This question almost always comes up at the same moment: when someone realizes the first piece is going to work and starts wondering what the second piece will cost. That's a sensible question to ask before you sign, not after.

There are two ways to answer it. One is a price list with components you can buy on top. The other is a way of calculating that decides, each time, what the work is worth. We do the second, and below is why, plus what that means for you in terms of unpredictability.

01 / the methodHow is the price of an expansion determined?

Per bottleneck, not per feature. We look at which part of your work costs the most time or money right now, work out how much effort it takes to fix that, and place it in a class: small, medium, or large. That class sets the price, and it's fixed before anything gets built.

The difference from a price list is bigger than it looks. A price list says what a component costs. A scope says what it costs to solve your problem. The same component is an afternoon's work at one company and two weeks at another, because the mess underneath differs.

02 / the lineWhen does something fit within the build, and when does it cost extra?

Small custom tweaks fit within the price you already paid. An extra field, a reworded label, a report in a different format: that's fine-tuning and it comes with the territory. Substantial work is extra, and we say so before we start on it, not afterwards on the invoice.

The honest caveat: the line between those two can't be defined exactly, and we haven't managed to capture it in a single rule either. What we do is name it the moment it comes into view. If someone gives you a precise line without knowing your situation, that line is made up.

A price you only hear after the work is done isn't a price. It's an invoice.

03 / the honestySo where does your revenue actually come from?

From the second through the eighth expansion. The first build is rarely the most lucrative piece of work; it mainly exists to prove that it works. The money is in the bottlenecks that come after, once you know the system and can see where the next one sits yourself.

That's not a confession, it's an explanation of how our interests line up. We have an interest in the first piece working so well that you want a second one. You have an interest in only taking a second one once the first has paid for itself. Those two don't clash, but you should know they're both there.

What it doesn't mean: that an expansion is needed. If the first bottleneck is gone and the rest is running fine, then the rest is running fine. Anyone who tells you your system needs to be "finished" is selling volume.

04 / the alternativeWhy not packages with an upgrade path?

Because a package has you paying for what's in it, not for what you need. With a ladder from bronze to gold, at some point you're buying six components just to reach the seventh. That sells nicely and rarely buys economically.

There's also a version where you buy loose components one at a time, each with its own price tag. That's more transparent, but it shifts the question: you're then choosing from a catalogue instead of from your own workweek. The risk is that you collect components that each work on their own and together solve nothing.

Our assumption is that you're better off with a scope determined each time, even though it's harder to compare. That's a choice with a downside, and the downside is this: you can't line up our prices side by side like a menu. What you get in return is that you never pay for parts you don't use.

05 / what to askWhat should you ask any provider about this?

Four questions you can ask in any conversation, including with us. They cost nothing and they make the difference between a price and an expectation.

• What does an expansion cost, and do I hear that before or after the work? Only "beforehand" counts as an answer.
• What fits within what I've already paid? If no example of that can be given, chances are nothing fits within it.
• What exactly is the monthly fee for? Management and rent are two different things; you only notice the difference once you want to leave.
• What happens if the build runs over? Does the provider carry that risk, or do you?

That last one is the sharpest and is almost never asked. A fixed price with an open end isn't a fixed price.

More questions of this kind, about ownership, customer data, and what disappoints in the first few weeks, are covered in the questions business owners should ask before buying an AI system.

This text was produced with AI assistance and checked and approved by a human before publication.