From Beta to Launch: The 47 Users Who Carried Us

There is a habit in software where companies write blog posts thanking their “community” without naming anyone, citing anything, or admitting that the community was, in fact, forty-seven people. We want to do the opposite. The early life of Baseline Maps was not a movement. It was a survey, a hundred installs, and forty-seven replies that we read line by line on the couch, on phones, on a laptop balanced on a kitchen table while coffee got cold.

This is the gratitude post we owe those forty-seven. Not abstract thanks for “supporting us on the journey.” Specific thanks, for specific feedback, that shaped specific screens. If you have ever wondered whether one well-written email actually moves the needle on a piece of software, the rest of this post is for you. The short version: it does, and the long version is everything below.

The survey we almost didn’t send

We almost didn’t send the beta survey. The app was rough, the install count was small, and the working theory was that we’d wait until things felt “more ready.” Then a friend pointed out that “more ready” is what you tell yourself when you’re afraid of the answer. We sent it on a Tuesday morning to all one hundred installs, attached to a plain email with no header image and no marketing copy.

Replies started arriving within the hour. By the end of the first day there were nineteen. By the end of the week, forty-seven. The remaining fifty-three never replied, and a quick check showed that most of them had also stopped opening the app. The survey was, accidentally, a perfect retention probe. The people who cared enough to write were the people who cared enough to stay.

Three questions, no Likert scales

The survey was three questions. What’s broken? What’s missing? Would you tell a friend about this app, and if so, what would you say? We deliberately skipped Likert scales, NPS, star ratings, and every other instrument that lets a respondent answer without thinking. The cost was that we couldn’t graph anything in a slide.

The benefit was that every reply was a paragraph or a voice memo, and every paragraph contained an actual sentence about an actual moment of using the app. We learned about a missing label on a screen because someone described squinting at it at dawn. We learned a download had failed silently because someone walked us through the seventeen seconds before they gave up. None of that survives a five-star rating.

What “what would you tell a friend?” actually surfaced

The third question did something the first two couldn’t. It made people describe the product instead of evaluate it. “I’d tell them it’s the only map that shows me where the boundary actually is.” “I’d tell them it loads when the cell tower doesn’t.” “I’d tell them it doesn’t try to be a social network about my fish.” Reading forty-seven of those in a row is how we learned what we had built.

Which turned out to be slightly different from what we thought we were building. We thought the product was an outdoor app. Users told us the product was an honest map. We thought the brand was the differentiator. Users told us the data was. We rewrote the home screen, the onboarding, and the App Store description in the week after we finished reading. None of those changes would have come from a satisfaction score.

Patterns in 47 voice memos

Three patterns showed up. First, the data was the product — accurate boundaries, current regulations, real elevation — not the interface or the brand or anything we could have A/B tested. Second, users kept asking for features we hadn’t considered, several of which were obvious in retrospect and embarrassing in foresight. Third, the people who answered carefully were the same people who stayed for months.

The third pattern was the one that quietly reorganized how we think about the whole business. The willingness to write a paragraph turned out to be the single strongest retention signal in the dataset — stronger than session count, stronger than install date, stronger than any segmentation we’d been told to track. The lesson generalized: the user who is talking to you is also the user who is staying.

The first design backlog

We printed the responses, cut them into strips, and sorted them on a kitchen table. Bugs went in one stack. Missing features went in another. Misunderstandings — places where the app worked but didn’t communicate that it worked — went in a third stack that turned out to be the largest. By the end of the weekend we had a design backlog with forty-one entries, each tagged with the name of the person who’d surfaced it.

Of those forty-one, thirty-four shipped before public launch. Four shipped in the first three months after launch. Two are still in flight a year later, visible in the in-app Development Queue with the original requester’s first name attached. One was retired because the underlying need turned out to be solved by a different feature entirely, and we wrote the requester a long email explaining why.

Who’s still here today

Twenty-three of the original forty-seven are still active Founders. Several of them are now the most prolific feature requesters in the entire app — we recognize their names instantly when a new request lands, and we know roughly what kind of feedback to expect before we open the email. A handful have become informal beta testers for risky changes, the kind of work where you need someone who will actually open the build and find the broken thing.

The twenty-three are not a demographic. They are a mountain biker in Montana, a steelheader in Oregon, a retired surveyor in Idaho, a father-and-son elk hunting pair in Wyoming, a guide in northern California, and eighteen others whose stories we know in fragments. We are not going to name them publicly. They know who they are, and so do we, and that mutual recognition is the whole point.

The signal nobody talks about

There is a signal most product teams ignore because it doesn’t fit in a dashboard. It’s the difference between a user who taps “Submit” on a feedback form and a user who records a forty-five-second voice memo from a riverbank. Both technically count as “engagement.” Only one of them is telling you anything.

The forty-seven were almost entirely the second kind. Looking back, the survey was less a research instrument than a self-selection filter for the kind of person we wanted to build alongside. The format mattered too — long-form, open-ended, plain text — because it filtered out anyone who only had the patience to click a star. If you only ever ask for clicks, the people who would have written you a paragraph never get the chance.

How we run beta now

We’ve kept the survey almost unchanged. Three questions, no scales, no required fields, sent by plain email. The install count has grown, the response rate has dropped slightly, and the percentage of respondents who become long-term users has held remarkably steady. The shape of the work has changed — there is a real backlog now, and the in-app Development Queue handles most of what used to live on a kitchen table.

But the spine is the same. Ask three questions. Read every answer. Build for the people who wrote back. We will keep doing this for as long as the app exists, because the alternative — guessing what users want from aggregate metrics — is how most outdoor apps end up feeling like they were designed by people who have never been outside. We would rather know forty-seven people well than ten thousand barely at all. That is not a humility flex. It is an operating decision, and it is the one decision from the beta era that has cost us nothing and given us everything.

---

If you are one of the forty-seven: thank you. If you are not, but you are reading this on a phone with Baseline Maps installed, the door is still open. Send a voice memo. Send a screenshot. Send a two-paragraph email about the moment the app failed you. We read everything, we triage everything, and the things we ship next are visible to every user in the in-app Development Queue. The product is built out in the open, by a small group of people who care, and that has not changed since the first Tuesday morning.