The AI Learning Hub Journal

What a Thousand Users Cost

The steep jump is getting real, not getting bigthe cost of getting real is not the cost of getting big — after the jump, cost grows more slowly than usageTHE BILL, IN THREE HONEST LINES — PER MONTH£0–15prototypefree tiers and pocket change£40–60~200 customersa real product£120–200~1,000 usersillustrativethe proportionally steep jump:from prototype to productthis is where you start paying for theproperties Module 3 taught you to wantWHAT THE MONEY ACTUALLY BUYSthe surprise in the bill: mostly, not computersthe computers serving your pagesbackups you have actually restoredemail that reaches inboxes, not spammonitoring that messages you when it dies— the cheap partyou are paying for the properties, not the pixelsTHE FREE-TIER MOMENT THAT CATCHES EVERYONEthe free version — built for prototypesgenerous while nothing depends on itthe day you start depending on itreal bookings, real reliancethe free tier ends — exactly thennot a scam, and not bad luckit is the pricing telling you the truth a little before you were ready to hear it: your product got real, and real products pay for what they rely onCHEAP TO BUILD, EXPENSIVE TO RUN — THE SECOND NUMBER IS THE ONE THAT MATTERSthe clearest example: a feature that calls an AI service each time it is used — a friendly reminder written for every bookingpennies per use × every booking, forever = a line on the bill that grows exactly as fast as your successbuild such things when they earn their keep, not because they were easy to add — this is Module 1’s not-list wearing its other hatbefore you add anything, ask what it would cost at a thousand users — then decideFOR A BUSINESS TAKING ITS BOOKINGS ALL WEEK, £40–60 A MONTH IS NOT A TECHNOLOGY BILLit is a phone bill — and every feature you decline protects it
Honest illustrative figures for the worked example, not promises about your product.

The Bill, in Three Honest Lines

Let us put real numbers on it, with the usual caution: these are honest illustrative estimates for the worked example, not promises about your product. As a prototype, the booking page cost roughly £0–15 a month — free tiers and pocket change. As a real product with around 200 customers, it costs roughly £40–60 a month. If it grew to around 1,000 users, illustratively £120–200 a month. Notice the shape of that line. The cost of getting real is not the cost of getting big: the jump from prototype to product is proportionally the steep one, because that is where you start paying for the properties Module 3 taught you to want. After that, cost grows more slowly than usage does. For a business taking its bookings all week, £40–60 a month is not a technology bill. It is a phone bill.

  • Prototype: roughly £0–15 a month, mostly free tiers and pocket change
  • Real product with around 200 customers: roughly £40–60 a month
  • At around 1,000 users: illustratively £120–200 a month
  • The steep jump is getting real, not getting big — then cost grows slowly

What the Money Actually Buys

Here is the surprise in the bill: mostly, you are not paying for computers. The machines that serve your pages are the cheap part. The money goes on the database that keeps real backups you have actually restored, the email service whose messages reach inboxes instead of spam folders, and the monitoring that messages you when the page dies — the properties, in other words, not the pixels. That explains the free-tier moment that catches everyone. The free versions of these services were built for prototypes, and they end exactly when you start depending on them. This is not a scam, and it is not bad luck. It is the pricing telling you the truth a little before you were ready to hear it: your product got real, and real products pay for the things they rely on.

  • Computers are the cheap part; you are barely paying for them
  • The bill is backups, email that arrives, and monitoring — the properties
  • Free tiers were built for prototypes and end when dependence begins
  • The end of a free tier is not a scam; it means the product got real

Cost Is a Design Input

Some features are cheap to build and expensive to run, and the second number is the one that matters, because you pay it every month for as long as the feature exists. The clearest example is anything with AI inside it. A feature that calls an AI service each time it is used — say, writing a friendly reminder message for every booking — costs pennies per use, and pennies multiplied by every booking, forever, is a real line on the bill that grows exactly as fast as your success does. Build such things when they earn their keep, not because they were easy to add. This is Module 1's not-list wearing its other hat: every feature you decline protects the bill as well as the ledger. Before you add anything, ask what it would cost at a thousand users — then decide.

  • Cheap to build but expensive to run is a common, dangerous combination
  • Per-use AI features grow the bill exactly as fast as your success
  • Build features when they earn their keep, not because they were easy
  • The not-list protects the bill too — ask the cost at 1,000 users first

Prefer slides, quizzes, and saved progress? Read this lesson in the library — free, no sign-up.