The question that comes after the survey
You sent the survey, isolated the users who are very disappointed at the idea of losing you, and now you hold a list of conflicting requests. The hard part starts here: what do you build first, and what do you refuse? Our guide on segmenting customer feedback covers the earlier step of sorting respondents. This article covers the next one, the roadmap decision.
Example: you build invoicing software for agencies. Thirty respondents used the product twice or more in the last two weeks. Twelve say they would be very disappointed to lose it, fifteen somewhat disappointed, three not disappointed. You have sixteen feature requests. Which one goes first?
What Superhuman did, according to its founder
The account published by First Round Review gives precise figures. The survey targets people "who recently experienced the core of our product", meaning those who used it at least twice in the last two weeks. It asks four questions: how would you feel if you could no longer use the product, what type of people would benefit most, what is the main benefit you receive, and how can it be improved for you. According to the article, results become "directionally correct" at around 40 respondents.
The benchmark is 40%: companies that struggled to grow almost always had fewer than 40% of users answer "very disappointed", while those with strong traction almost always exceeded it. At Superhuman the starting score was 22% in summer 2017. It rose to 33% after the segmentation phase, then to 58% within three quarters of product work.
Two halves of a roadmap
The principle, as First Round quotes it: spend half your time doubling down on what users already love and the other half on what holds others back. At Superhuman the first half covered speed, shortcuts, automation and design polish. The second half covered gaps: a mobile app, integrations, attachment handling, calendaring, a unified inbox, search and read receipts.
The safeguard against drift sits in the second half. The team split the "somewhat disappointed" users into two groups, depending on whether speed, the fans' main benefit, mattered to them. Those for whom speed was not the main benefit were politely disregarded. For the others, the team paid very close attention to what held them back. Users who would "not be disappointed" are not listened to: per the article, the team chose to set their feedback aside.
| Respondent group | What Superhuman says | Effect on the roadmap |
|---|---|---|
| Very disappointed | They are your fans, their main benefit defines the product | First half: strengthen what they already love |
| Somewhat disappointed, same main benefit as fans | The team paid very close attention to this group | Second half: remove the blockers they name |
| Somewhat disappointed, different main benefit | The team chose to politely disregard them | No project for them |
| Not disappointed | The team chose to set their feedback aside | No project for them |
Ranking projects: cost and impact
Once the two halves are set, Superhuman used a deliberately simple analysis. Each project gets a cost rating (low, medium or high) and an impact rating (low, medium or high). The team tackles low-cost, high-impact projects first, to ship improvements right away. The score is then tracked weekly, monthly and quarterly, and the roadmap is rebuilt every quarter with the same method.
Back to the agency example: among the sixteen requests, automatic expense import is requested by four somewhat-disappointed users who share the fans' main benefit (saving data entry time), at medium cost and high impact. It joins the second half. A payroll module requested by a single not-disappointed respondent does not become a project.
What this method does not tell you
- It describes one company and one product, a professional-grade email client. Nothing in the article proves the exact 50/50 split suits a B2B product sold to a few dozen accounts. Treat it as an order of magnitude, not a law.
- The 40% threshold is a benchmark observed by its authors, not a growth guarantee.
- A sample of a few dozen answers steers a decision, it does not prove it. Always cross-check stated opinion against actual product usage.
Keeping a record of the trade-offs
The hidden cost of this method is forgetting. Three months later, nobody remembers why the loud customer's request was refused, and it comes back. In Second Brain, you can gather survey answers, interview notes and your decisions in one folder, attach up to 20 files as sources and write instructions that apply to every conversation in that folder. You can then ask "why did we refuse this request?" with your own documents in front of you. Second Brain does not compute your fit score for you: the survey and its threshold remain yours.
For the next step, our use case on Creation for founders seeking product-market fit shows how to present these results.
Sources
FAQ
How many respondents do you need to start?
According to the First Round account, results become directionally correct around 40 respondents. Below that, read answers one by one rather than computing percentages, since a single answer moves the score by several points.
Should the roadmap really be split evenly?
Superhuman did it, with half the time on strengths and half on blockers. It is a starting principle, not a rule proven for every product. Adjust the share if your very disappointed users mostly ask for improvements to what they already love.
What about a request from an important customer who is not among the fans?
Superhuman's method is to politely disregard somewhat-disappointed users whose main benefit differs from the fans'. If the customer is strategic for other reasons, decide that consciously and write it down, rather than dressing it up as a survey result.
How do you rank projects once the two halves are set?
Give each project a cost and an impact rating, each low, medium or high, then start with low-cost, high-impact projects. Superhuman did this to ship improvements immediately.
How often should you repeat the exercise?
Superhuman tracked the score weekly, monthly and quarterly, and rebuilt its roadmap every quarter. A quarterly rhythm is a good starting point if you have fewer users.
Can Second Brain run the survey or compute the score?
No, that is not its role. It helps you keep your answers, notes and decisions in one place and reread them. The survey and the very-disappointed percentage stay in your usual tools.