Is AI consulting worth it for a UK business?

Where the fee goes

What the £2,000 buys, line by line.

Fourteen days, three sessions, four documents. Here is the work behind it, in order.

1. Days one to three: the map of what actually happens.We sit with the people doing the work and write the process down as it runs, not as the procedure says it runs, including the hand-offs nobody has documented. Most businesses have never seen this in one place.
2. Days four to seven: the measurements underneath it.Volumes, waiting time against working time, rework, and where a job queues, taken from your own exports, timestamps and diaries rather than sector benchmarks. A benchmark tells you about somebody else.
3. Days eight to eleven: opportunities, separated and tested.Viable AI opportunities are separated from the ones that are really a process fix, a pricing decision or a system nobody configured properly. Each survivor gets its dependencies, failure conditions and the point where a human must stay in control written down.
4. Days twelve to fourteen: the economic case and the ranking.Upside, cost, assumptions and risk for every opportunity worth considering, then a ranked investment order and a fixed price for the highest-value scope the evidence justifies. The assumptions are shown, so you can disagree.

All four documents are yours outright: build with us, take them elsewhere, or file them and do nothing.

Exclusions

What is never in the price.

Run this against every quote, ours included. The difference between three quotes is usually here, not in the headline fee.

• Software, model and hosting costs. Licences, API usage and hosting are billed to you by the vendor at their price, not marked up and resold. The expected monthly figure is named in the build proposal so it is a decision rather than a surprise.
• Your team’s time. Three sixty-minute sessions, plus the hours spent pulling exports and checking that what we wrote down matches what happens. Real cost, and nobody carries it for you.
• The build itself. The audit prices the build. It does not include it, and the audit fee is not deducted from it.
• Data you were always going to have to fix. Where records are inconsistent enough that no system could read them reliably, sorting that out is its own piece of work, scoped separately rather than absorbed into a build.
• Training programmes and change management. We hand over systems your team can run and show them how. A structured training programme is a different product and other firms sell it better.
• Anything the proposal does not name. If it is not in the written scope it is not in the price, and the change process is agreed before the build starts.

The awkward outcome

What happens if we find nothing?

It happens, and it is a result rather than a failure. Sometimes the process is already tight and the money is going somewhere else. Sometimes the constraint is a pricing decision, a staffing decision or a rule nobody has written down, and software on top of any of those makes the problem faster rather than smaller.

When that is the answer you get it in writing with the measurements behind it, you keep all four documents, and you owe nothing further. No second proposal, no consolation retainer, no smaller build invented to justify the fortnight. We are paid the same either way, which is the only arrangement that makes "do not build this" trustworthy from the firm that would have built it.

It is not a refund. You bought fourteen days of measurement and received them. What changed is the decision they point to.

The other side of the ledger

Every number above is a cost. Here is the honest return.

In the illustrative week our audit graphic reads, 11.5 of 36.5 working hours are collection, retyping and chasing. Put one £28,000 admin salary against those hours and roughly £9,000 a year flows through work software could carry, before you count the jobs that waited. Your number will be different, which is the point: the £2,000 audit exists to measure it, and a build is only proposed when the fixed price stands a serious chance of repaying itself inside a year.

That arithmetic is illustrative and labelled as such, because it is built from a worked example rather than a client. It is worth spelling out what the return is made of, because "efficiency" is not a line in anybody's accounts.

• Hours that move from people to software. The clearest and the easiest to overstate. It becomes money only when those hours are redeployed onto work that earns, or never hired for. If the same people do the same jobs slightly faster, the saving is invisible in the accounts, and we say so rather than counting it.
• Work that stops waiting. Usually the larger number and almost never the one a business measures. A quote that goes out the same day rather than four days later saves no labour; it changes the odds on the job. The audit measures queueing separately from working time for that reason.
• Errors that stop reaching the customer. Rework already has a cost visible in your own data: credits raised, jobs revisited, the second site visit nobody billed for.
• Capacity you do not have to hire. Not "replaces a person". The next twenty per cent of volume arrives without the next hire, a number your own growth plan already contains.
• Decisions made on figures rather than instinct. Named here because leaving it out would be dishonest, and counted at zero because we cannot measure it.

What we will not do is show you a percentage from somebody else's case study. AGMM has no published client outcome it can link, so there is no ROI multiple here. Ask any firm quoting one for the baseline and who measured it.

Doing the sum yourself

How do you work out whether it is worth it for you?

You can get most of the way to an answer in an afternoon, without buying anything. Four numbers.

1. Pick the one process that hurts. Not the whole business: the one where work piles up, where somebody is the single point of failure, or where customers chase you.
2. Count how often it runs a week, and how long it takes.Rough is fine. Ask the person who does it rather than the person who owns it: the two answers are rarely the same.
3. Multiply by a loaded hourly cost. Salary plus employer costs over working hours. A £28,000 salary is roughly £18 an hour loaded, and being ten per cent out does not change the decision.
4. Add what the waiting costs. Jobs lost to a slow quote, invoices sent late, the visit that happened twice. Usually larger than the hours, and almost always the number nobody has.

If that total is comfortably under £2,000 a year, keep your money. If it is several times the audit fee, the question is no longer whether to measure it properly but who you trust to do the measuring. And if you cannot produce the numbers at all, that is itself the finding: a business that cannot see a process cannot price fixing it.

Evidence

Can you show one thing you have actually built?

Yes, and it is open. A real issued job from a quoting system we built and run: a first-floor build-over in Stockport, 63 measured lines, £98,214.72, published in full so every line traces to a dimension somebody can check. Read all 63 lines of it.

It is not a case study and claims no saving. It is the kind of artefact this market does not produce, and the thing to ask every firm you brief for. What we cannot show yet is a named client outcome with a published figure, and we will not imply one.

The market

What does everyone else charge?

Useful context, because the range in this market is enormous and nobody publishes it side by side. These are the bands UK firms advertise, read from their own pricing pages.

Named firms, sorted by what each one publishes, with sources and dates:the UK published price index.

Our proposal follows the scope the audit proves, from one focused system to a larger programme spanning teams and workflows. There is no artificial ceiling, and the people who scope the work are the people who lead the build.

Where we stop

When will we tell you not to build?

The audit sometimes finds the constraint is not a software problem. When it is a pricing decision, a staffing decision, or a process nobody has written down yet, putting software on top of it makes the problem faster rather than smaller, and more expensive to unpick later. When that is the answer, that is the answer, and you still keep the audit.

Before you decide

What is the worst case?

You spend £2,000. Fourteen days later you know what the process actually costs you, which steps are queueing, and what we would build, priced. Then we tell you not to build it, because the numbers land small. You keep every document and owe us nothing further. That is the worst case, and it is more than most firms in this market deliver as their best.