Pricing AI services for local businesses

Pricing is the part of running an AI services agency that new owners get wrong most often, and they nearly always get it wrong in the same direction. They price low, win a few clients quickly, feel encouraged, and then discover three months later that the two busiest accounts are costing more to run than they bring in. The problem was never the selling. It was the number on the proposal.

This article is about how to arrive at that number. It does not tell you what your market will pay, because nobody can tell you that from a distance. What it can do is tell you what to consider before you commit to a figure, and which mistakes are expensive enough to be worth avoiding on someone else's account rather than your own.

Price against the alternative, not against other AI vendors

The instinct is to look at what other AI service providers charge and position yourself somewhere in that range. Resist it. Most of those providers are selling to a different buyer, in a different market, and quite a few of them are themselves underpriced and will not be trading in a year.

The relevant comparison is the option the business is actually weighing up. A plumber deciding whether to pay you for call handling is not comparing you against another AI company. He is comparing you against three things: doing nothing, paying a call answering service, or eventually hiring somebody part-time. Those are the real competitors.

Doing nothing is free but it loses jobs, and he knows it, because he has stood in a loft with his phone buzzing in his pocket and no way to answer it. A traditional call answering service has a per-call or per-minute cost that adds up quickly, and it typically takes a message rather than booking anything. A part-time employee costs a great deal more than any of this once you include employer's National Insurance, holiday, sick pay and the management overhead of a person.

Once you frame it that way, the pricing conversation changes. You are not selling a piece of software that ought to cost what software costs. You are selling coverage of a job that would otherwise cost considerably more to cover.

Always charge a setup fee

Charge a setup fee on every engagement, without exception. The revenue matters, but it is not the main reason.

A setup fee filters. A business owner who will not pay anything up front is, in the great majority of cases, a business owner who has not really decided. He is curious. He wants to try it. He will be gone by month three, and you will have spent hours on configuration, call flows, testing and hand-holding for two months of a small retainer. That is a loss dressed up as a win.

The setup fee also sets the tone. It says that work is being done, that the work has value, and that this is a professional engagement rather than a free trial with a subscription attached. Clients who pay a setup fee behave differently afterwards. They turn up to the onboarding call. They send you the information you asked for. They have skin in it.

There is a related trap worth naming: the free pilot. It sounds sensible and it almost never converts at the rate you expect. A prospect who was not willing to pay to start is rarely willing to start paying once the free period ends, and you have taught them that your time is available at no cost. If you genuinely need a first client to build confidence, discount the setup fee rather than removing it, and be explicit that it is a one-off introductory rate.

Suggested starting points

The figures published on the pricing page exist as a starting point, not a promise. They reflect what services of this kind commonly sell for. Whether your particular market bears them depends on where you are, who you sell to, and how well you present the offer.

• Entry package - around £495 setup and £149 per month. Suits a sole trader or a very small firm who mainly needs the phone answered and messages captured reliably.
• Middle package - around £895 setup and £299 per month. Call handling plus follow-up and booking, which is where most local businesses actually sit.
• Full package - around £1,495 setup and £549 per month. Broader automation across enquiries, quotes, reminders and review requests, for firms with staff and volume.

Treat these as anchors to think against. In a market with low commercial rents and small firms, the entry tier may be the ceiling. In a city with competitive trades who spend real money on advertising, the middle and full tiers are the sensible defaults and the entry tier mostly serves as a comparison point. See what's included for the detail of what sits behind each level, and think about how to package those levels so a local business understands them before you quote any of them.

Usage costs are the thing that catches people out

This deserves its own section because it is the single most common way an otherwise healthy account turns into a loss.

AI voice minutes cost money. SMS costs money per message. Email at volume costs money. These are not fixed overheads you can ignore - they scale directly with how busy the client is, and the clients who value the service most are the ones generating the most usage. Your best account by satisfaction can be your worst by margin.

Consider two clients on the same flat monthly fee. One is a small electrician taking a handful of calls a day. The other is a busy multi-van heating firm taking calls constantly through winter, each one running several minutes, each one followed by confirmation texts and reminders. The second client might consume many times the voice minutes of the first. On an identical flat fee, one is comfortably profitable and the other may not be.

There are three sane responses:

1. Include a stated allowance. The monthly fee includes a defined number of minutes or messages, with a clear per-unit rate beyond it. This is honest, easy to explain, and standard in telecoms, so business owners already understand the shape of it.
2. Tier by volume. Put higher-volume clients on a higher tier as a matter of policy, rather than as an argument you have to win later.
3. Quote after observing. For a client you suspect is high volume, ask what their call numbers actually look like before quoting, and price accordingly.

Whatever you do, do not sell an unlimited flat fee to a business whose call volume you have not established. Unlimited is a word that sounds generous in a sales meeting and looks very different on an invoice you have to pay.

Ask about volume during discovery, always. How many calls a day in a normal week? What about your busiest month? Do you run any advertising that spikes enquiries? A business that pays for search advertising in the spring will have a very different March to its January.

What you can safely say about value

Be careful with claims. This matters commercially and it matters legally.

"This will get you thirty per cent more leads" is a claim you cannot substantiate, and you may one day have to defend it to an unhappy client who counted. "This means the phone gets answered when you are up a ladder" is simply true, observable, and entirely sufficient. The second sentence sells better anyway, because it describes something the buyer has personally experienced.

The safest ground is always description rather than prediction. You can say what the system does. You can say when it does it. You can say what the client no longer has to do themselves. You cannot say what it will do to their revenue, because you do not know their close rate, their pricing, their capacity or their competition.

If a sentence in your proposal contains a number about the client's future business, delete it. If it contains a number about what the system does, keep it.

Discounting, and why to avoid it

You will be asked for a discount. Local business owners negotiate; it is a normal part of how they buy, and price is only one of the objections you will hear regularly. The question is what you give away.

Do not discount the monthly fee. The monthly fee is the business. A reduced retainer is a permanent reduction in the value of that account for as long as it exists, and it is very difficult to raise later without a conversation nobody enjoys. If you must move, move the setup fee, or spread it across the first two invoices, or add a month of service rather than removing money.

Better still, have something to give that costs you little. An extra automation, a second phone number, a review request flow. Concessions in scope preserve the headline price and preserve your position with the next prospect who asks the same question.

Raising prices later

Assume your first few clients are underpriced. Nearly everyone's are. That is acceptable as tuition, but do not let it become permanent.

The practical approach is to raise prices for new clients while leaving existing ones alone. Grandfather them quietly. It avoids awkward conversations with the people who took a chance on you early, and it means your pricing improves steadily without any single confrontation. Over a year or two, the underpriced accounts become a small proportion of the whole.

When you do need to raise an existing client, give notice, give a reason connected to something real such as increased usage or added scope, and give it in writing. Most reasonable clients accept a modest increase from a supplier who has been doing the job properly.

When to walk away

Some prospects should not be sold to at any price. The business that argues over the setup fee before understanding what it covers. The one who wants everything customised before paying anything. The one who has already told you, unprompted, about the last three suppliers who let them down. The one whose volume is so high and margin expectations so low that no version of the deal works.

Walking away is a pricing decision. Your time is your only real constraint in the early months, and an account that consumes it disproportionately is more expensive than an empty slot in the diary. Say politely that you do not think it is the right fit and move on.

Finding clients at all is your own work, and it takes persistence - cold calls, local networking, follow-up with people who said not now. Nothing about the pricing model changes that. But it is far easier to sustain that effort when the accounts you win are actually profitable. Look at the services themselves and at how the delivery side works before you settle on your numbers, because what you can charge follows from what you can genuinely deliver.