The quote for an AI tool shows one figure, the licences. Some of the rest is cash: set-up, connecting the tool to existing systems and checking the contract. Most of it is people's time: learning a tool, checking and fixing its output, answering colleagues' questions and keeping track of how AI is used. Those hours are covered by salaries already paid, so they rarely show as a new cost, yet they come out of other work and need counting.
Suppose a licence costs about as much each month as an hour of that person's time, counting salary and on-costs. An hour a week learning, checking and fixing in the first months then costs more than four times the licence, before anyone's time answering questions. Whether that is repaid depends on what the tool saves. Work out both, the hours it takes and the hours it saves, with the organisation's own figures.
No new moneyLink to this section
People work with the AI already built into tools the organisation pays for, and a team that needs more makes its case for a future budget. Oversight stays simple, though someone decides which built-in features are switched on, and for whom. Suppliers keep adding them, sometimes on by default and sometimes with a price rise at renewal.
Easy to miss. The time people put in is still paid for, and because it never appears in a budget, it is rarely considered.
When to choose it. It makes sense as a starting point, since existing tools may include AI features few people use yet, with an end date and someone tracking the time.
From team budgetsLink to this section
Managers pay for small tools from budgets they already run, so teams with a keen manager start quickly and what people can use depends on who they work for. Costs spread across many budgets, and nobody sees the whole bill.
Easy to miss. Small amounts tend to be treated as a small commitment. Each purchase gets little scrutiny, yet brings a contract, a renewal date and terms for the organisation's information, agreed by whoever bought it. Some lower-priced plans let the supplier use what is put in to improve its products, where plans sold to organisations often rule that out.
When to choose it. Choose it where someone keeps one list of what each team pays for, what information goes into each tool and when it renews, so useful tools are bought once and checked properly.
A set annual budgetLink to this section
One person holds the budget and a record of what it pays for, and anything outside it needs its own case. Costs are predictable and easy to report, and mid-year ideas wait for the next round.
Easy to miss. The budget is usually built on licence prices, the figure on the quote, before anyone knows which uses will pay for themselves. Once people's time is counted, as above, licences are usually the smaller part.
When to choose it. It is best where the budget counts people's time to learn, someone's time to answer questions, and the time to check how AI is used. Keeping a share unallocated leaves money for ideas that come up mid-year.
Budget tied to resultsLink to this section
Teams that show results get more, so people have reason to track and share what works, and teams compete for the same money. Someone agrees what counts as a result and checks the figures before money is released, a regular leadership task.
Easy to miss. It rewards whatever is easiest to count. Time saved shows within weeks, while fewer errors reaching clients, or staff staying longer, take months, so those uses lose funding even when they matter more. Teams competing for money also have reason to overstate their results, so someone with nothing to gain from them should check the figures.
When to choose it. It suits organisations where the leadership team trusts the measures and agrees them, slower results included, before any money is allocated.
Whatever it takesLink to this section
People get the tools and help they ask for, quickly, giving the most freedom to find value. There is no limit on costs, and with usage-based pricing, where the bill rises with use, it can grow faster than anyone tracks. Expectations, and the pressure to show it was worth it, rise with the money.
Easy to miss. Money buys tools within weeks. How quickly people change the way they work depends on their managers, their confidence and their time, and money changes those slowly.
When to choose it. It is worth choosing where a large share goes to people's time and support, each commitment has a named person and a review date, and the leadership team agrees upfront what result would end the funding.
How to chooseLink to this section
- How much of the AI in tools the organisation already pays for is used?
- Who will see the whole bill, including what teams buy for themselves?
- How many hours will learning, checking and fixing take in the first months, and does the plan count them?
- Is money what limits adoption, or managers' time and people's confidence?
Tensions with other decisionsLink to this section
- No new money with an Open or higher appetite and a Quick or faster pace. The appetite relies on new tools on real work, with most teams starting together, while the list barely changes and nobody has time set aside to sort out problems. The three conflict. Set a budget for tools and support, or bring the pace or appetite down.
- From team budgets with the same appetite and pace. Problems appear in several places, with nothing set aside centrally for better tools or someone to sort them out. Set a budget for both, or slow the pace.
- Whatever it takes with On their own time or Occasional sessions. Renewals are due before most people have found out what the tools can do. Put part of the money towards their time to learn, or delay purchases until people have learnt to use the tools they already have.
- Budget tied to results with efficiency and better work for people as the aims. Better work shows slowly, so at each funding decision it tends to get less money than efficiency. Agree how to measure it first, or fund it separately.
When to revisit itLink to this section
- At each renewal date and budget round.
- At the end date set for No new money, or the end point for Whatever it takes.
- When the bill grows faster than use, one tool appears in several teams' costs, or the hours run over the plan.
If the leadership team would like support with this decision, see Getting outside help.
Something missing or out of date here? Tell us.