How to Raise Your Rates on an Existing AI Client

Announce the new rate, give a date, and do not explain your reasoning at length. That is the whole mechanic, and almost every failed rate increase fails by doing the opposite: a long justification, an apologetic tone, and an implied invitation to negotiate that the client accepts.

The awkwardness is specific to existing clients. New clients get quoted a number and either buy or do not. An existing client has an anchor, usually the number you quoted when you were less experienced and more nervous, and every month you leave it there makes the eventual correction larger.

Get the timing right, because it does most of the work

There are three good moments and one bad one.

**Good: at a natural boundary.** Contract renewal, the start of a new engagement phase, or a calendar year. The increase reads as a scheduled event rather than a reaction to something.

**Good: after a visible win.** The month after you shipped the thing that worked is the month your value is least abstract. This is not manipulation, it is the only time the client has fresh evidence.

**Good: when scope has genuinely drifted.** If you are now maintaining a system you originally only built, the engagement changed and the price should reflect the engagement that exists. This is the easiest conversation of the three because you are describing reality rather than asking for more.

**Bad: immediately after a problem.** Even a problem you fixed well. The two events will be linked in the client's memory no matter how you phrase it, and you will spend the next call defending a connection you did not make.

Pick the number without flinching

Two failure modes, and the second is more common than people admit.

Too small is the frequent one. A five percent increase costs you the same awkward conversation as a twenty percent increase and buys you almost nothing, which means you get to have the conversation again next year. If you are going to spend the social capital, spend it once.

Too large without a scope change is the other. An increase over roughly thirty percent on unchanged work invites the client to re-run the buy decision from scratch, and you do not want that decision reopened. Above that threshold, restructure the engagement rather than repricing it, so the client is comparing a new thing to a new price rather than comparing your old price to your new one.

For AI work specifically, there is a trap worth naming. Your API costs have probably fallen rather than risen, and a client who reads the news knows it. Never anchor an increase to your costs. Anchor it to scope, demand or the value delivered, all of which are true and none of which invite a spreadsheet. If the cost question does come up directly, whether to pass an AI price cut to your client sets out a coherent position to hold.

The message

Short, dated, unapologetic. Email, so the client can absorb it without performing a reaction, and follow the next call rather than leading with it.

From 1 January my rate for this work moves to 1,400 a day, from 1,150. The scope stays as it is and so does everything else about how we work. Happy to walk through the plan for the next quarter on Thursday's call.

Three things that message does. It gives a specific date at least thirty days out, which is the professional minimum and gives the client time to plan rather than react. It states the number plainly, without a paragraph of reasons that reads as negotiating against yourself. And it separates the increase from the relationship, so the client is not being asked to evaluate whether they still like working with you.

What it deliberately omits: an apology, a comparison to market rates, and any sentence beginning "unfortunately". Each of those signals that you expect resistance, and clients are good at reading that signal.

The three responses

**Acceptance.** Most common by a wide margin, especially from clients who value the work. Say thank you once and move on. Do not over-thank, which retroactively frames the old rate as the correct one.

**A counter.** Usually a request to split the difference or delay a quarter. Decide your floor before you send the email, not during the reply. Delaying by a quarter is often a reasonable concession because it costs you little and preserves the number, which is the thing that compounds. Splitting the difference on the rate itself is worse, because it sets the precedent that your stated price is an opening position.

**Quiet acceptance, then shopping.** The one people do not plan for. The client agrees, then starts evaluating alternatives without telling you. You will notice it as slower approvals and more questions about documentation and handover. The response is not to lower the rate, it is to make the next two months visibly excellent and to make sure the client knows what switching would actually cost them in context and continuity.

When they come back with a competitor's quote

This deserves its own answer because the instinct is to match it, and matching is almost always wrong.

A cheaper quote from someone else is not a comparison of like for like, and the honest response says so without disparaging anyone. The things a replacement does not have are specific and worth naming out loud: knowledge of why the system is built the way it is, the undocumented decisions, the context that makes a two-hour fix two hours rather than two days. That is not a loyalty argument, it is a switching-cost argument, and it is the only one that holds up.

Then stop talking. Do not add a discount to the end of that paragraph. If the client still wants the cheaper option after hearing it, the work was a commodity to them, and commodity work at your old rate was not a business worth protecting.

One genuine exception: if the competing quote is for a meaningfully smaller scope, say so and offer the smaller scope at a matching price. Competing on scope is fine. Competing on rate for identical work resets your pricing permanently, because a rate you discounted once is a rate you will be asked to discount again.

Occasionally a client leaves. If your rate increase was reasonable and the client leaves over it, they were a price-sensitive account that would have left at the next increase anyway. Losing one of those is a smaller loss than the three years of underpricing that preceded it. When a departure needs managing rather than accepting, how to fire a client on an AI project covers the same conversational territory from the other direction.

Avoiding the situation next time

The reason this conversation is hard is almost always that it has been deferred. Three structural habits remove most of the pain.

• Put an annual adjustment clause in the contract from the start. An increase that was agreed in advance is an administrative event rather than a negotiation.

• Reprice at every scope change, not on a calendar. Scope creep priced at the old rate is the single largest source of underpaid engagements.

• Quote new clients at your current rate, always, so your existing clients are the only ones anchored low and the gap stops widening.

There is also a growth path that removes the ceiling entirely, which is moving from time-based billing to something that scales with outcomes. That changes what you are selling rather than what you charge for it, and the options are laid out in AI monetization strategies and, for recurring work specifically, how to turn AI consulting into recurring revenue.

Questions

How much notice should I give?

Thirty days minimum, sixty for a retainer the client budgets for. Enough that they can plan, not so much that the conversation stays open for a quarter.

Should I raise rates for every client at once?

No. Handle them individually, starting with the client where you are most underpriced and least worried about the outcome. The first conversation is the hardest and you want it to be a rehearsal.

What if the client asks me to justify it?

Answer briefly and factually once: demand, experience, the scope as it now stands. Do not produce a cost breakdown. A justification that invites line-by-line examination turns a price into a budget review.

Can I raise rates mid-contract?

Not unilaterally, unless the contract provides for it. Wait for the renewal, or trade the increase against a scope change both sides agree to. Breaking a price commitment costs more trust than the increase is worth.

What if they are already my cheapest client and my best reference?

Price the work properly and value the reference separately. A discount that is never named is a discount that is never appreciated, and if the reference is genuinely worth something, say so explicitly and agree what you get for it.