the exact words they use for it.

We read and coded more than 500 complaints and pieces of praise about app and website developers, from Reddit, review sites, forums and video comments. Most of the complaints came after a build had gone wrong, not before, which changed who we talk to and when. ^[2]

3. Search data

Ahrefs, Semrush, Google’s own autocomplete and AlsoAsked. Keyword tools show you what people type into Google before they buy anything. Searches like “X vs Y” and “X alternative” are people about to switch. “How to cancel X” is churn. The search volume tells you how big the demand is, and the wording tells you how people describe the problem when nobody is listening.

We learned how people search before hiring a developer, and wrote the pages they land on around those words.

4. Feature-request boards

Canny, UserVoice, the “ideas” portals and GitHub issues. Most platforms run a public board where users post the features they want and vote on them. Read it and you get a ranked list of what users are asking for, and a record of what the vendor has refused to build.

The market leader in trainer apps closed the request “Allow clients to upload video of their form” in June 2020, with the reply that clients could send a video message instead. Six years on, the platform still has no client video upload. That gap is one of the first things RepGarden does. ^[1]

5. Help centres and changelogs

Their help centre, their pricing page and their release notes. A competitor’s homepage is marketing. Its help centre is the honest feature list, because support articles do not exaggerate. Add the pricing page and the release notes and you know what is an add-on, how long the trial is, and what shipped last month.

The homepages did not say which trainer apps let a client upload a video of a set. The help centres and feature pages did: five of the fourteen. ^[1]

6. Ad libraries

Meta’s Ad Library, Google’s Ads Transparency Center and TikTok’s Creative Center. Meta, Google and TikTok publish every ad a company is running. You can see the promise they think sells, who they are targeting, and which ad has run the longest, which is the one that works. What none of them say is your opening.

Eight trainer platforms were running 318 Meta ads between them on the day we looked. Every one sold a free trial and less admin. Not one led with reviewing a client’s video. ^[1]

7. Your first user’s current tool

Sit next to your first user while they use whatever they use today. You will see the clunky parts they have stopped noticing, and the features that exist but nobody touches. Neither shows up in an interview.

The app our first trainer pays for can already take a video of a set. He never uses it. So the upload is not the product; the review is: how fast it is to film a set and get a reply, and whether the reply lands on the moment in the clip. ^[1]

8. The people who hear customers all day

Shop staff, receptionists, support teams and the tutors who train your users each talk to a hundred customers a week. They know the questions asked every day, the objections, and what gets returned. One coffee with them is worth a month of surveys.

This is the one we have not dug yet. Next on the list is the reception desk at a gym.

9. Papers and engineering blogs

Someone has usually tested the thing you are about to guess at. Google Scholar and the engineering blogs of the big apps hold the results, with numbers, for free.

Two finds shaped RepGarden’s reminders. In a randomised trial with 17,500 app users, each extra push notification a week raised the uninstall rate by 2.5 percentage points. ^[3] And in a study of 61,293 gym members that tested 53 different nudges, the one that worked best was a small reward for coming back after a missed workout. ^[4]

Dig before you build

None of this takes money, and most of it takes a day. Read the reviews and the forums first, because that is where users speak for themselves. Then read what the market offers, in its help centres and its ads, to find what everyone says and what nobody says. Then sit with one real user, talk to the people on the front line, and check the science for the part you were about to guess. By the end you will know which feature to lead with, what to promise, and where to find the first people to tell.

Sources

1. RepGarden market research — fourteen trainer-client platforms, 26 October 2026. Our own research for the RepGarden build, read from each platform’s website, pricing page, help centre, store listings and public ad library on one day. The platforms are named in the report and left unnamed here. The report in the RepGarden repository.

2. GadirLabs — what buyers complain about, October 2026. 521 complaints and pieces of praise about app and website developers, collected from Reddit, Trustpilot, Clutch, forums and video comments posted between January 2024 and October 2026, and coded by theme. An internal study; the headline finding is that most complaints are made after a build, about the build.

3. Wohllebe, A., Hübner, D.-S., Radtke, U. and Podruzsik, S. — “Mobile apps in retail: Effect of push notification frequency on app user behavior”. Innovative Marketing 17(2), 2021. A randomised field experiment with 17,500 users of a retail app, split into five groups and sent pushes at different frequencies for seven weeks. The paper (PDF).

4. Milkman, K. L., Gromet, D., Ho, H. and colleagues — “Megastudies improve the impact of applied behavioural science”. Nature 600, 2021. Fifty-three interventions tested on 61,293 members of a US gym chain, measured on check-ins. An author copy of the paper (PDF).