When AI changes what you sell, not just how you work
Two different things get called digital transformation. One makes your existing work cheaper; the other changes what you can charge for it. One question tells them apart, and most businesses are sold the wrong one.
Two completely different things get called digital transformation, and confusing them is expensive in both directions. One is making the work you already do take less effort. The other is the work you do changing shape, because the economics underneath it moved. Both are worth doing. Only one of them is a strategy question, and most businesses are sold the second when they need the first.
The distinction is not about how much technology is involved. It is about whether your price list has to change at the end of it.
The efficiency change
This is the common one, and there is nothing wrong with it. Quotes stop being rebuilt from scratch. Supplier invoices stop being typed in twice. The monthly report assembles itself instead of eating a day. Your team does the same work for customers who notice no difference at all, and you get hours back.
The test is simple: when it is finished, you sell the same thing, to the same people, at the same price, with fewer hours behind it. That is a good outcome. It is also not a transformation, whatever the proposal called it, and treating it as one tends to make the project bigger and slower than the saving justifies.
The model change
Occasionally something different happens. The cost of doing one specific thing falls far enough that work you used to turn down becomes worth taking, or a service you charged for by the hour stops making sense priced that way.
A business that quotes on complex jobs, and declines the small ones because the estimating effort is the same either way, is a good example. If producing a defensible quote genuinely gets cheap, the small jobs stop being unprofitable. That is not a faster quoting process. That is a different addressable market, and it probably needs different pricing, different marketing and possibly different staff.
Or the reverse: an advisory firm that bills for the hours junior staff spend reviewing documents has a problem if that review stops taking hours. The efficiency is real and it points straight at the revenue line. Pretending otherwise does not make it go away; it just means a competitor gets to raise the question with your clients instead of you.
If it works and your price list does not change, it was efficiency. If it works and your price list has to change, it was the other thing.
Telling them apart
| Signal | Efficiency change | Model change |
|---|---|---|
| What the customer notices | Nothing, or faster service | A different offer, or a price that moved |
| Who needs to approve it | An operations manager | Whoever sets pricing |
| What it threatens | Some internal admin time | A revenue line you currently rely on |
| If a competitor did it first | They are a bit more efficient | They can undercut you and still make money |
| Where it fails | Adoption: people go back to the old way | Strategy: the economics were never going to move |
The competitor test
The most useful hour you can spend on this is uncomfortable. Sit down and design the business that would take your customers if it started today, with your market knowledge and none of your constraints: no legacy system, no staff whose roles assume the current process, no revenue line it would hurt to lose.
If what you design looks much like what you already have, you are in efficiency territory and you can relax. If it looks materially different, somebody will eventually build it, and you would rather be the one who did.
Note that this test says nothing about AI specifically. That is deliberate. The question is whether the economics of serving your customers have moved. AI is currently the most common reason they do, but the question is older than the technology and will outlast it.
Why most businesses should not start here
Having said all that: for the majority of small and medium businesses, the honest finding is that the model is fine and the operations are a mess. The quoting is slow, the same detail is entered into three systems, nobody can see the state of every job at once. Fixing that is worth real money and does not require anyone to rethink what the company is for.
Starting with a business-model review when what you need is three automations is a way of spending a lot of money to be told to do the obvious thing. We would rather tell you that at the first conversation than at the end of an engagement.
The exception is when something in your market has already moved. A competitor pricing in a way you cannot explain, customers asking for something your model cannot deliver, or a service line where the effort behind it is visibly collapsing. Those are worth taking seriously straight away.
What to do with the answer
If it is efficiency, treat it as such. Pick the processes with real volume behind them, fix the ones that are inconsistent before automating anything, and be suspicious of any plan that requires rethinking the business to deliver a saving in accounts payable.
If it is a model change, the first move is almost never technology. It is working out what you would charge, who you would charge it to, and what you would stop doing. The build comes after that, and it is usually smaller than people expect, because by then you know precisely what it has to do.
The short version
Ask one question of any AI proposal in front of you: if this works perfectly, does our price list change? If the answer is no, you are buying efficiency, and you should scope it, cost it and judge it as efficiency. If the answer is yes, stop treating it as an IT project, because it is not one.
Not sure which one you are looking at?
Describe what has changed in your market and we will give you a straight read on whether this is a business-model question or three processes that need fixing. The second answer is more common and we will say so.
How AI disrupts a business model: four patterns
Business model disruption is not a competitor being cheaper. It is the expensive part of your work ceasing to be expensive. Four recognisable patterns, what tends to survive, and three questions that tell you where you stand.
Before you automate anything
Four questions worth answering about a process before you spend a dollar automating it. Most of the automation projects we are asked to rescue skipped at least two of them.