Calculating the real cost of AI subscriptions for a law firm
By the ChamberOne team 4 minute read
The monthly price of an AI tool is easy to find and easy to compare. It is also the smallest number in the calculation. A firm that buys ten seats at a modest fee and then discovers that only three people use them, that everyone spends time checking output, and that the usage meter is running has not saved money. This article sets out how to count the real cost before you commit.
Key points
- Count users, not heads, and include the time spent learning and checking.
- Per-seat suits regular users; usage or pooled plans suit occasional ones.
- Read the allowance and what happens when you exceed it.
- Review real usage after three months and adjust the plan.
The three pricing shapes
Most legal AI products fall into one of three patterns, and some combine them.
- Per seat. A fixed fee per user per month. Easy to budget, wasteful when seats go unused, and awkward when staff come and go.
- Usage-based. You pay for what you consume, measured in documents, pages, queries or tokens. Cheap for light use, unpredictable for heavy use, and a worry if a large matter suddenly needs analysing.
- Firm-wide or tiered. One price for the firm, often with an allowance and a cap. More predictable and simpler to administer, though the allowance may not suit your pattern.
Start with who will really use it
Do not count heads; count users. Make three lists: people who would use the tool most days, those who would use it occasionally, and those who would never open it. A per-seat plan suits the first list. For the second, a pool or usage-based plan may be cheaper. The third list should cost you nothing.
Be realistic about partners. Many will be keen at a demonstration and then ask an associate to do the work. If so, count the associate's seat, not theirs, unless you want partners to review output on screen.
The costs that are not on the invoice
Add these to the subscription before you compare options.
- Training time. Each person needs time to learn the tool and agree how the firm uses it. Count hours multiplied by the cost of those hours, not just the fee earner rate you could have billed.
- Review time. AI output needs checking. If the saving on a task is forty minutes and checking takes twenty, the gain is twenty, and only if the output is usable.
- Set-up and integration. Importing templates, connecting to your document store, setting permissions.
- Policy and governance. Someone has to write and maintain the firm's rules for use.
- Switching costs. If you leave, what happens to your data and any custom set-up?
- Tax and currency. Check whether quoted prices include tax, and in which currency you will be billed.
A worked illustration
Suppose a twelve-person firm is comparing a per-seat tool and a firm-wide plan. These numbers are made up to show the method, not to suggest real prices.
The per-seat tool costs a fixed sum per user per month. Eight people would use it, so the firm buys eight seats. The firm-wide plan costs a single monthly amount that happens to be a little more than six seats. Add training time for eight people, plus a monthly hour of review time per user for output checking.
On the face of it the firm-wide plan costs more than six seats and less than eight, so it wins if eight people really use the tool. But if only four people use it regularly, the per-seat plan for four is cheaper. The right answer depends on adoption, which is why the first trial should be small and measured.
Where a firm-wide plan wins
A firm-wide plan tends to win when use is broad and unpredictable. If everyone from the intake team to the partners might ask for a summary, the administrative saving of not counting seats is worth something. It also helps where staff turnover is high, since no one has to reassign licences.
It tends to lose when a small group does most of the work, or when the allowance is so low that heavy users hit a ceiling and pay overage. Read the cap carefully and ask what happens when it is reached: are you throttled, charged more, or asked to upgrade?
- Use is broad and unpredictable
- Staff turnover is high, so no licences to reassign
- A small group does most of the work
- The allowance is so low that heavy users pay overage
Questions to ask before signing
- What exactly is metered, and can I see usage by user and matter?
- What happens when we pass the allowance?
- Can we change plan mid-term, up or down?
- Which AI features are included, and which cost extra?
- What happens to our data and templates if we cancel?
- Can we switch AI features off for some matters or clients?
Measuring after you buy
Set a review date three months after rollout. Look at who used the tool, how often, and for what. Ask users how much time they saved and how much they spent correcting output. Compare against what you assumed in the business case.
Adjust the plan then, not at renewal. If half the seats are idle, drop them. If usage is heavier than expected, a different plan may be cheaper. The aim is to pay for the work the tool actually does for the firm.
Finally, do not forget the client. If AI changes how long a task takes, check the guidance from your regulator or bar and your engagement terms on how technology costs and time savings should be reflected in what you charge.
- 1Set a review date three months after rollout
- 2Look at who used the tool, how often and for what
- 3Ask users about time saved and time spent correcting
- 4Compare with the business case
- 5Adjust the plan, then review again
- Then back to step 1.
Small firms and sole practitioners
For a firm of two or three, the maths is different. Administration is a larger share of the cost than the fee, and a month-to-month plan that you can cancel is worth more than a discount for a long commitment. Avoid annual contracts until you have used the tool for a quarter.
Equally, a sole practitioner may find that a usage-based tool costs next to nothing in quiet months and a lot in a month with a large file. Set a spending alert if the vendor offers one, and keep a note of which matters drove the cost so you can recover it where your terms allow.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
AI and the billable hour: what actually changes and what does not
How faster drafting puts pressure on hourly billing, and which fee models suit which work.
5 minute readFixed-fee pricing with AI-assisted scoping
How to use past matter data and AI summaries to scope a fixed fee, define exclusions and set a review point.
5 minute readRecovering unrecorded time: how AI time capture works and its limits
How suggestions from calendar, email and documents help fill timesheet gaps, and why the fee earner must still confirm each entry.
5 minute read