The Blinking Cursor
Picture the first Monday. You’ve just been hired into your first job as a product marketing manager at a startup (or the field marketing manager, same panic, different lanyard). Nobody on the team has held the job before, so there’s no predecessor, no handoff doc, and nobody to ask what the role is supposed to do here.
You have a laptop, a Slack invite, and an AI assistant that will answer anything you type.
The problem is the typing.
If you’ve never done the job, you don’t know what to ask for, so you can’t tell a great answer from a confident but incorrect one. That’s the question I’m intrigued by:
If you have the most capable assistant ever built but no idea what to ask it, how far does it get you?
A while back I wrote about what to do when the founder is the PMM, and the short version was “you ARE the PMM, so just start.” Seemed like a fair enough glib statement … but the obvious follow-up to “just start” is “start with what?” This post is my attempt at an honest answer to that question, and the answer turned out to be more interesting (and more uncomfortable) than I expected.
So I ran the experiment. Five prompts, a blank thread, a (kinda) real product, no coaching. Then I graded every output the way I’d grade a new hire’s first week. I’ll spoil the outcome up front … the outputs were better than I expected… and that is exactly what makes them a little scary.
Note: this post is long on purpose. I’m showing the full outputs, not summaries of them, because the details are where the interesting insights are.
The Setup and the Rules
The test subject is The Flexible Cyclist, a recovery companion app for cyclists. This was a little side project I built, it’s a real app I have running on my devices via TestFlight, it just isn’t published yet.
I figured this made it a safe stand-in for a startup product: real enough to have a web page and a point of view, early enough that there are no customers, no sales calls, and no win/loss notes. Hold onto that last part. It pops up later.
The rules:
- Blank thread. I stripped out everything the assistant knew about me, my sites, and my opinions on product marketing, and explicitly instructed Claude to ignore all CLAUDE.md and history it knew about me.
- Five prompts, typed the way a new hire would type them. No prompt-engineering tricks or leading the AI.
- No follow-ups. No corrections, no coaching, no “try again, but better” between rounds.
- Outputs unedited. What you read below is exactly what the prompt returned.
- Graded like a first week. For each round I ask what the new hire could use on Monday and what could potentially steer them wrong.
Here’s the flow:
| Round | What the new hire types | What they’re really trying to get |
|---|---|---|
| 1 | “I just got hired as the first product marketing manager at a startup and I have no idea what that means. Where do I start?” | Orientation |
| 2 | “Can you make me a 30/60/90 day plan?” | A plan to show the boss |
| 3 | “Write me a positioning statement for our product.” | The core of the job |
| 4 | “Who are our competitors and how do we beat them?” | Ammunition for sales |
| 5 | “My CEO wants a launch plan for next month’s release. Can you write one?” | Something to survive the first real deadline |
Note: this is one run, one model, one product, and a seasoned PMM pretending to be new … it’s likely there is bias and slanted intent even with what I wrote. Where the assistant needed to know about the product (rounds 3 to 5), it was working from the (very basic) web page.
Round 1: Where Do I Even Start?
The prompt:
I just got hired as the first product marketing manager at a startup and I have no idea what that means. Where do I start?
What came back, unedited:
Round 1 output: full text
# Starting as the First Product Marketing Manager at a Startup
## What product marketing is
You're the translator between the product, the market, and the go-to-market teams. In practice, you own four things:
1. **Positioning and messaging:** who the product is for, what problem it solves, why it's different, and how to say that in a way buyers actually care about.
2. **Launches:** getting new products and features to market in a coordinated way, so sales, support, and marketing all say the same thing.
3. **Sales enablement:** giving sales what they need to win, such as pitch decks, one-pagers, objection handling, and competitor comparisons.
4. **Market and customer insight:** feeding what you learn about customers, competitors, and the market back to product and leadership.
It sits between product management (what to build) and demand generation (how to get leads). Your job is making sure the right people understand why the product matters.
## First 30 days: listen before you write anything
- **Talk to 10-15 customers.** Ask why they bought, what they tried before, what almost stopped them, and how they describe the product to colleagues. Their exact words are your best messaging raw material. Include some who churned or chose a competitor if you can.
- **Sit in on sales calls** (or listen to recordings). Note which questions come up repeatedly, where deals stall, and which objections sound familiar.
- **Interview internal stakeholders:** the founders, the head of product, sales, customer success, and support. Ask each what they think the product does, who it's for, and who the top competitors are. The gaps between answers show you where the work is.
- **Audit what exists:** website, deck, pitch, sales materials. Look for inconsistency in who the product is for and what it promises.
- **Map the competition.** Find out who customers actually compare you to, which isn't always who the founders think.
## Days 30-90: produce the foundation
- **Write a positioning document.** Cover target customer, the problem, your alternatives, your unique value, and why it matters. Get the founders to agree on it. At an early startup, this is the most valuable thing you can ship.
- **Turn it into core messaging:** a one-line pitch, a few value pillars, and proof points.
- **Fix the highest-leverage asset first.** That's usually the website homepage or the sales deck, wherever the most deals are won or lost.
- **Build one repeatable launch process,** even a lightweight one, so product releases stop being surprises.
- **Create a simple competitive cheat sheet** for sales.
## Things that trip up first-time PMMs
- **Trying to do everything.** You'll be asked for blog posts, decks, event help, and more. Anchor to positioning and launches, and say no or "later" to the rest. Ask your manager for your top three priorities.
- **Writing before researching.** Messaging based on your assumptions rather than customer language is the most common failure.
- **Working in isolation.** Positioning only works if sales, product, and leadership believe it. Get them involved early.
- **Not defining success.** Agree on metrics with your manager, such as win rate, sales cycle length, launch adoption, or pipeline quality, so your impact is visible.
## Two questions to ask your manager this week
1. What's the one business problem you most want me to help solve in the first six months?
2. Who owns what between product, marketing, and sales, so I know where my lane starts and ends?
What worked
- A plain-English job description a beginner can follow: four things you own, and where PMM sits between product management and demand generation.
- “Listen before you write anything” is a good instinct, and it’s specific (10 to 15 customers, including the ones who left).
- The “things that trip up first-time PMMs” list names some real traps for new players, including writing before researching and trying to do everything.
- Two questions to take to your manager this week. Concrete, doable, and what a nervous new hire needs on Day 1.
What didn’t
- It never asked a single question about the company: B2B or consumer, sales-led or self-serve, pre-launch or three years old. Every piece of advice assumes there are customers to interview, sales calls to sit in on, and a sales team to enable.
- Because it can’t tell what it doesn’t know, it sounds equally sure about everything. Nothing in here says “this depends on your company and product.”
- It’s a good description of the job at a typical B2B startup. For the product I gave it, that’s not what this is.
TL;DR Good textbook response, wrong approach for the scenario. (Which it couldn’t have known, because nobody told it yet, but that’s the red flag to see.)
Round 2: Okay… Can You Make Me a Plan?
The prompt:
Can you make me a 30/60/90 day plan?
What came back, unedited:
Round 2 output: full text
# 30/60/90 Day Plan: First Product Marketing Manager
**Goal of the first 90 days:** Learn the business deeply, establish agreed positioning and messaging, ship a first launch, and put repeatable foundations in place that make sales and marketing more effective.
---
## Days 1-30: Learn
**Theme: Listen, absorb, and build credibility. Ship nothing big yet.**
### Objectives
- Understand the product, customers, market, and competitors from the people closest to them
- Build relationships across product, sales, marketing, customer success, and leadership
- Agree with your manager on priorities and how success will be measured
### Actions
- [ ] Meet with your manager in week 1 to confirm top three priorities, success metrics, and decision rights
- [ ] Complete a full product walkthrough and use the product yourself as a new customer would
- [ ] Interview founders and leaders: what is the product, who is it for, who are the competitors, and why do customers win or lose?
- [ ] Interview 10-15 customers (include a few churned customers and lost deals if possible)
- [ ] Listen to 10+ recorded sales calls and shadow live calls
- [ ] Interview sales, customer success, and support leads about common questions and objections
- [ ] Audit existing assets: website, pitch deck, sales materials, emails, onboarding, and past launches
- [ ] Build an initial competitor list based on who customers actually compare you to
- [ ] Set up a shared place (doc or folder) for customer quotes, objections, and insights
### Deliverables
- Insights summary: top themes from customers, sales, and internal interviews (1-2 pages)
- Messaging and asset audit: what's inconsistent, outdated, or missing
- Draft list of priorities for the next 60 days, reviewed with your manager
### Success check at day 30
- You can explain the product, customer, and market clearly without notes
- Your manager agrees on your priorities and success metrics
- You have a clear view of the biggest gaps between how the company describes itself and how customers see it
---
## Days 31-60: Define
**Theme: Turn what you learned into positioning, messaging, and the first usable assets.**
### Objectives
- Get leadership alignment on who the product is for and why it wins
- Translate positioning into messaging the team can actually use
- Start fixing the highest-leverage assets
### Actions
- [ ] Write the positioning document: target customer, problem, alternatives, unique value, proof, and why it matters
- [ ] Workshop the draft with founders, product, and sales; revise until there is real agreement
- [ ] Develop core messaging: one-line pitch, elevator pitch, 3-4 value pillars, and proof points for each
- [ ] Define the ideal customer profile and key buyer personas, using real customer data
- [ ] Build a one-page competitive cheat sheet for sales (strengths, weaknesses, how to win)
- [ ] Update the single highest-leverage asset first (usually the sales deck or website homepage)
- [ ] Pick an upcoming product launch or release and start planning it
- [ ] Set up a regular channel with sales for feedback (e.g., short monthly sync, shared Slack channel)
### Deliverables
- Positioning document, signed off by leadership
- Messaging framework
- Competitive cheat sheet
- Updated sales deck or homepage (first version)
- Launch plan for the first launch
### Success check at day 60
- Founders and sales leaders can state the positioning the same way
- Sales has at least one new asset they are actually using
- A first launch has an owner, a timeline, and a clear goal
---
## Days 61-90: Deliver
**Theme: Launch, enable, and build repeatable processes.**
### Objectives
- Execute your first launch end to end
- Enable sales to use the new positioning
- Put in place lightweight systems so the work scales
### Actions
- [ ] Run the first launch: internal briefing, customer-facing messaging, announcement, and sales enablement
- [ ] Train the sales team on positioning, messaging, and competitive handling; collect feedback
- [ ] Create a simple launch process: tiers (major, minor), checklist, timeline, and owners
- [ ] Build a basic win/loss feedback loop (short post-deal notes or interviews)
- [ ] Set up a repeatable customer-insight routine (e.g., 2-3 customer conversations per month)
- [ ] Review early results: sales feedback, launch performance, and any change in win rate, sales cycle, or lead quality
- [ ] Present a 90-day readout to leadership and propose the next-quarter roadmap
### Deliverables
- First launch completed, with a short retrospective
- Sales enablement session delivered and materials in a shared location
- Launch process template
- Win/loss and customer-insight process
- 90-day readout and next-quarter plan
### Success check at day 90
- One launch shipped with sales and marketing aligned
- Sales uses your messaging and assets in real deals
- Leadership understands what PMM does and has seen the impact
- You have a prioritized next-quarter roadmap
---
## Metrics to agree with your manager
Pick a small set, and agree on baselines early:
| Area | Example metrics |
|---|---|
| Sales impact | Win rate, sales cycle length, competitive win rate |
| Launch | Feature adoption, launch-driven pipeline, sales readiness |
| Messaging | Website conversion, sales feedback, message consistency |
| Insight | Number of customer conversations, insights adopted by product |
## Risks to watch
- **Scope creep:** protect time for positioning and launches; push back politely on ad hoc requests
- **Skipping research:** resist drafting messaging before talking to customers
- **Lack of buy-in:** involve sales and product early so positioning isn't a surprise
- **No baseline:** capture current numbers in the first month so you can show change later
What worked
- The shape of the response is correct. Learn, Define, Deliver, with a theme per phase and success criteria at each milestone. “Ship nothing big yet” is the good restraint a new hire needs to hear.
- It’s consistent with Round 1. Same advice, now with deliverables and dates attached.
- You could paste this into a doc and hand it to a manager tomorrow, and it would start the right conversation. The metrics table and the “agree on baselines early” line are the kind of detail people usually forget.
- The risks list names real problems: scope creep, skipping research, no baseline.
What didn’t
- It’s the same plan for any generic company. There are 24 checkboxes (!) and most of them assume an established sales team, recorded sales calls, lost deals to study, and a sales deck to fix. In this case, our company has none of that.
- It never says what to do when a box can’t be checked. If there are no sales calls to listen to, the plan never says what replaces them. A new hire won’t know to ask.
- The timeline is very hopeful, not a sequential plan. “Positioning document, signed off by leadership” by Day 60 is a fine goal and an enormous assumption about how fast people can agree on anything.
- Same confident tone as Round 1. Nothing says “adjust this once you know more.”
TL;DR A plan is a hypothesis you take to your manager to communicate intent. This one is a good hypothesis about a company that isn’t ours.
Round 3: Write Me a Positioning Statement
The prompt:
Write me a positioning statement for our product.
What came back, unedited:
Round 3 output: full text
# Positioning Statement: The Flexible Cyclist
*Draft based on the homepage at https://theflexiblecyclist.app. The homepage describes a recovery companion app for cyclists on iPhone, iPad, and Mac, listed as "Coming soon" on the App Store.*
## Positioning statement
> For cyclists who put in the miles but skip the recovery work, The Flexible Cyclist is a recovery companion app that makes stretching, mobility, strength, and core work easy to build, log, and stick with in short sessions, right alongside your ride data in Apple Health. Unlike generic fitness or stretching apps, it's built around what riding does to your body, and it treats recovery as a small daily habit rather than a second training plan.
## One-line version
> Recovery for cyclists, in small habits that fit around your riding.
## Breakdown
| Element | Draft |
|---|---|
| Target customer | People who ride regularly and want to stay ready for the next ride |
| Problem | Riding alone doesn't do the recovery work, and a full routine feels like a second training plan |
| Category | Recovery companion app for cyclists |
| Key benefit | Consistent recovery habits with little time and effort |
| Differentiators | Cycling-specific focus, short-session streaks, Apple Health and iCloud integration |
| Proof | None yet |
## What the homepage says today
- **Headline:** "Your legs did the work. Now give a little back."
- **Tagline:** "Recovery for cyclists"
- **Section header:** "Small habits. Smoother miles."
- **Features:** custom routines (stretching, mobility, strength, core), session logging alongside riding data, streak tracking where short sessions count, Apple Health sync, iCloud sync
- **Platforms:** iPhone for logging, iPad for following along, Mac for planning
- **Pricing, social proof, named competitors:** none stated
## Gaps to close before this is final
- **Alternatives aren't named.** The page doesn't say what riders do today (YouTube stretching videos, generic apps like Strava or Apple Fitness+, a physio's handout, nothing). The "unlike" clause is a guess until confirmed with riders.
- **The audience is broad.** "People who ride" covers commuters through racers. Picking one primary segment would sharpen the statement a lot.
- **There's no proof or pricing.** No testimonials, user counts, or results. Early beta users would be the fastest way to get quotes.
- **Apple-only is both a strength and a limit.** It's a clear differentiator, but it excludes Android riders. Say so deliberately.
## Next steps
1. Talk to 5-10 riders about how they recover today, and validate or rewrite the statement using their words.
2. Choose a primary segment and write a version for it.
3. Collect early user quotes for proof.
4. Align the homepage headline and subhead with the final statement.
What worked
- It says where it got its information: “Draft based on the homepage.” It lists what the homepage says today, and it separates what it read from what it guessed.
- “Proof: None yet” is an honest call, and it didn’t invent a testimonial to fill it. That alone puts it ahead of a lot of human first drafts.
- The gaps section is good: alternatives not named, audience too broad, no proof or pricing, Apple-only as both strength and limit. A seasoned PMM would write most of that list.
- The statement itself is clean and the one-liner is usable.
What didn’t
- It’s my web page in a trench coat. The statement is assembled from what I already said about myself, so it can’t be wrong … and it can’t tell me anything new either.
- “People who ride regularly” is not a segment. A commuter, a gravel racer, and a weekend roadie have three different problems, and the statement changes with each one. The assistant noticed and left it to me.
- The order is backwards. Round 1 told this person that writing before researching is the most common failure. Round 3, on request, writes first, and puts “talk to 5-10 riders” at the bottom as step 1 of the next steps. The caveats live in the last third. The copy-pasteable statement lives at the top.
- The “unlike generic fitness apps” line is a guess. The assistant says so, but only in the gaps section.
TL;DR The tell is that the assistant obeyed my prompt over its own advice. That’s a critical thing to remember about AI. By default it will agree and amplify your approach.
Round 4: Who Are Our Competitors, and How Do We Beat Them?
The prompt:
Who are our competitors and how do we beat them?
What came back, unedited:
Round 4 output: full text
# Competitive Landscape: The Flexible Cyclist
*Based on web research (November 2026) and the homepage at https://theflexiblecyclist.app. Pricing and feature details change often; verify before using any of this in sales or marketing material.*
## The short version
The Flexible Cyclist doesn't have one clear head-to-head rival. It sits where three categories overlap, and each category has strong players that only partly solve the problem:
1. **Cycling-specific strength and mobility apps** (closest rivals)
2. **General stretching and mobility apps** (biggest by audience)
3. **Cycling training and recovery-data platforms** (own the rider's attention and data)
The biggest competitor is probably **doing nothing, or watching free YouTube videos**. Most riders already know they should stretch. The product's job is to make them actually do it.
---
## Tier 1: Cycling-specific competitors (closest)
| Competitor | What it is | Notes |
|---|---|---|
| **Dynamic Cyclist** | Cycling-specific stretching, mobility, strength, yoga, and foam rolling programs, with daily 15-20 minute routines | Free with in-app purchases. A reviewer found its Strava link rudimentary. It is video-program based, so it's a "follow the content" product rather than a habit and logging tool. |
| **Dialed Health** | Strength and mobility programs for cyclists, in 4-16 week cycles, with a "minimum dose" philosophy | Small ratings base (about 40 ratings at 4.65 when I checked). Syncs with Strava and has a calendar. Leans toward strength more than recovery. |
| **RideStrong** | Cycling strength and mobility platform from a performance and rehab expert | One comparison site ranked it the strongest dedicated cycling-gym option available. Strength-led, with credibility from its founder. |
| **GOWOD** (cycling page) | Personalized mobility and stretching routines, with 8 or 12 minute protocols and a mobility assessment | Sport-specific pages for cycling, but it is a general mobility product at heart. Has a free tier of 300+ exercises. |
| **Roadman** (upcoming) | iPhone app coordinating strength and recovery around your riding week | Early access, not launched when I checked. The most direct future threat: iPhone, cycling-specific, and positioned around fitting recovery into the training week. |
## Tier 2: General stretching and mobility apps
| Competitor | Positioning | Price signals |
|---|---|---|
| **Pliability** (formerly ROMWOD) | Mobility and recovery for athletes, with a mobility test and 12-20 minute sessions | Roughly $15-18/month depending on source and date |
| **StretchIt** | Goal-oriented stretching programs (e.g., splits), personalized by quiz | About $10/month or $90/year per one comparison |
| **Bend, Down Dog, Stretch Mode, and others** | Daily stretching for general wellness | Mostly freemium |
**Where they're weak** (from a review analysis of 1-3 star reviews across popular stretching apps): paywalled routines, sessions built for gymnasts, hard-to-cancel subscriptions, and content too long for real life. A cyclist after a ride wants none of those things.
## Tier 3: Training and recovery platforms
| Competitor | Role | Gap for cyclists |
|---|---|---|
| **Strava** | Ride log and community | Tracks rides, not recovery work |
| **TrainingPeaks** (including Strength) | Structured training and coach workflows | Strength and stretching plans are available, but it's a coach and plan tool, not a recovery habit app |
| **Wahoo SYSTM** | Indoor training with strength and yoga modules | Tied to its own training ecosystem |
| **Whoop, Wattson Blue, Athlytic** | Readiness and recovery scores from HRV and sleep | Tell you *how recovered you are*, not *what to do about it* |
| **Apple Fitness+** | Guided workouts for the Apple ecosystem | General fitness, not cycling-specific |
These aren't direct rivals today, but they own the rider's data and attention, and any of them could add a recovery-routine feature.
---
## How to win
### 1. Own "the smallest habit that sticks"
Competitors sell *programs* and *content* (15-20 minute videos, multi-week plans). The homepage promise is "small habits," with streaks where 5 or 10 minutes counts. Lean into this fully. The gap competitors leave open is real: even stretching-app reviews complain about sessions that are too long and not built for real life.
**Play:** Make "five minutes counts" the core message and the core product experience. Position against programs that demand a second training schedule.
### 2. Be the recovery layer next to the ride, not another training plan
Roadman and the training platforms fit recovery around a training calendar. Your edge is the opposite: logging recovery *beside* ride data in Apple Health, with no coaching system required.
**Play:** "Works with whatever you already use to ride." Don't ask riders to switch ride trackers or training plans.
### 3. Win on the Apple ecosystem
iPhone for logging, iPad for following along, and Mac for planning, with Apple Health and iCloud sync. Most cycling-specific competitors lean on video content or Strava links rather than deep Apple Health integration.
**Play:** Make Apple Health integration visibly excellent (recovery sessions appear next to rides and workouts), and ship the Apple Watch and widget features riders in the ecosystem expect. Treat Android exclusion as a deliberate focus, not a gap.
### 4. Sharpen the cycling specificity
General apps are the cheapest to beat on this. They teach generic stretches; you can speak to hip flexors after hours in the saddle, neck and shoulders, lower back, and so on. Dynamic Cyclist and RideStrong already do this, so specificity gets you in the door but won't win alone.
**Play:** Build routines named for riders' real complaints ("after a long ride," "tight hips on the bike," "pre-ride 5 minutes") rather than body parts.
### 5. Fix what users hate about subscription apps
Hard-to-cancel subscriptions, paywalled core features, and content that doesn't fit real life are the recurring complaints in this category.
**Play:** Be transparent on pricing, easy to cancel, and generous in the free tier. Pricing isn't on the homepage yet, so this is a decision still open to you.
---
## Where you're exposed
- **No visible proof.** The homepage has no reviews, user counts, or press. Competitors have ratings and expert-backed credibility (RideStrong, Dynamic Cyclist).
- **Unclear content model.** The homepage emphasizes *building* routines, but doesn't say whether the app ships with guided content. If riders have to build everything themselves, that's a heavy lift next to video-led rivals. Confirm this and make it clear in messaging.
- **Pre-launch.** The app is listed as coming soon, while competitors are established and Roadman is moving in the same direction.
- **Apple-only.** A strength for focus, but it caps the market.
- **Easy substitutes.** Free YouTube content and the Strava or Apple Fitness+ subscriptions riders may already pay for set a low price anchor.
## Suggested next steps
1. **Download and use the top five competitors** (Dynamic Cyclist, Dialed Health, RideStrong, GOWOD, Pliability). Take notes on onboarding, session length, pricing, and cancellation.
2. **Interview 8-10 riders** about how they recover today and why they do or don't stretch. This validates which of the plays above matters most.
3. **Read competitor app store reviews** (1-3 star especially) for language to use in your messaging and problems to solve.
4. **Decide the primary segment** (e.g., weekend riders, commuters, gravel, serious amateurs). Competitors split along this line, and it changes who your real rival is.
5. **Monitor Roadman's launch** and any recovery features from Strava, Whoop, and TrainingPeaks.
## Sources
- GOWOD cycling page: https://www.gowod.app/sports/cycling
- Dynamic Cyclist (App Store): https://apps.apple.com/be/app/dynamic-cyclist/id6444422797
- Dynamic Cyclist review (Road Bike Rider): https://www.roadbikerider.com/dynamic-cyclist-app-review/
- Dialed Health (App Store): https://apps.apple.com/us/app/dialed-health/id1623502533
- RideStrong profile: https://movement.so/customers/ridestrong
- Best cycling strength training apps (Roadman, disclosed competitor): https://roadmancycling.com/best/best-cycling-strength-training-apps
- Roadman app: https://roadmancycling.com/app
- Pliability vs StretchIt: https://pliability.com/stories/stretchit-vs-pliability
- Stretching app complaints analysis (Unstar): https://unstar.app/blog/bend-stretchit-pliability-gowod-stretching-apps-ranked-2026
- Whoop recovery: https://www.whoop.com/ae/en/recovery/
- Wattson Blue (App Store): https://apps.apple.com/us/app/wattson-blue/id1270093102
- Cycling apps overview (GOWOD blog): https://www.gowod.app/pt/blog/best-cycling-apps-for-training-tracking-and-performance
What worked
- The frame. Three overlapping categories instead of a fake head-to-head list, with “doing nothing, or watching free YouTube videos” named as the biggest competitor. That is the most seasoned sentence in the whole experiment.
- It went and got information. This is the first round where the assistant did research instead of restating what I gave it, and it dated that research (November 2026) and told me to verify before using any of it.
- Mining 1-3 star reviews for what people hate is a classic PMM move, and “too long for real life” is a real insight for a product built around five-minute sessions.
- The “where you’re exposed” list is candid, and it raises a question I now have to answer: whether the app ships with guided content or is a blank builder.
- The sources are listed so you can check every claim.
What didn’t
- Look at who is vouching for whom. One source is a “best apps” article from Roadman, which the output itself labels as a disclosed competitor. Another is a comparison site’s ranking, which became a table cell (“the strongest dedicated cycling-gym option available”) with no hint that it’s one site’s opinion. A seasoned PMM asks “who benefits from this source?” before using it. A new hire might not.
- Some of it is vague in ways that look precise. “Roughly $15-18/month depending on source and date.” “About 40 ratings at 4.65 when I checked.” When you checked is not a date I can cite, and this is the kind of detail that gets into a sales deck and then into an argument with a customer.
- It’s outside-in only: marketing pages, app store listings, and review sites. No customer has said any of this to us. For a pre-launch product that’s the best available, but the output never says “this is the weakest kind of evidence.” It just sounds thorough.
- Five “plays” with no ranking. If you only get one, the output never says which. The new hire is left to guess.
Note: I have not actually verified these competitor claims, and you shouldn’t take my word that they’re right either. This is a thought experiment that just happens to use a side project I built for myself.
TL;DR The assistant did the legwork in minutes, but every line still needs a human to check it and be accountable. That’s Zero-Trust fact-checking and not something a typical newbie knows.
Round 5: The CEO Wants a Launch Plan
The prompt:
My CEO wants a launch plan for next month’s release. Can you write one?
What came back, unedited:
Round 5 output: full text
# Launch Plan: The Flexible Cyclist (December 2026 Release)
**Prepared for:** CEO review
**Status:** Draft v1, built from the public homepage and the positioning and competitor work. Items marked **[CONFIRM]** are assumptions that need an answer from the team before this plan is final.
---
## 1. Summary
The Flexible Cyclist is a recovery companion app for cyclists on iPhone, iPad, and Mac. The homepage currently lists it as "Coming soon" on the App Store. This plan covers the run-up to, and the first 30 days after, the December release.
**Launch goal:** get the right cyclists to download the app, complete a first recovery session, and come back for a second. Early *activation and retention* matter more than raw downloads for a new habit app.
**Core message:** *Recovery for cyclists, in small habits that fit around your riding.* Five-minute sessions count, and recovery sits alongside ride data in Apple Health.
**Launch tier:** Major launch (first public release). Lean execution, sized for a very small team.
---
## 2. Assumptions to confirm [CONFIRM]
| # | Assumption | Why it matters |
|---|---|---|
| 1 | Target release date is a specific day in December. **Date: ______** | Every milestone below is counted back from it (T-minus). |
| 2 | The app is already submitted, or will be submitted, to App Store review at least 7-10 days before release | A review delay is the biggest schedule risk. |
| 3 | Pricing and model are decided (free, freemium, paid, subscription). **Model: ______** | Messaging, the page, and the App Store listing all depend on it. |
| 4 | The app ships with starter routines, not only a blank builder | Determines whether we can promise "start in 60 seconds." |
| 5 | We have a small beta or waitlist group to draw testimonials from | Needed for social proof, which the site currently lacks. |
| 6 | Budget for paid promotion is minimal or zero | The plan below relies on owned and earned channels. |
| 7 | There is no Android version, and we say so openly | Avoids confusion and negative reviews. |
---
## 3. Goals and success metrics
**Targets are placeholders. Set real numbers with the CEO once the baseline (waitlist size, current traffic) is known.**
| Stage | Metric | Target (30 days post-launch) |
|---|---|---|
| Awareness | Landing page visits, waitlist signups | **[set]** |
| Acquisition | App Store page views to installs conversion | **[set]** |
| Activation | % of new users who complete a first session within 48 hours | **[set]** |
| Habit | % of users with 3+ sessions in the first 14 days | **[set]** |
| Retention | Week-4 retention | **[set]** |
| Proof | App Store ratings and written reviews | **[set, e.g., first 25 reviews]** |
| Learning | Rider interviews completed | 10+ |
**North-star metric:** users who complete 3+ recovery sessions in their first two weeks.
---
## 4. Audience
**Primary (launch focus):** Regular riders who ride several times a week, own an iPhone, and know they should stretch but don't consistently. **[CONFIRM segment: road, gravel, commuter, indoor/trainer]**
**Why this segment first:** They feel the problem (tight hips, back, and legs), already track rides in an Apple or Strava ecosystem, and talk about gear and habits with other riders.
**Secondary (opportunistic):** Newer riders building a routine, and cycling coaches or physios who might recommend it.
**Out of scope for launch:** Android users, non-cyclists, and competitive racers who need a full coaching platform.
---
## 5. Positioning and messaging
**Positioning statement:** For cyclists who put in the miles but skip the recovery work, The Flexible Cyclist is a recovery companion app that makes stretching, mobility, strength, and core work easy to build, log, and stick with in short sessions, right alongside ride data in Apple Health.
**Message pillars**
1. **Small habits, smoother miles.** Five or ten minutes counts. Recovery shouldn't feel like a second training plan.
2. **Built for how cyclists ride.** Routines for the body demands of the saddle.
3. **Sits next to your ride data.** Sessions appear in Apple Health. Nothing to replace.
4. **Works across your Apple devices.** Log on iPhone, follow along on iPad, plan on Mac. iCloud keeps it in sync.
**Proof points needed:** beta user quotes, a cycling coach or physio endorsement **[CONFIRM if available]**, and early usage stats once we have them. Do not make health or injury-prevention claims we cannot support. The homepage already carries a general-guidance disclaimer, so keep launch copy consistent with it.
---
## 6. Timeline (counting back from launch day, "L")
### Phase 1: Foundations (L-4 weeks to L-3 weeks)
- [ ] Confirm launch date, pricing, and the assumptions in Section 2
- [ ] Finalize positioning and one-line pitch with CEO and product
- [ ] Interview 5-10 beta users or riders; collect quotes and permission to use them
- [ ] Draft the App Store listing: name, subtitle, description, keywords, screenshots, preview video
- [ ] Define analytics events (install, first session, session count) so the metrics above are measurable
- [ ] Choose launch-day owner for each channel (Section 8)
### Phase 2: Build assets (L-3 weeks to L-1 week)
- [ ] Update homepage headline, proof, pricing, and a clear App Store download button
- [ ] Produce App Store screenshots and a 15-30 second preview video showing the first-session flow
- [ ] Write the launch email to the waitlist, plus a short follow-up sequence (Section 7)
- [ ] Prepare social posts and short demo clips for launch week
- [ ] Write a founder story or "why we built this" post
- [ ] Create a one-page press and creator kit: description, screenshots, logo, founder bio, contact
- [ ] Build a short list of 30-50 cycling creators, bloggers, podcasts, clubs, and coaches for outreach; send early access to the best 10-15
- [ ] Submit the app for App Store review (target **L-10 days** at the latest)
- [ ] Prepare a support inbox and FAQ (Apple Health permissions, iCloud sync, pricing, Android plans)
### Phase 3: Pre-launch warm-up (L-1 week to L-1 day)
- [ ] Send "launching on [date]" teaser to waitlist
- [ ] Share early-access links with creators and coaches
- [ ] Run a dry run: install from the App Store (or TestFlight), complete a first session, check Apple Health sync
- [ ] Final check of listing copy, links, analytics, and support readiness
- [ ] Brief the team on launch-day roles and the message
### Phase 4: Launch week (L-day to L+7)
- [ ] **L-day:** send launch email, update homepage, post on all channels, notify creators and partners
- [ ] Founder post with a personal story and a direct download link
- [ ] Monitor installs, crashes, and review feedback several times a day
- [ ] Reply to every review and comment, and fix urgent bugs fast
- [ ] Ask happy early users for an App Store rating at a sensible moment (after a completed session, not on first open)
- [ ] Share user quotes and screenshots as they come in
### Phase 5: Sustain (L+1 week to L+30 days)
- [ ] Send onboarding and habit-building emails or in-app prompts to new users
- [ ] Publish 2-3 content pieces (e.g., "5-minute post-ride routine," "why cyclists get tight hips")
- [ ] Interview 10+ new users: what brought them in, what nearly stopped them, what they would change
- [ ] Review metrics weekly; adjust messaging and the App Store listing based on results
- [ ] Write a launch retrospective for the CEO by **L+30**: results vs. goals, learnings, and the next-quarter plan
---
## 7. Channels and tactics
| Channel | Tactic | Owner | Notes |
|---|---|---|---|
| **Waitlist / email** | Teaser, launch-day email, 3-email onboarding series | **[CONFIRM]** | Highest-intent audience; our best day-one lever |
| **Website** | Clear hero, download button, proof, FAQ | PMM + design | Page must match the App Store listing |
| **App Store** | Keyword-tuned listing, screenshots, preview video | PMM + product | Where most installs happen |
| **Social** | Short demo clips; posts in cycling communities (without spamming) | **[CONFIRM]** | Lead with the problem, not the app |
| **Creators / coaches / clubs** | Early access, honest ask for feedback | PMM | Trust beats reach for a niche audience |
| **Cycling communities** | Authentic founder participation on cycling forums and subreddits that allow it | Founder | Follow each community's rules on self-promotion |
| **Press / newsletters** | Short pitch to cycling and Apple-focused outlets | PMM | Low odds; low cost. Founder story is the hook |
| **Product Hunt or similar** | Optional launch listing | **[CONFIRM]** | Skews tech-first, not cyclist-first; lower priority |
**Timing note:** December is gift season and many riders are in the winter off-season or riding indoors, when recovery and mobility work are on their minds. January resumes "new year, new habits." Both are natural content hooks. A gift angle is only worth using if pricing makes it practical.
---
## 8. Roles
| Role | Responsibility |
|---|---|
| **CEO / Founder** | Founder story, outreach to creators, final calls on pricing and date |
| **PMM (you)** | Positioning, messaging, launch plan, App Store copy, enablement, post-launch learning |
| **Product / Engineering** | Release readiness, App Store submission, analytics events, bug fixes |
| **Design** | Screenshots, preview video, website updates |
| **Support** | Inbox, FAQ, review replies **[CONFIRM who]** |
---
## 9. Risks and mitigations
| Risk | Mitigation |
|---|---|
| App Store review delay | Submit by L-10 days; keep a fallback launch date; avoid announcing hard dates until approved |
| Low awareness at launch | Build the waitlist and creator relationships now; focus on one channel done well |
| Users install but never do a session | Make the first session under 2 minutes; test onboarding with real riders; track activation |
| No social proof at launch | Collect beta quotes now; ask for ratings after a completed session |
| Unclear value vs. free videos and bigger apps | Lead with "small habits" and "sits next to your ride data" rather than feature lists |
| Health-claim risk | Keep copy to general guidance; avoid promises about injury prevention or treatment |
| Apple-only limits the audience | Say it plainly; capture Android interest through a waitlist to size demand |
| Small team stretched thin | Cut to the essentials: App Store listing, homepage, waitlist email, creator outreach |
---
## 10. Budget and resources
**Assumed lean:** mostly time. Possible costs: video or screenshot production, email tool, small paid test on one channel **[CONFIRM]**. If there is no budget, drop paid promotion and keep the channels above.
---
## 11. Decisions needed from the CEO this week
1. Confirm the launch date and who owns the App Store submission.
2. Confirm pricing and model.
3. Confirm the primary audience segment for launch.
4. Confirm who owns social, email, and support.
5. Agree on the three success metrics that will define a good launch.
---
## 12. After launch
At **L+30**, deliver a one-page readout: what we hit, what we missed, what riders told us, and the proposed next 90-day priorities (retention improvements, content, partnerships, and whether to explore Android or Apple Watch support).
What worked
- This is the most honest output of the five. It labels itself Draft v1, lists seven assumptions it can’t know, marks them [CONFIRM], and ends with five decisions for the CEO. That’s what a good first draft looks like.
- It left the targets as [set] instead of inventing numbers. I was expecting a bunch of fake benchmarks to be used and it didn’t do that.
- The north-star metric (3+ sessions in the first two weeks) is the right call for a habit app. Downloads would have been the lazy answer.
- The T-minus timeline, the L-10 App Store submission deadline, and the health-claims caution are things a first-time PMM would otherwise learn the hard way.
- It admits “small team stretched thin” as a risk and says to cut to the essentials.
What didn’t
- It tells you to cut to the essentials and then doesn’t. There are 31 checkboxes across five phases, 8 channels, analytics event definitions, a press kit, a 30 to 50 name creator list. And the roles table hands most of it to “PMM (you)”. Round 1 warned this person about trying to do everything. Round 5 builds them the plan for doing literally everything.
- It never picks. The risks table says cut to four things, but the plan never says which is the one thing that matters most if you only get one.
- It builds on the guess from Round 3. The positioning statement I never validated is now Section 5 of a launch plan headed to the boss. Draft on top of draft… and by the time it reaches the CEO, it looks like a decision.
- The [CONFIRM] questions are right, but the plan gives a new hire no idea how to walk into the boss’s office and get them answered. Knowing what to ask and having the standing to ask are different skills.
TL;DR Twelve sections, honest flags, and a new hire could easily send it to the boss on Tuesday. But that’s the risk. It’s highly confident without underlying context.
The Scorecard
The assistant did not fail this test, and it was better than I expected in every round. The risk sits in what the new hire does with the outputs.
| Round | Usable on Monday? | Asked the new hire anything? | Flagged its own gaps? | Specific to this company? | Biggest risk |
|---|---|---|---|---|---|
| 1. Where do I start | Yes, as orientation | Only questions to take to the manager | No | No | Sounds sure about a company it knows nothing about |
| 2. 30/60/90 plan | Yes, as a draft to discuss | No | A little (the risks list) | No | A 24-checkbox plan for a company with no sales team |
| 3. Positioning | As a first draft | No | Yes, in the last third | Yes, from the homepage | Copying the statement before talking to a single rider |
| 4. Competitors | As a research starting point | No | Yes (“verify before using”) | Yes | Sources with a stake in the answer and unverified numbers landing in a sales deck |
| 5. Launch plan | As a draft for the boss | Yes (7 assumptions, 5 decisions) | Yes | Yes | An overloaded plan resting on the Round 3 guess |
Three patterns jumped out.
- Every output pointed toward real people. None started there. Talk to customers, interview riders, validate with beta users… the advice was right every single time, and it sat in the footnotes of a document that looked finished. A seasoned PMM treats the footnote as the assignment. A new hire treats the beautiful document as the assignment.
- The assistant got more honest as it got more context. Rounds 1 and 2 had nothing to be honest about, so they were confident about everything. Once it had a real homepage to read (rounds 3 to 5), it flagged gaps, labeled assumptions, and left targets blank. Give it context and you get calibration. Starve it and you get a confident textbook.
- Errors compound. The positioning statement nobody validated became Section 5 of the launch plan. Each draft inherits the guesses of the one before it, and by the time it reaches the CEO it looks like a decision.
TL;DR We got a finished-looking document with the real work buried in the footnotes, which is a reliable way to attain a feeling of confidence and “done”. The confident format is the trap that concerns me.
Cool, You Started… Now Make It Real
So our new hire now has five documents. But none of it is in production, it’s just a bunch of drafts.
Production means other people act on it. The sales rep says the sentence out loud. The homepage matches the App Store listing. The CEO picks a date. A number moves. The distance between a good-looking folder of drafts and that moment is the job.
Here’s what each draft needs before it counts:
| Artifact | What it is today | What makes it production | Who signs off |
|---|---|---|---|
| 30/60/90 plan | Generic, written for a company with a sales team | Rewritten after a real week-one conversation with the manager, with this company’s priorities and three agreed metrics | Your manager |
| Positioning | The homepage, restated | Built from 5 to 10 rider conversations, one chosen segment, alternatives named in the riders’ own words | Founder, plus anyone who builds or sells |
| Competitive view | Desk research from sources with a stake in the answer | You’ve used the top five apps, checked each claim against a primary source, dated every number, and put an owner and a refresh date on it | You |
| Launch plan | 31 checkboxes, 8 channels, built on an unvalidated guess | One goal, one metric, the three channels you can run, and a name and a date next to every item | CEO |
And here’s the conversion path, in order:
- Close the [CONFIRM] loop. Take every assumption the assistant flagged (launch date, pricing, whether the app ships with starter routines) and don’t move until each one has a name and a date next to it. If the answer is “we haven’t decided,” that’s the finding. It’s now a decision the CEO owes you.
- Go talk to people. The assistant told you to in every single round, and it was right every time. Use it to prep the interview guide and the outreach note, and to synthesize afterward, but you’re the one in the room. That’s the 20% in Humans + AI, and it’s the part that can’t be delegated.
- Zero-Trust the research. Open every source. Check pricing and ratings against the primary page. Ask who benefits from each source, and put a date and your initials next to every number you plan to repeat.
- Cut. Draft scope will always outrun your capacity, because the assistant has no idea how many hours you have. Tell it (“I have 10 hours a week and no budget”) and ask again. Then pick the one metric (the plan already named it: 3+ sessions in the first 14 days), pick the channels you can run, and write the “not doing” list where the CEO can see it.
- Build one source of truth. In Round 5 the positioning statement got copied into the launch plan, which means the day it changes there are two versions. Create one MSD (messaging source document, the single page every deck, page, and email pulls from) and point everything else at it.
- Put it on a calendar. Dates, owners, a weekly check-in. A plan that isn’t on a calendar is a wish.
- Set a baseline and a refresh. Capture today’s numbers (waitlist size, traffic) so you can show change later, review the competitive view every quarter, and put the post-launch retro on the calendar before launch day.
None of this makes the assistant less useful. It’s where it earns its keep: summarizing interview transcripts, drafting message variants, checking that the homepage, the App Store listing, and the launch email all say the same thing, and watching competitor reviews. That’s the execution tax going down, and it’s real. But “does this ring true to a rider?” is still a person’s call.
The assistant got you to a draft. Getting to production is where you start.
Same Assistant, Different Order
A seasoned PMM doesn’t write cleverer prompts. They give the assistant the thing it was missing (context) and use it to collect evidence before asking for answers. So what if we ran this same play, but this time we bring the context.
Note: I haven’t run this half yet, so treat the right-hand column as what I’d type, not a result.
| Round | What the new hire typed | What a seasoned PMM types first | What changes |
|---|---|---|---|
| 1 | “Where do I start?” | “I’m the first PMM at a pre-launch consumer app for cyclists. No customers, no sales team. Before you advise me, ask me 10 questions about the company and its stage.” | Advice fits the company instead of a generic B2B textbook |
| 2 | “Make me a 30/60/90 plan.” | “There are no sales calls to learn from here. Who should I talk to in the first 30 days, and in what order?” | The plan fits the company that exists |
| 3 | “Write me a positioning statement.” | “Here are my notes from eight rider conversations. What patterns do you see in the riders’ own words, and where do they contradict each other?” | The statement is built from evidence, not from the homepage |
| 4 | “Who are our competitors?” | “Here are the 1 to 3 star reviews for the five apps riders mentioned. What do riders complain about, in their words? Which of these sources has a stake in the answer?” | Customer language, with source bias flagged |
| 5 | “Write me a launch plan.” | “I have 10 hours a week and no budget. Given this positioning and one goal (3+ sessions in 14 days), what’s the smallest launch that tests it?” | A plan sized to the person who has to run it |
Notice the pattern. Every seasoned prompt comes after a conversation with a human, and every one hands the assistant something real to chew on. The prompts aren’t necessarily smarter. They improve the outcomes.
The Footnote Is the Job
The most useful sentence in all five outputs wasn’t in a heading. It was a bullet near the bottom of almost every document: talk to 5 to 10 riders. A new PMM who reads that as an afterthought ships the draft. A new PMM who reads it as the assignment has a real shot.
So, could an early-career hire with no PMM heritage figure this out? They can get moving faster than ever for sure. The assistant will hand them a plan, a positioning statement, a competitor map, and a launch calendar before lunch. What it can’t hand them is the instinct to distrust all of it until more context and confirmation is gathered.
That’s the Humans + AI split in one example. The assistant took a big chunk of the execution tax off the table. The 20% that decides whether any of it matters, the conversations, the judgment, the “does this ring true?”, is still a person’s job. The bottleneck isn’t the prompt. It’s the context.
If you were our new hire, which footnote would you have read first?
Adam