Your best AI customer is probably not your biggest one
Most MSPs go after the largest logo in the base. The AI conversation usually lands somewhere less obvious — and the signals that predict it are not the ones you are measuring.
Ask most MSP owners who they serve and the honest answer is a version of “anyone in the region who needs IT support”. It sounds inclusive and it feels commercially sensible. Why turn business away?
In practice it means nobody in particular hears anything specific. Marketing describes the industry rather than a customer. Sales conversations start from scratch every time. And when a genuinely good opportunity appears, nothing in the business was set up to notice it.
The cost is not the business you turn away. It is the business you never recognised in time.
Broad feels safer. Defined performs better.
Everyone who needs IT
Messaging that could belong to any MSP in the country. Sales effort spread evenly across accounts of wildly different potential. Every proposal written from a blank page because no two customers look alike.
Nothing is ever obviously not a fit, so nothing is ever obviously a priority.
A profile with edges
You can say who this is for and who it is not. Offers are built for a known situation. The team can spot a good opportunity in a routine service review, because they know what one looks like.
Effort concentrates where it compounds.
The cost is not the business you turn away. It is the business you never recognised in time.
Technical maturity is a poor predictor.
Here is where the usual targeting advice breaks down. The instinct is to go after the most technically mature customers — the ones already on modern workloads, already talking about automation, already asking questions.
They are often the worst first customers for an AI practice. They have opinions, internal capability, and frequently a preferred adviser already. What you want is not technical readiness. It is commercial pain that a change to how the work is done would relieve — and an executive who owns that pain.
A distribution business drowning in manual order processing is a better AI customer than a software firm with a modern estate and no particular problem. The first has a business case waiting to be written. The second has a technology conversation.
Five signals worth more than sector or size.
Most of these are already visible in your installed base. Nobody is looking at them, because nobody defined what they were looking for.
Process-heavy operations
Work done repeatedly by people, at volume, with rules. Order processing, claims, quotations, onboarding, case handling. Where the economics are visible before anyone builds anything.
An executive who owns the pain
A COO who knows exactly how many hours that process consumes, or a CFO who has been asked to reduce cost without reducing service. Somebody whose problem it already is.
A business event in motion
Growth outstripping headcount, an acquisition to integrate, a system reaching end of life, a regulatory deadline. Events create budget and permission that steady states do not.
You already have the relationship
Most first AI engagements should come from the installed base. You know the estate, you have credibility, and you are already in the building. New logos are a harder first proof point.
They have paid for outcomes before
A customer who has only ever bought hours and licences from you will price this as hours and licences. One who has bought a project with a business case attached has already crossed that line.
Score the base you already have.
This does not need a research project. Take your top forty accounts, put the five signals across the top, and mark each one honestly. It takes an afternoon with the leadership team and it usually produces the same surprise: the best-scoring accounts are rarely the largest, and several of them have not had a commercial conversation in two years.
Then do the harder thing. Write down who this is not for, and mean it. A profile that excludes nobody is a description, not a profile — and it will not change a single decision about where the team spends next quarter.
Four things worth remembering.
- “Anyone who needs IT” is not a market. The cost is the opportunity you never recognised, not the business you turned away.
- Technical maturity is a poor predictor of a good AI customer. Commercial pain with an executive owner is a good one.
- Five signals beat sector and size: process-heavy operations, an executive who owns the pain, a business event, an existing relationship, and a history of buying outcomes.
- Score your top forty accounts against them. The best ones are rarely the biggest.
Which of your customers is the first AI conversation?
A Growth Review is a structured conversation with you and your leadership team: where your AI practice is constrained, which accounts are genuinely ready, and what removing that constraint is worth. Ninety minutes, no cost, and an honest answer — including when the answer is that you do not need us.
Win the business case before you win the build.