The WattPlan messaging house, how we built it and what it does

WattPlan is Manual Focus’s own cycling training app, and since 11 September 2026 every line of its copy has traced back to one document, a messaging house. One two-word tagline, Train. Adapt., sits over one product claim, three benefits and a proof row for every feature that names the code behind it. This is how we built it.

Key takeaways

The WattPlan messaging house has six levels. A two-word tagline, an internal emotional benefit, a 61-character short description, one core product claim, three benefits and the features that prove each one.
Every supporting feature in the house carries a reference to the code that ships it, so when the team removes a capability the claim retires the same day.
The house corrected five stale claims before they reached a store listing, including a workout library of 67 where older documents said more than 100, and a calendar export that had never been built.
Superlatives stay out of WattPlan copy because the ASA treats an unqualified superlative as a comparison with every competing product.

Why build a messaging house at all?

WattPlan builds a cyclist’s training plan around their goals and the hours they have, then adjusts it when a ride is missed or goes harder than planned. It has a Coach the rider can talk to, and it syncs and scores rides from Garmin, Strava and Wahoo. That is a lot of product to describe across two store listings, a website, an app and emails, and left alone each surface drifts towards its own version of the story.

A messaging house stops the drift. It is the communication hierarchy from our basics series applied to one product, with one rule added. If a line cannot trace to a benefit in the document, it does not ship. WattPlan is our own app, so this is a worked example from the inside, mistakes included.

What the house says

The house starts with the feeling WattPlan should create and gives the cyclist more detail as they move towards action. The top is emotional and memorable, the bottom is proof, and copy leads with a feature only when the cyclist already understands the proposition. Each level has a job.

The WattPlan messaging house · revised 11 September 2026Six levels, each proving the one above

1
Taglinethe line people rememberTrain. Adapt.
2
Emotional benefitthe feeling it should leave, kept internalConfidence that your training is right
3
Short descriptionwhat it is and what it does for youA cycling training app built to make you stronger on the bike
4
Core product claimthe sentence every benefit supportsWattPlan builds your training around your goals and the time you have, adjusting it as things change.
5
Benefitswhat changes for the cyclist, each a group of features
- Personalised. TrainingThe week. When life or riding changes, WattPlan keeps you on track.
- Questions. Answered.The Coach. Ask a normal question and get an answer based on your plan and rides.
- Rides. Counted.The data. Connect once, then ride. WattPlan brings in the file and scores it.
6
Featureshow it delivers each benefit, with the code as proof
- For the week
  - Training plans built around your days and hoursplan_generator.js · week_coherence.js
  - Adjusted for life after a missed workoutrepairWeek() · rebalanceWeekAfterMiss()
  - Adjusted for the rides you didovershoot, undershoot and fatigue rules
  - Strength placed around the ridingstrength_sessions.js
- For the Coach
  - Talk to a coach that knows your plan, rides and eventai.js assembles the context
  - Change the plan by askingCoach action tags · updatePlanDay()
  - Long-term memory of goals, injuries and schedulecoach_memories
  - Proactive coaching after a ride or personal bestwebhook_shared.js · coaching_patterns.js
- For the data
  - Automatic ride sync from Garmin, Strava and Wahoogarmin.js · strava.js · wahoo.js
  - Weighted Power, Intensity and Load for every rideshared_calculations.js
  - Fitness Explorer trendspmcHistoryDays · pmcForecastDays
  - Personal bests and FTP promptspb_shared.js · ftp_prompt.js

Each row proves the one above it and widens as it goes, from two words to the code that ships them. If a file moves, the reference is fixed. If a capability goes, the claim goes the same day.

Level	What it is	WattPlan’s
Tagline	The line people remember	Train. Adapt.
Emotional benefit	The feeling the product should leave, kept internal	Confidence that your training is right
Short description	What it is and what it does for the buyer	A cycling training app built to make you stronger on the bike
Core product claim	The one sentence every benefit has to support	Builds your training around your goals and the time you have, adjusting it as things change
Benefits	What changes for the cyclist, each a group of features	The week, the Coach, the data
Features	What the product does to deliver each benefit, with proof	Week repair, Coach memory, automatic ride sync

Benefit 1 leads because personalising the week is WattPlan’s point of difference. The emotional benefit never appears in customer copy as written, because it would read as a performance promise.

How two words carry four readings

Cyclists train so their bodies adapt, and WattPlan uses the same word for its core mechanism. The adaptation engine carried the word adapt in the code before the brand did, so the tagline describes the product rather than decorating it.

The taglineFour readings of Train. Adapt.

TrainDo the work. Ride.The cyclist
Adapt, the bodyYour body adapts through training.The sport
Adapt, the planThe plan adapts to what you actually did.The product
Train and adaptThe loop that makes you faster.Both

The second and third readings do the work. The stops are a typographic treatment for lockups and end frames. In body copy it is train and adapt.

Two earlier lines were retired. “Your Cycling Coach That Never Switches Off” went because uptime does not sell a training app. “Turn cycling commitment into real performance improvement” went because WattPlan cannot guarantee performance, and the CAP Code requires documentary evidence for any claim a consumer would read as objective.

How was it built from the code?

Traceability is what separates the house from a brand deck. We wrote the proof column first, against the shipped product, so every feature names the function or file that delivers it. Pricing was checked against the subscription configuration on 18 August 2026 and the store field limits against Apple and Google’s guidance on 11 September 2026. Writing that way turned up five claims that older internal documents still made and the product no longer supported.

Stale claim	What the code showed	What the house says instead
More than 100 workouts	67 workouts across seven zones	”A structured library across every training zone”, because the count will change
Calendar export	No iCal or .ics writer exists	Roadmap work, kept out of store and pricing copy
Zwift export is Premium	The export ships free	Kept out of the Premium list
Ride Review as a feature	Shipped on 2 September 2026 as a Coach review	”A Coach review after every ride”, because the benefit is the review
Garmin and Wahoo push as Premium	Both ship free	Claim the capability without a tier

Each would have gone into a store listing. Google Play’s Deceptive Behavior policy requires an accurate description of an app’s functionality in all parts of the metadata, so an overclaiming listing risks rejection, then refunds.

Where does each message go?

The house maps a path rather than a script, and no asset has to walk all of it.

The message pathFrom attention to action

AAttentionStop the right cyclist and establish the idea.Train. Adapt.Next move, read the short description.
IInterestName the product and the rider outcome.A cycling training app built to make you stronger on the bikeNext move, see how WattPlan delivers it.
DDesireTurn the proposition into outcomes the cyclist wants.Personalised. Training · Questions. Answered. · Rides. Counted.Next move, explore the features and proof.
AActionMake the first step clear and low-risk.Start free. Upgrade when you're ready.Create a plan for free and add Premium for the complete plan.

No asset has to carry all four. A store listing can. A banner often carries the first and last only.

A routing table then names which benefit leads on each surface. The home hero carries the tagline and short description with Benefit 1 beneath. Onboarding leads with Benefit 3, because watching the plan fill itself in from a connected Garmin is the fastest proof the product offers. Adaptation notifications are Benefit 1, factual. Personal-best notifications are Benefit 2, warm and brief.

Behind the routing sit two feelings every surface should leave. Understood, because the plan takes the rider’s real hours and has read their ride. Consistent, because one missed workout does not decide the week.

The internal benefitUnderstood and consistent, at every stage

StageUnderstoodConsistent

Setting upUnderstoodIt takes your real hours, per day, not a fantasyConsistentThe week it builds is one you can actually ride

Mid-weekUnderstoodYour ride is already in, already scoredConsistentMiss one and the rest of the week still makes sense

AfterUnderstoodIt saw the personal best before you didConsistentThe block completes instead of collapsing

These are product tests, not headlines. A feature that strengthens neither feeling is a feature looking for a reason.

What the house refuses to say

The foundationThe rules every level obeys

Tone of voiceAn experienced club rider who knows the science. Warm, direct, one clear recommendation.
Product substantiationShipped claims with their code references, and a corrections log for the stale ones.
Claims WattPlan does not makeMedical, guaranteed gains, a replacement for a human coach, safety, invented popularity.

Everything above the foundation is copy. The foundation is the set of rules the copy is checked against before it ships.

The longest section of the document is about silence. WattPlan makes no medical claims, promises no watts, FTP points, times or placings, and never presents the Coach as a replacement for a human coach. Its own instructions call it an advisor, not an autopilot, so the copy cannot say “fire your coach”. The house also bans invented popularity, so no “trusted by thousands”, and bans advertising a free trial, because there is a permanent free tier instead.

The banned vocabulary runs to three lists. Sport clichés such as epic, pain cave and beast mode. Technology filler such as seamless, disruptive and next-generation. And AI filler, where AI-powered is out because every rival can say it and it tells a rider nothing. A vocabulary table from the product’s own screens sits alongside.

Say	Not	Because
Plan	Programme, schedule, block	The screen is called Plan and the product has no “programme” in it
Coach	Assistant, bot, AI	The tab is Coach and the placeholder reads Ask your coach
Cyclist, rider	Athlete, user, triathlete	WattPlan is cycling only

Our view at Manual Focus is that a messaging house earns its keep only when every claim can be traced to something the product does today, because a brand document that cannot be checked against the product drifts from it within a release. The brand guardrails as code playbook turns a document like this into a check that runs before copy ships.

How to build one for your product

Write the proof column first. List what the product does today, with the feature, file or setting that delivers it, and let the benefits come from the list. Anything the list cannot support goes in a corrections log.
Retire lines in writing. Name each dropped tagline and the reason, so nobody revives it from an old deck. The tagline system playbook covers generating and testing the replacement.
Put the channel limits in the document. Apple’s product page guidance and Google Play’s listing help publish every field’s limit. Record how much of each your copy uses.
Route the benefits. One table naming which benefit leads on each surface ends most arguments about what a page says first.
Write the silence. The claims you will not make, with the regulation or product limit behind each, protect you from the enthusiastic release-day draft.

What the house hasn’t settled

The document ends with open decisions rather than a flourish. Garmin and Wahoo workout push ship free, so the product gate must change before any Premium claim about them. iOS push notifications stay out of App Store copy until a build passes delivery tests on a physical iPhone. Some product strings still mix British and American spelling. A house that admits what it hasn’t decided is easier to trust than one that reads as finished.

Frequently asked questions

What is a messaging house?

A messaging house is a one-page hierarchy for everything a brand says. One memorable line sits at the top, a plain description and a core claim sit beneath it, and three or four benefits hold the house up, each proved by named features. Every piece of copy, from a store listing to a push notification, leads with the top and goes as far down as it has room for.

What does WattPlan's tagline Train. Adapt. mean?

Train. Adapt. carries four readings in two words. Train means do the work and ride. Adapt means the body adapts through training, which is the sport, and it means the plan adapts to what the rider did, which is the product. Read together, training and adapting is the loop that makes a cyclist faster. The product used the word adapt in its code before the brand did.

Why does WattPlan avoid saying AI-powered?

WattPlan's messaging house bans AI-powered, powered by AI and our advanced algorithms as filler, because every rival can say them and none of them tells a rider what changes. The house also keeps the Coach from disclosing its model or vendor. Copy names what WattPlan does instead, such as repairing the week after a missed workout, and uses AI Cycling Coach only in the store app name to establish the category.

How long can App Store and Google Play copy be?

Apple allows an app name of up to 30 characters, a subtitle of 30, promotional text of 170 and keywords of 100 characters, with up to 10 screenshots. Google Play allows an app name of 30 characters, a short description of 80 and a full description of 4,000, with up to eight screenshots per device type. WattPlan's house records the approved copy for each field and how much of each limit it uses.

Does WattPlan promise to make you faster?

No. WattPlan's short description says it is built to make you stronger on the bike, which states the purpose of the product without a guaranteed result or timeframe. The messaging house bans promises of watts, FTP points, times or placings, bans medical and safety claims, and bans comparing the Coach with a human coach, because none of them can be substantiated for every rider.

Sources

WattPlan. Train. Adapt., WattPlan by Manual Focus
WattPlan pricing, WattPlan by Manual Focus
Creating your product page, Apple Developer
Add preview assets to showcase your app, Google Play Console Help
Create and set up your app, Google Play Console Help
Deceptive Behavior, Google Play Developer Policy Center
The CAP Code, section 3, Misleading advertising, Advertising Standards Authority

From the podcast

James and Hannah go inside the messaging house for WattPlan, the cycling training app Manual Focus builds. It has six levels, from the two-word tagline Train. Adapt. down to the features, each naming the code behind it, and writing it that way caught five stale claims before they reached a store listing. They walk each level with its definition, the four readings of the tagline, the message path and routing table, the understood and consistent test, the three banned lists, the foundation, the App Store and Google Play limits, and three jobs to hand to AI if you are building a house for your own product.

Get AI on this

Paste these into whichever AI assistant you use: ChatGPT, Claude, Gemini or Copilot. Swap the bracketed parts for your own details. The first two need file uploads. The third needs an assistant that can run on a schedule, or a calendar reminder to run it on release day.

1. Write the proof column

I run marketing for [product], [one line on what it does] at [website],
selling mainly in [market]. I'm building a messaging house, a one-page
hierarchy where every customer-facing claim traces to a feature the product
ships today.

Attached are:
- our feature documentation, or the codebase (attach the repository, or
  export the feature list and settings)
- every piece of live customer copy we have (store listings, website pages,
  onboarding screens, emails, release notes), as text files

Go through the copy and list every claim it makes about what the product
does. For each claim, find the feature, file, function or setting that
delivers it.

Give me a table with the columns: Claim | Where it appears | Feature or
code that delivers it | Status (shipped / partly shipped / not found).

Then list the orphans separately: every claim with nothing behind it, or
with a number the product no longer matches. For each orphan, suggest the
plainest truthful replacement. Don't guess at a feature you can't find in
the files; mark it "not found" and I'll check.

2. Audit live copy against the house

Attached is our messaging house for [product]: the tagline, short
description, core claim, benefits, the vocabulary table (say this, not
that), the banned words and the claims we never make.

Also attached is our live customer copy: [store listing, website pages,
onboarding, emails, release notes].

Check every line of the copy against the house. Flag:
1. Any claim that doesn't trace to a benefit in the house.
2. Any banned word or phrase, including superlatives such as "the only",
   "most advanced" or "best", and any performance guarantee.
3. Any term the vocabulary table says to avoid, with the word to use
   instead.
4. Any surface that leads with a benefit other than the one the routing
   table assigns it.

Give me a table with the columns: Surface | Line as it stands | Rule it
breaks | Suggested rewrite. Keep each rewrite in the house's voice, with no
colons, dashes or exclamation marks. Where you're unsure whether a line
breaks a rule, say so rather than deciding for me.

3. Keep the house current on release day

Run this every time we ship a release of [product], or set it as a routine
on release day.

Attached are our messaging house for [product] and the release notes or
changelog for the new version.

Compare the two and tell me:
1. Which claims in the house the release strengthens, and the new proof
   (the feature, file or setting) to record under them.
2. Which claims the release weakens or removes, so the claim should be
   retired or reworded the same day.
3. Any new capability in the release that the house doesn't yet mention,
   with a draft line for it that follows the house's rules (no superlatives,
   no guarantees, product vocabulary only).
4. Any free or paid tier change that affects the pricing section.

Give me the answer as a short list of edits to make to the house, each with
the section it belongs in. If the changelog doesn't say enough to decide,
say what you'd need to check rather than guessing.

And one for you

Write the retired lines down with the reason each one went. The house records both of WattPlan’s, and it is the cheapest insurance against someone reviving a dropped tagline from an old deck.

Read the episode transcript

James: We don't usually turn the microphone on ourselves, but this week we are, because WattPlan, the cycling training app we build at Manual Focus, now runs every line of its copy from one document. I'm James Vickers, this is Manual Focus, and Hannah's here to keep me honest about our own product.

Hannah: Hi James. And up front, this is our app, so it's a worked example from the inside, mistakes included.

James: So, the document. A messaging house, revised on the eleventh of September. Walk us down it.

Hannah: Six levels, each one proving the one above. Level one is the tagline, the line people remember. Ours is two words, Train. Adapt. Level two is the emotional benefit, the feeling the product should leave, and it stays internal. For WattPlan, confidence that your training is right.

James: Internal because written out, it reads like a performance promise.

Hannah: Level three is the short description, what it is and what it does for you. "A cycling training app built to make you stronger on the bike", sixty-one of the eighty characters Google Play allows. Level four is the core product claim, the one sentence every benefit has to support. WattPlan builds your training around your goals and the time you have, and adjusts it as things change.

James: And then the benefits.

Hannah: Level five. What changes for the cyclist, each a group of features. The week, the Coach and the data. And level six is the features, each naming the code that delivers it, the function or the file. If the file moves, someone fixes the reference. If the capability goes, the claim goes the same day.

James: It's the communication hierarchy from our Basics episode on what marketing is, applied to a real product, with one rule added. If a line can't trace to a benefit in the document, it doesn't ship.

Hannah: And we wrote the proof column first, against the shipped product, which turned up five things older documents still said and the product no longer did.

James: Embarrass us.

Hannah: Older documents said more than a hundred workouts. The code has sixty-seven, so the house says "a structured library across every training zone". Calendar export was in a feature table, and there's no calendar writer in the code. Zwift export was listed as Premium, and it's free. So were Garmin and Wahoo workout push. And Ride Review had shipped on the second of September, so the house names the benefit, a Coach review after every ride.

James: Every one of those was a sentence away from a store listing.

Hannah: Which is the regulatory point. Google Play's Deceptive Behaviour policy requires an accurate description of what an app does in every part of the listing. And the CAP Code treats an unqualified superlative as a comparison against every competitor. So no "the only".

James: Back to the tagline. Why does Train. Adapt. earn its place?

Hannah: Four readings in two words. Train is the cyclist doing the work. Adapt, the body, is the sport, because your body adapts through training. Adapt, the plan, is the product, because the plan adapts to what you actually did. And together, train and adapt is the loop that makes you faster.

James: And the product used the word first. The adaptation engine carried "adapt" in the code before anyone wrote a tagline. Two earlier lines were retired, with the reason written down. One about a coach that never switches off, because uptime doesn't sell. One promising performance improvement, because we can't guarantee it.

James: So where does it go? A notification and a store listing can't carry the same amount.

Hannah: The house maps a path, attention, interest, desire, action. Attention is the tagline. Interest is the short description. Desire is the three benefits. Action is "start free, upgrade when you're ready". No asset has to walk all of it. Then a routing table says which benefit leads on each surface. The home page gets the tagline and the week. Onboarding leads with the data, because watching the plan fill itself in from a connected Garmin is the fastest proof there is.

James: And behind the routing, the internal benefit breaks into two feelings. Understood and consistent.

Hannah: At every stage. Setting up, it takes your real hours, and builds a week you can actually ride. Mid-week, your ride is already in and scored, and if you miss one the rest still makes sense. After, it saw the personal best before you did, and the block completes instead of collapsing.

James: Then the longest section, which is about silence. No medical claims. No promised watts. The Coach is an advisor, not an autopilot, so no "fire your coach". No "trusted by thousands". And no free trial, because there's a permanent free tier.

Hannah: Plus three banned lists. Sport clichés, so no pain cave. Technology filler, so no seamless. AI filler, so no "AI-powered", because every rival can say it. And a vocabulary table from the app's own screens. Plan, not programme. Coach, not bot. Rider, not athlete.

James: And under all of that sits the foundation. A tone of voice, an experienced club rider who knows the science. Product substantiation, the shipped claims and a corrections log. And the claims WattPlan never makes. Everything above it is copy, checked against the foundation.

Hannah: The house also carries the store limits. Apple allows a name of thirty characters, a subtitle of thirty, promotional text of a hundred and seventy and keywords of a hundred. Google Play allows a name of thirty, a short description of eighty and a full description of four thousand. The house records how much of each limit the approved copy uses.

James: Our view at Manual Focus is that a messaging house earns its keep only when every claim traces to something the product does today, because a brand document you can't check against the product drifts from it within a release. The article gives five steps. Proof column first. Retire lines in writing. Put the channel limits in the document. Route the benefits. Write the silence. Most of that can go to AI, so, get AI on this.

Hannah: Go on.

James: First, get AI to write the proof column. Give it your feature documentation, or the codebase, and every piece of live copy, and have it map each claim to a shipped feature.

Hannah: Back comes a table of claims with the feature behind each, and a list of orphans, the claims with nothing underneath. That's your corrections log on day one.

James: Second, audit the copy you already have against the house. Store listing, website, emails.

Hannah: Give it the house and the copy, and it hands back each line that breaks a rule, which rule, and a rewrite. The decision stays with a person.

James: Third, make it a routine. After every release, have AI read the changelog against the house and list the claims to add and to retire.

Hannah: Set it to run on release day. Our five mistakes are the kind that build up between reviews.

James: The prompts are in the show notes. What hasn't the house settled?

Hannah: It ends with open decisions. Garmin and Wahoo push ship free, so the product gate has to change before any Premium claim. iOS push stays out of App Store copy until a build passes delivery tests on a real iPhone. And some product strings still mix British and American spelling.

James: So the takeaway. One document, six levels, every claim traced to the code that ships it. The full house, with the diagrams, is on manual-focus dot co dot uk. If you'd like the next episode to find you, follow Manual Focus in Apple Podcasts, Spotify or wherever you listen. And if your copy has drifted from your product, book a free working session at manual-focus dot co dot uk and we'll start with the proof column. Thanks, Hannah.

Hannah: Thanks, James.