Business & Money
Pricing Your Product When AI Cuts Your Delivery Costs
By Jim Vernon, Editor, AI Intelligence International · Published 18 January 2026 · Reviewed against our editorial standards · About the author
When delivery costs drop sharply, the reflex is to pass it on. Sometimes that is right, often it destroys margin for no gain in volume, and occasionally it repositions your product into a category you did not want to be in.
The decision depends on three things: what your customers are actually buying, how visible your cost reduction is to them, and whether your market has price-elastic demand.
Key takeaways
- Establish what the customer is buying: If customers buy the artefact, cost transparency will drag your price down eventually.
- Capture margin first, deliberately: The default should be to keep the saving for two to four quarters.
- When to cut price instead: Cut when demand is genuinely elastic and you can serve the extra volume, or when a competitor's cost structure means the price is going to move whether you like it or not.
- Repackaging as the third option: Often the best answer is neither price nor margin but structure: move from project fees to retainers, from per-seat to usage, from deliverables to guarantees.
Establish what the customer is buying
If customers buy the artefact, cost transparency will drag your price down eventually. If they buy the outcome, the risk transfer or the relief from managing it themselves, your cost base is largely irrelevant to what they will pay.
Test this by asking why the last three customers chose you. If the answer is speed or price, you are in an artefact market. If it is trust, integration or fear of getting it wrong, you are in an outcome market.
Artefact markets deflate. Outcome markets hold, and in fact often expand as buyers become more nervous about unsupervised automation.
Capture margin first, deliberately
The default should be to keep the saving for two to four quarters. It funds the transition, absorbs the quality risk, and gives you room to discover what quality level the market actually accepts.
Cutting price immediately is irreversible in practice. Raising a price back after a public reduction costs more goodwill than never having cut it.
Use the retained margin to invest in the moat: better data, better review, deeper integration with the customer's systems. Those are what defend the price later.
When to cut price instead
Cut when demand is genuinely elastic and you can serve the extra volume, or when a competitor's cost structure means the price is going to move whether you like it or not.
The strongest case is when a lower price opens a new segment you could not previously serve at all — smaller customers, new geographies, self-serve tiers. That is share capture, not discounting, and it should come with a different packaging.
Never cut price on the same package to existing customers as a first move. Introduce a cheaper tier with visibly less service instead.
Repackaging as the third option
Often the best answer is neither price nor margin but structure: move from project fees to retainers, from per-seat to usage, from deliverables to guarantees.
Repackaging resets the comparison. A client comparing your monthly ownership fee against a free draft is making an obviously silly comparison, and most of them know it.
It also aligns your incentives with the efficiency gain, because you keep the upside of getting faster while the customer keeps the certainty they wanted.
Handling the awkward client question
Expect 'but you use AI now, so why does it cost the same?' Answer with what they are buying: the accountability, the review, the fact that the output is right and on brand and defensible.
Volunteer a written statement of your process, including what is model-assisted and what is not. Clients who ask are usually managing their own internal risk, and a clear answer removes the objection.
If a client only wants the raw draft, let them go to the cheaper option and keep the door open. A surprising share return within a year having discovered what review costs.
Modelling the alternatives
Run the cost comparison tools with your real numbers to see the spread between in-house, agency and assisted delivery. It is usually narrower than expected once review is included.
Then model each pricing option against volume assumptions. The point is not precision; it is discovering how much volume a price cut needs to generate before it beats simply keeping the margin, which is often more than the market can supply.
A worked example: an agency retainer
An agency billing a four-thousand-pound monthly retainer for twelve content pieces found AI drafting cut production time by about a third. The instinct was to cut the price by a third and keep the client happy.
Instead they held the price and changed the deliverable: the same twelve pieces plus a monthly performance review and two rewrites of underperforming pages. The client's outcome improved, the agency's margin improved, and the price never entered the conversation.
Where a client did push on price — knowing full well what the tools cost — the fallback was a tiered offer: the old scope at a lower price, or the expanded scope at the old price. Most chose the expanded scope.
Price the outcome, not the hours saved
Hourly pricing converts every efficiency gain into a pay cut. If your cost of delivery falls by a third under an hourly model, your revenue falls with it, and you have effectively worked to reduce your own income.
Fixed-price or outcome-linked pricing lets you keep the gain until competition arrives, which for most niches takes longer than the discourse suggests. The transition is easier than it sounds: price the next project as a fixed fee based on your old estimate and see what happens.
When competition does compress the market, the defensible position is the part AI does not supply — accountability, taste, domain knowledge and the willingness to say a plan is wrong. Price that explicitly rather than hiding it inside a rate card.
Frequently asked questions
Will competitors force the price down anyway?
In artefact markets, yes, over time. In outcome markets the price floor is set by trust and switching cost, both of which are slow to erode.
Should I tell customers my costs fell?
Not as a headline. Communicate improvements in speed, consistency or scope instead, which is what they actually experience.
Is usage pricing a good fit?
Only when the customer can predict usage. Unpredictable bills damage retention more than a higher fixed price does.
How do I protect margin in a tender?
Compete on scope definition and risk terms rather than unit price. Tenders scored purely on price are usually not worth winning in this market.
Should we tell clients we use AI?
Yes, plainly. Being found out is far more damaging than disclosure, and most clients care about the result and the accountability, not the toolchain.
What if a competitor undercuts on price?
Compete on scope and reliability rather than following them down. Clients who choose purely on price were rarely profitable to begin with.