Learn Marketing The horsepower rule
PositioningNobody in 1780 knew what a steam engine could do. So James Watt stopped selling steam and started selling horses. Two centuries later we still rate engines in animals that left the job long ago.
The original translation
Drag the engine bigger. Watch which sentence a mine owner can price.
What the engineer says
Our engine produces 7,457 watts of mechanical power.
Accurate. Also unpriceable by anyone in the room.
What Watt said
This engine replaces 10 horses.
Same machine. A unit the buyer feeds, stables and pays for.
“Ten horses working every hour of every day, and none of them tire.”
One horsepower was defined as 745.7 watts — a figure Watt got from watching mill horses turn a wheel all day, then rounding in the customer's favour.
Why it works
This is the whole mechanism, and it is countable. Every unfamiliar term is a small tax the reader pays before they can decide anything. Most readers stop paying.
Our engine produces 24,000 watts of mechanical power at the flywheel.
Three terms that did not exist in the buyer's working life. The sentence cannot be answered — only nodded at.
This engine does the work of ten horses, and it never sleeps.
Nothing to look up. The buyer already knows what a horse costs to keep, and can do the arithmetic before you finish the sentence.
Your job is not to explain the product. It is to find the thing the buyer already understands and stand next to it.
Finding your horse
Pick one. Arrow keys work too.
The person they would otherwise pay to do this.
The sentence
It is like adding a ______ who ______.
Filled in
It is like adding a researcher who works every night and hands you a finished list by 9am.
A salary is the one number every buyer already knows by heart. Next to it, most software is rounding.
Cheap products. Nobody weighs a $12 tool against a $108,000 salary — the gap is so wide it reads as a joke.
The anchor does the pricing
Move your price next to a salary instead of next to a competitor. Same number, completely different conversation.
$5,000 a month is 62% of one $75,000 hire.
Under two-thirds of one hire — at that ratio the comparison does the selling for you.
Costed at 1.3× base for payroll tax, benefits, software seats and equipment — the range finance teams usually plan with.
18 translations
Every left-hand sentence is true. Not one of them helps a stranger decide anything.
Showing all 18 examples.
We generate qualified pipeline through omnichannel paid acquisition.
It is a salesperson who works nights and weekends and never asks for a raise.
Now the retainer competes against a headcount req, not against four other agencies.
$5,000 per month, twelve-month minimum commitment.
Less than half of one mediocre salesperson — and this one does not quit in month seven.
Same price, different shelf. Expensive is always expensive relative to something.
Advanced speech recognition and LLM-based summarisation.
An executive assistant in every meeting who never forgets a single line.
The buyer stops evaluating the model and starts evaluating the outcome.
Searchable transcript archive with semantic retrieval.
Version history — but for every conversation your company has ever had.
One product, two good horses. Pick by which one your buyer touches more often.
SMTP handshake plus MX record validation with confidence scoring.
Spell-check for your contact list — run before your domain pays for the typos.
Spell-check is the rare reference literally everyone has used and nobody had to learn.
Waterfall enrichment across multiple providers with per-field confidence scores.
It fills the blanks in your spreadsheet, and tells you which answers to trust.
Waterfall and confidence score are real things. They are also the seller’s vocabulary, not the buyer’s.
Continuous deduplication and field normalisation across objects.
You know the Friday afternoon you lose merging duplicate accounts? You get it back.
Naming the day of the week does more work than naming the algorithm.
Distributed tracing with span propagation across microservices.
A flight recorder on every request. When it crashes you read the black box instead of guessing.
A physical object beats an abstraction, even for a technical buyer.
Automated failover with point-in-time recovery and immutable snapshots.
A database that carries a spare tyre and fits it for you at 3am.
Engineers buy on outcomes too. They just have a higher bar for the analogy being accurate.
Immutable snapshot retention with configurable RPO windows.
A time machine for the afternoon someone deletes the wrong folder.
Apple shipped a backup product literally called Time Machine. The horse was the product name.
AES-256 encrypted vault with zero-knowledge architecture.
One key for the whole building, instead of forty on a ring you keep losing.
The encryption is why it is safe to believe you. It is not why anyone wants it.
Continuous external attack-surface discovery and drift detection.
A night watchman who walks the whole building every ten minutes — not a consultant who walks it once a quarter.
The comparison is doing double duty: it names the value and buries the incumbent.
ML-based receipt categorisation and automated ledger reconciliation.
What your bookkeeper does on the third of the month — done every night while you sleep.
Anchoring to a person you already pay converts the price into a comparison, not a cost.
Employee lifecycle orchestration with role-based task routing.
The checklist your best manager keeps in their head — handed to every other manager.
Buyers can picture the good manager. Nobody can picture lifecycle orchestration.
5mg of highly bioavailable melatonin per serving.
Like brushing your teeth — nobody expects a perfect smile after one brush.
This is the anchor that makes a 90-day pack feel sensible instead of expensive.
A structured twelve-week supplementation protocol.
Couch to 5K, for your sleep.
Four words. It explains the length, the difficulty curve, and the payoff at once.
800-volt architecture with 350 kW peak charge rate.
Ten minutes plugged in gets you from Boston to New York.
Specs answer how. The journey answers whether the buyer can stop worrying.
Branching and merge-conflict resolution for binary design files.
Undo — but for the whole team, going back a year.
Undo is possibly the most universally understood interaction in software. Use it.
Nothing in that category.
Test your own line
Paste a headline. Two signals get counted: mechanism words the reader must already understand, and comparison markers that hand them something they do.
Mechanism words
0
Comparisons
0
Words
0
Your line, marked up
You have a horse.
A comparison the reader already owns, with no vocabulary they have to be taught first.
Nothing is sent anywhere — the check runs in your browser. None of the flagged words are wrong, they are simply the seller's vocabulary rather than the buyer's.
The other half of the rule
A comparison imports everything about the thing you compared yourself to — including its price and its ceiling.
You swapped one unfamiliar thing for another. The reader now has two things to look up instead of one.
"It is Terraform for your CRM." Ask a VP of Sales what Terraform is.
Test the reference on someone in the buyer’s actual job title, not someone in yours.
Every comparison imports a price as well as a picture. Anchor to something cheap and you have just published your ceiling.
"Like a spreadsheet, but nicer." Spreadsheets are free. Now you are arguing about $9.
Anchor up. Compare to the salary, the agency, or the outage — not to the free thing.
A comparison that got used a thousand times stops being a picture and becomes a category label.
"Uber for X" said nothing by 2016 except that the founder read TechCrunch.
If you can name five companies using your analogy, it is scenery. Find a fresher one.
A good analogy sticks so hard it becomes the limit of what people imagine. Watt won so completely that we still rate engines in horses two centuries after the last one left the mine.
Early cars were sold as "horseless carriages" — a frame that hid what the car would actually change.
Use the familiar anchor to get in the door. Once they are inside, show them the thing the horse cannot do.
Practising what we preach
We write spec lines by accident too. Here is the same exercise run on the things we sell.
Multi-provider verification with syntax, MX and mailbox checks inside Google Sheets.
Spell-check for the list you are about to email — before the bounces cost you the domain.
Scheduled Stripe API import of MRR and churn metrics into a spreadsheet with native charts.
The Monday morning numbers your finance person used to rebuild by hand, already in the sheet.
Pixel-based open tracking with per-recipient event history.
A read receipt, for the follow-up you were going to send blind anyway.
Rich-table composition inside the Gmail compose window.
The table you gave up on and pasted as a screenshot — except it is real text.
The mental model
Explain what the product does.
Name the thing it replaces, and let the reader do the explaining.
The more precise the wording, the more credible it sounds.
Precision that needs a glossary reads as noise. Accuracy is the floor, not the pitch.
Our price should look small next to competitors.
Choose what your price sits beside. A salary makes software look free.
Analogies are for consumer products.
Flight recorder, spare tyre, undo. Technical buyers are people at 4pm too.
Write the spec line honestly
One sentence, all the mechanism, no marketing. You need to see what you are actually asking people to decode.
Ask what the buyer already prices
A person, a tool, a chore, a habit. Pick the one they touch most often, not the one that flatters you most.
Write the comparison, then cut the spec
Not "our AI does X, like a researcher". Just the researcher. The mechanism can live three scrolls down.
Read it to someone outside the company
If they can repeat it back in their own words a minute later, you have a horse. If they ask a clarifying question, you have a spec.
The best marketers do not invent new languages. They translate new ideas into old ones.
Sources
The example copy on this page is written for illustration, not quoted from the companies named.
Marketing, finance, system design and algorithms — all explained the same way.