How to Price an AI Training Workshop

Price an AI training workshop on the value of the material, not the hours in the room. A one day session costs you two to five days of preparation the first time you run it and about half a day every time after that, so any price built from the delivery day alone loses money on delivery one and leaves money on the table from delivery three onward. The three workable structures are a flat day fee, per-seat pricing, and a fee plus a follow-on retainer, and which one fits depends entirely on whether you intend to run the material again.

Why hourly pricing breaks for workshops

Consulting rates assume the work is consumed once. Training does not behave that way. The deck, the exercises, the sample data, the prompts you tested the week before and the three questions you now know every room asks: all of that is an asset you build once and rent repeatedly. Charging by the hour prices the rental as though it were construction.

The number that matters is your true cost for the first delivery. Count it honestly:

• Content build: 12 to 30 hours for a full day, depending on how much you are customising to their stack.

• Customisation per client: 3 to 6 hours to swap in their tools, their data shapes, their actual workflows.

• Delivery: the day itself, plus travel and the dead time around it.

• Follow-up: the questions that arrive for two weeks afterwards, which you will answer whether or not you charged for them.

A first delivery of a one day workshop is realistically four to six working days of your time. If you quote a single day rate against that, you have taken a 75 percent pay cut and called it a booking.

Three ways to price an AI training workshop

Flat fee per delivery

The default, and the right choice most of the time. Quote one number for the session regardless of who sits in the room. It is easy for a buyer to approve, it survives the headcount changing, and it captures the reuse economics: your margin improves each time you run it, which is exactly the reward for having built something worth running twice.

Set the first-delivery fee to cover your real build cost with a normal margin, not your day rate. If a day of your consulting time is X, a first full-day workshop delivery is rarely below 3X, and 4X is not aggressive when the content is custom.

Per seat

Per-seat looks attractive because the arithmetic flatters you in the room: 20 people at 400 each is 8,000, and that beats most day rates. The problem shows up in month three, when the same client wants a second session for 6 new starters and your identical effort now bills 2,400. You have taught the buyer that your price tracks their headcount rather than your work, and the second sale is the one that hurts.

Per-seat works when the buyer genuinely has a per-head training budget and the group is large and stable. Put a floor under it: a minimum charge equal to your flat fee, with the per-seat rate applying above that. That converts per-seat from a discount mechanism into an upside mechanism.

Fee plus retainer

The honest structure for a subject where one day of training does not survive contact with real work, which describes most AI enablement. Charge a delivery fee, then a small monthly amount for a standing office-hours slot and prompt reviews for three months. It solves the problem that getting a team to actually use AI is a habit problem rather than a knowledge problem, and it puts a price on the support you would otherwise give away.

A worked example of the amortisation

Say your consulting day rate is 1,200. You build a one day workshop on using AI for internal documentation. Build plus customisation plus delivery plus follow-up on the first client is 5 days, so your true cost basis is 6,000. Compare the two structures across three deliveries to three different clients, assuming 1.5 days of effort on each repeat:

The totals land in roughly the same place, which is the point people miss. The difference is the shape. Per-seat front-loads your revenue onto the delivery where you did the most work and then punishes every repeat, which is backwards: repeats are where the asset you built is finally earning. Flat fee is a loss on delivery one and a strong margin thereafter, which is the correct incentive, because it pays you to run the thing again rather than to chase ever-larger rooms.

If you only ever intend to run a workshop once, for one client, the flat-fee loss is real and you should price delivery one at cost recovery plus margin, closer to 6,500. The same logic that governs productised AI services applies: the price is defensible when the thing you sell is repeatable, and fragile when it is not.

What to put in the quote

1. A fixed scope: number of sessions, length, maximum attendees, and what happens if they exceed it.

2. What the client supplies: room, sample data, accounts and licences for the tools you are teaching.

3. Materials licence: whether they may re-run your deck internally without you. If yes, that is a separate and higher number.

4. Recording rights. A recorded session competes with your next delivery to the same company. Price it or prohibit it, do not leave it unsaid.

5. A cancellation window, typically full fee inside 7 days and half inside 14, because your build cost is already sunk by then.

The materials licence is the clause people forget and regret. A client who can re-run your workshop internally has bought your product, not your day, and should pay accordingly. It is the same asset-ownership question as who owns the prompts you wrote for a client, and it is much cheaper to settle in the quote than afterwards.

Setting the first number

If you have never run one, work backwards rather than guessing. Take your consulting day rate, multiply by the real days in delivery one, add 20 to 30 percent margin, and quote that as the flat fee. Then decide in advance what you will charge for delivery two, and hold it: the most common failure is quoting the second one off the memory of the first one's effort rather than its value.

One pricing note specific to AI training. Buyers frequently ask whether the content will still be correct in six months, and the honest answer is that the tool-specific parts will not be. Say so, and offer a refresh at a stated price rather than pretending otherwise. That framing also sets up the retainer conversation naturally, and it sits well alongside how you scope a productised service so it does not become custom work.

Frequently asked questions

How much should I charge for a one day AI training workshop?

Work from your true first-delivery cost rather than a market figure. If your consulting day rate is 1,200 and delivery one takes five days of real effort, a flat fee in the 4,500 to 6,500 range is defensible, with the lower end assuming you will reuse the material and the higher end assuming you will not. Training sits alongside the other ways to make money with AI skills rather than replacing them.

Should I charge per person or a flat fee for AI training?

Flat fee by default. Per-seat pricing ties your revenue to their headcount, which means your second and third sessions for the same client pay far less for identical work. If you do use per-seat, set a minimum charge equal to your flat fee so small groups cannot drag the price below your cost.

Do I charge extra to customise a workshop to a client's tools?

Yes, as a separate line. Customisation is three to six hours of work that does not transfer to your next client, so it should not be buried in a fee whose margin depends on reuse. Quoting it separately also lets a price-sensitive buyer choose the generic version.

What if the client wants to record the session?

Price it or refuse it, but decide before you quote. A recording lets them train future hires without you, which replaces your repeat business. A common approach is to allow an internal-use recording with a twelve month expiry for a meaningful uplift on the fee.

How do I price a follow-up session after the first workshop?

At the same flat fee, not a discounted one, unless the scope is genuinely smaller. Your effort on a repeat drops to roughly a third, and that improved margin is the return on the content you built. Discounting repeats means the asset never pays for itself.