What an AI consultant costs in the UK

Ask what AI consultancy costs in the UK and the quotes come back an order of magnitude apart. The number is the least useful part of the quote. What matters is the pricing model it sits in, because the model decides who carries the risk.

The three pricing models

Nearly every quote you receive will be one of three shapes:

• —Day rate. You pay for time. The quote looks precise, but the total is open. The deliverable arrives when it arrives, and the meter runs until it does.
• —Retainer. You pay for availability, by the month. Sensible for ongoing advice; a poor fit for a one-off question like “where should we use AI?”, which has an end.
• —Fixed price. You pay for a defined deliverable. The number is the number. If the work turns out harder, the consultant absorbs it.

What a day rate hides

A day rate moves the scope risk to you. If the consultant underestimated, your invoice grows. If they need time to learn your industry, you fund the education. Neither appears on the quote.

The pattern to watch is the open-ended discovery phase: weeks of workshops before anyone commits to a finding. Discovery is real work, but on a day rate nobody is paid to finish it. Ask, before signing, what the discovery phase produces and when it ends.

What drives the price

Four things, roughly in order:

• —Overhead. A large firm’s rate funds offices, partners, and pitch teams. An independent’s rate funds one person. The advice is not ten times better because the invoice is.
• —Advice versus delivery. Telling you what to do costs less than doing it. Be clear which you are buying; quotes blur it.
• —Scope clarity. A vague brief gets a padded quote. The tighter you can state the question, the less you pay for the consultant’s uncertainty.
• —Who does the work. At larger firms the person who pitched is rarely the person who delivers. Ask who is on your engagement, by name.

Fixed price, and who carries the risk

A fixed price is a claim about competence. To offer one, the consultant must know the work well enough to bound it, and absorb the cost of being wrong. As a worked example: the Opafex AI audit is one week and £2,500 fixed, paid 50% on booking and 50% on delivery. An audit that turns out harder is our problem, not your invoice.

We cite our own price because we can vouch for it. Do not read it as the market rate. Whatever quote you receive, apply the same tests. Is the deliverable defined, is the price closed, and who pays for overrun?

How to tell what you are paying for

Before you sign anything, get written answers to five questions:

• —What is the deliverable? A named document with named sections, not “insights” or “a roadmap”.
• —Who owns it? You should, outright, including the right to hand it to another supplier.
• —What happens if it takes longer? The answer tells you which pricing model you are in, whatever the quote says.
• —How much of our time does it need? A defined number of hours shows the process exists. “As much as we can get” shows the process is missing.
• —Can we act on it alone? The report should contain steps you can take without the author. If every recommendation routes back through them, see the callout above.

What good value looks like

Cheap and expensive are the wrong axis. Good value is a closed price for a defined deliverable you own. It takes a bounded amount of your team’s time, and you are free to ignore every recommendation. If you want to see what a deliverable like that contains, we have written up what happens in a week-long AI audit. And before buying advice, know the decision it should inform: see build, buy, change the process, or hire.