Discovery Report:
a teardown of a real idea
Cycling Cult is an AI coach for amateur road cyclists. We handed the idea to BHAG and four hours later had a finished Discovery Report: verdict, Segment Map, strategy, and a risk-testing plan. We're publishing it unedited.
The idea going in
Cycling Cult
A personal AI coach for amateur road cyclists: custom training plans, daily workouts, detailed analytics, an ask-anything AI chat, and motivation through streaks and comparing yourself to others.
- Market
- EU, UK, US, Canada, Australia
- Model
- B2C, $14.99/month subscription
The report's verdict
Course-correct — narrow the focus
Segment and job confirmed. Serious amateurs on an FTP plateau are a mass pattern and the idea's strong suit. The direction is right.
The entry point needs focus. Weekly planning is the one critical job where current solutions systematically fail. Everything else is nice-to-have that dilutes the product.
Gamification is an anti-pattern. For skeptical professionals aged 35–45, streaks and badges undercut trust. What motivates them is performance on the club ride, not a badge on a screen.
Next step: rebuild the MVP around a weekly AI planner. Three load-bearing assumptions with a cheap test for each are already in the risk map: two weeks of testing instead of six months of building blind.
Positive feedback isn't validation — it's politeness. The report answers with data: who needs it, why, what to build, and what to test first.
Inside the report
The report opens with the verdict; the Report Map leads to the rest — from methodology to segment dossiers. Every section opens in the inspector, just like in the product.
How to read this
The Executive Summary is the only must-read: the decision, the reasoning, and the next step. Everything else shows the work — where each conclusion comes from.
Executive Summary (TL;DR)
Verdict: Course-correct — narrow the focus
The original Cycling Cult idea — an adaptive AI app for cyclists — lands on a real job and a real segment. The direction is right. But the product as it stands is unfocused: a broad feature set (an ask-anything AI, daily workouts, streaks, comparing yourself with other riders) lowers the odds of fast PMF.
Why "course-correct" and not "stay the course"
1. The segment and the job are confirmed — that's the strength. Serious amateurs on an FTP plateau (35–45, an engineer's mindset, 3–5 workouts a week) are a widespread pattern. Their core pain is "The data's there — the answer isn't": they can't turn their power history into a concrete plan for the three windows they actually have. The original idea acknowledges the pain but doesn't build around it.
2. The entry point needs focus. The research shows that planning the training week is the only critical job where the current solutions (TrainerRoad, TrainingPeaks) systematically fail. That's the place to attack. The original idea spreads itself thin: "daily workouts," "ask the AI anything," "detailed analytics," "motivation through streaks." All of it is nice-to-have; none of it is core.
3. Gamification is an anti-pattern for this segment. The ICP is a skeptical 35–45-year-old professional, motivated by results on the club ride, not by virtual badges. Streaks, badges, and comparing yourself with others are mass-market B2C mechanics that can make the product come off as unserious and cost it trust.
What to change
Narrow to weekly planning as the core. The first screen after onboarding: a ready-made weekly plan (three sessions, zones, the logic behind them). Not "pick a workout from the library." Not "look at the charts and decide for yourself." A concrete plan, tied to your data and your schedule, instantly.
Cut or defer the motivational gamification. Streaks, badges, leaderboards — not at launch. Value gets proven by a public win on a club ride 4–6 weeks in, not by virtual rewards.
Test three critical assumptions before betting big:
- The planning pain is real and big enough to pay for (€15–30/month)
- FTP stagnation causes chronic anxiety, not a shrug of "that's just how it goes"
- AI plans earn the skeptical segment's trust from the very first one
The next step
Rebuild the concept around weekly planning: the product's first screen is a ready-made weekly plan built on the user's data. Streaks, badges, and the ask-anything AI stay out of the MVP.
Before serious money goes into development, run the three load-bearing assumptions through the cheap tests in the risk map: build a prototype AI planner (simple heuristics are enough), show 10–15 people who fit the ICP a personal plan built on their real power data, and measure trust and willingness to pay.
If the assumptions hold, the product has a strong foundation for PMF. If they don't, you'll know in two weeks of cheap tests — not six months of development.
Report Map
This summary is the tip of the iceberg. The full research is laid out in self-contained sections below: read straight through, or dive into any block on its own.
Part 1 · Context & Methodology
- — the product in plain words: why the idea is compelling and why the idea alone isn't enough.
- — people don't buy products, they "hire" solutions for a job; the competition isn't between features but between ways to get the job done.
- — the A→B transition we tested: not "who needs the product," but what people already spend time and money on.
- — behavioral reconstruction instead of surveys: a panel of job performers, cross-checking, and the method's honest limits.
Part 2 · Market Map
- — three market segments with composite scores across Size & Growth, Purchasing Power, Competition, and Operational Complexity.
- Segment dossiers — the job performer, the language, the job graph with Job Stories, the solutions they hire:
- — focus: serious amateur on a plateau, 73/100
- — self-coached event prep, 63/100
- — DIY system on an age-related plateau, 60/100
Part 3 · Strategy & Recommendations
- — why the bet is on Segment A and where the alternatives fall short.
- — the door you can get through faster than anyone else: the job where the pain runs sharpest.
- — the through line: where to play → where to start → where to grow.
- — a living portrait: the language, the fears, how he chooses.
- — the promise of a transition to Point B that people will pay for.
- — must-have / nice-to-have / irrelevant, with a criterion to settle every argument.
- — the contexts where the job shows up: the message, the first mile, and what brings people back.
- — load-bearing assumptions and minimal tests with pass/fail thresholds — before development starts.
Research Inputs
- · · ·
Where to start
— how the segments stack up against each other. — how deep each dossier goes. — what to test before you build. In the product, Jim answers questions on any section — in chat, grounded in the report's data.
The same report — for your idea
Describe your idea in chat — Jim will gather the inputs and prep the research for launch, all free. A verdict, segments, and an action plan in about 4 hours.
Questions about the report
The questions that come up most often after reading the sample report.
Discovery Report:
a teardown of a real idea
Cycling Cult is an AI coach for amateur road cyclists. We handed the idea to BHAG and four hours later had a finished Discovery Report: verdict, Segment Map, strategy, and a risk-testing plan. We're publishing it unedited.
The idea going in
Cycling Cult
A personal AI coach for amateur road cyclists: custom training plans, daily workouts, detailed analytics, an ask-anything AI chat, and motivation through streaks and comparing yourself to others.
- Market
- EU, UK, US, Canada, Australia
- Model
- B2C, $14.99/month subscription
The report's verdict
Course-correct — narrow the focus
Segment and job confirmed. Serious amateurs on an FTP plateau are a mass pattern and the idea's strong suit. The direction is right.
The entry point needs focus. Weekly planning is the one critical job where current solutions systematically fail. Everything else is nice-to-have that dilutes the product.
Gamification is an anti-pattern. For skeptical professionals aged 35–45, streaks and badges undercut trust. What motivates them is performance on the club ride, not a badge on a screen.
Next step: rebuild the MVP around a weekly AI planner. Three load-bearing assumptions with a cheap test for each are already in the risk map: two weeks of testing instead of six months of building blind.
Positive feedback isn't validation — it's politeness. The report answers with data: who needs it, why, what to build, and what to test first.
Inside the report
The report opens with the verdict; the Report Map leads to the rest — from methodology to segment dossiers. Every section opens in the inspector, just like in the product.
How to read this
The Executive Summary is the only must-read: the decision, the reasoning, and the next step. Everything else shows the work — where each conclusion comes from.
Executive Summary (TL;DR)
Verdict: Course-correct — narrow the focus
The original Cycling Cult idea — an adaptive AI app for cyclists — lands on a real job and a real segment. The direction is right. But the product as it stands is unfocused: a broad feature set (an ask-anything AI, daily workouts, streaks, comparing yourself with other riders) lowers the odds of fast PMF.
Why "course-correct" and not "stay the course"
1. The segment and the job are confirmed — that's the strength. Serious amateurs on an FTP plateau (35–45, an engineer's mindset, 3–5 workouts a week) are a widespread pattern. Their core pain is "The data's there — the answer isn't": they can't turn their power history into a concrete plan for the three windows they actually have. The original idea acknowledges the pain but doesn't build around it.
2. The entry point needs focus. The research shows that planning the training week is the only critical job where the current solutions (TrainerRoad, TrainingPeaks) systematically fail. That's the place to attack. The original idea spreads itself thin: "daily workouts," "ask the AI anything," "detailed analytics," "motivation through streaks." All of it is nice-to-have; none of it is core.
3. Gamification is an anti-pattern for this segment. The ICP is a skeptical 35–45-year-old professional, motivated by results on the club ride, not by virtual badges. Streaks, badges, and comparing yourself with others are mass-market B2C mechanics that can make the product come off as unserious and cost it trust.
What to change
Narrow to weekly planning as the core. The first screen after onboarding: a ready-made weekly plan (three sessions, zones, the logic behind them). Not "pick a workout from the library." Not "look at the charts and decide for yourself." A concrete plan, tied to your data and your schedule, instantly.
Cut or defer the motivational gamification. Streaks, badges, leaderboards — not at launch. Value gets proven by a public win on a club ride 4–6 weeks in, not by virtual rewards.
Test three critical assumptions before betting big:
- The planning pain is real and big enough to pay for (€15–30/month)
- FTP stagnation causes chronic anxiety, not a shrug of "that's just how it goes"
- AI plans earn the skeptical segment's trust from the very first one
The next step
Rebuild the concept around weekly planning: the product's first screen is a ready-made weekly plan built on the user's data. Streaks, badges, and the ask-anything AI stay out of the MVP.
Before serious money goes into development, run the three load-bearing assumptions through the cheap tests in the risk map: build a prototype AI planner (simple heuristics are enough), show 10–15 people who fit the ICP a personal plan built on their real power data, and measure trust and willingness to pay.
If the assumptions hold, the product has a strong foundation for PMF. If they don't, you'll know in two weeks of cheap tests — not six months of development.
Report Map
This summary is the tip of the iceberg. The full research is laid out in self-contained sections below: read straight through, or dive into any block on its own.
Part 1 · Context & Methodology
- — the product in plain words: why the idea is compelling and why the idea alone isn't enough.
- — people don't buy products, they "hire" solutions for a job; the competition isn't between features but between ways to get the job done.
- — the A→B transition we tested: not "who needs the product," but what people already spend time and money on.
- — behavioral reconstruction instead of surveys: a panel of job performers, cross-checking, and the method's honest limits.
Part 2 · Market Map
- — three market segments with composite scores across Size & Growth, Purchasing Power, Competition, and Operational Complexity.
- Segment dossiers — the job performer, the language, the job graph with Job Stories, the solutions they hire:
- — focus: serious amateur on a plateau, 73/100
- — self-coached event prep, 63/100
- — DIY system on an age-related plateau, 60/100
Part 3 · Strategy & Recommendations
- — why the bet is on Segment A and where the alternatives fall short.
- — the door you can get through faster than anyone else: the job where the pain runs sharpest.
- — the through line: where to play → where to start → where to grow.
- — a living portrait: the language, the fears, how he chooses.
- — the promise of a transition to Point B that people will pay for.
- — must-have / nice-to-have / irrelevant, with a criterion to settle every argument.
- — the contexts where the job shows up: the message, the first mile, and what brings people back.
- — load-bearing assumptions and minimal tests with pass/fail thresholds — before development starts.
Research Inputs
- · · ·
Where to start
— how the segments stack up against each other. — how deep each dossier goes. — what to test before you build. In the product, Jim answers questions on any section — in chat, grounded in the report's data.
The same report — for your idea
Describe your idea in chat — Jim will gather the inputs and prep the research for launch, all free. A verdict, segments, and an action plan in about 4 hours.
Questions about the report
The questions that come up most often after reading the sample report.