Why Your Firm Cannot Bill for What It Knows
You sell hours and your clients buy answers. Everything valuable your firm has learned sits in old files and individual heads, and none of it is on an invoice.
A client rings with a question your firm has answered forty times over eleven years. Somebody spends two hours reconstructing the answer from memory and one old file, and bills two hours.
The forty previous times are worth nothing to you. That is the central economic problem of a professional services firm and almost nobody addresses it directly.
Hours are a proxy that stopped working
Billing by time made sense when time was a fair measure of the effort involved. It penalises exactly the thing that makes a firm valuable.
The partner who answers in fifteen minutes because they have seen it before bills a quarter of what the junior bills for the same answer arrived at slowly. Experience reduces revenue. Every firm knows this and most carry on regardless, because the alternative requires knowing something they do not: what the answer is actually worth.
Where the knowledge actually sits
Three places, and none of them is usable.
In old matter files. Technically retained, practically unreachable. Finding the 2019 engagement where this exact question came up requires remembering it existed, which is the hard part.
In email threads. The real reasoning, the version that explains why the obvious answer was wrong, is almost always in an email rather than in the formal advice.
In individual heads. The largest store by a wide margin, and it leaves at five thirty and permanently on retirement.
Every attempt to fix this has failed the same way
Precedent banks, knowledge management systems, internal wikis. Most firms have built at least one and it is empty.
The reason is always the same: capturing knowledge was a separate task, done after the billable work, by people whose compensation depends on billable hours. Asking a fee earner to spend forty unbilled minutes writing up a matter is asking them to pay for the firm’s asset out of their own bonus.
It is not a discipline problem. The incentive was always going to win.
Where the sector actually stands
Statistics Canada, in a release dated 16 June 2025, found that 12.2% of Canadian businesses had used AI to produce goods or deliver services in the previous twelve months. Professional and scientific services came in at 31.7%, one of the three highest sectors.
Two readings of that, both true. Professional services firms are ahead of the Canadian average, which is unsurprising given the work is largely reading, writing and judgement. And roughly two thirds of them still are not, in the sector where the fit is most obvious.
The survey measures use, not results, so being inside the 31.7% guarantees nothing. But it does mean that in this sector the advantage window is narrower than elsewhere, and a firm waiting another two years will be arriving late rather than early.
What AI does with the archive
The capture problem disappears, because nothing has to be captured. The knowledge was always there in the files and the threads. What was missing was any way to find it.
Ask the archive a question in plain language. Not keyword search across document titles. A question, answered from eleven years of matters, with the specific files cited so a fee earner can check the reasoning rather than trust it.
It finds the analogous matter, not the identical one. The 2019 engagement that is structurally the same problem in a different industry with different terminology. That is the connection an experienced partner makes and nobody else in the firm can, and it is now available to everybody.
It surfaces what a person forgot they knew. Frequently the useful result is reminding the partner of their own advice from six years ago, including the caveat they have since stopped mentioning.
The junior arrives at the partner’s answer faster. Which is the whole economics of the firm. Leverage has always been limited by how much supervision juniors require, and reducing that is worth more than any efficiency saving.
What this does to the pricing conversation
It makes fixed fees defensible, which is the change that matters commercially.
Most firms resist fixed pricing because they cannot predict the effort, and they cannot predict it because they have no reliable record of what similar matters actually took. Once you can see across the archive, you can price a category of work from evidence, with the variance visible.
Then the incentive inverts. Getting to the answer faster increases margin instead of reducing revenue, and every improvement in how the firm captures and reuses what it knows pays the firm directly.
That is the point at which knowledge stops being a cost centre nobody funds.
Start with one question type
Firms that try to make the whole archive searchable at once usually stall, because the archive is large and messy and the project has no obvious end.
Pick one question your firm answers repeatedly. Every practice has three or four: a particular structuring question, a specific compliance situation, a recurring dispute. Take that one, gather the matters where it came up, and make those answerable.
Two things follow. The value is visible within weeks rather than at the end of a year long programme. And you learn what your files are actually like, which is always worse than anybody expects and much better to discover on a small scope.
What it does to recruitment
An underrated consequence, and for some firms the largest one.
A firm where a two year qualified person can reach the right material in minutes is a firm where that person develops faster and stays longer. The alternative is the current arrangement, where juniors learn by interrupting seniors, seniors resent the interruptions, and both conclude the firm is badly run.
In a market where every professional firm is competing for the same small pool of people, that is a recruitment argument as much as an efficiency one.
The obvious professional caution
None of this replaces the judgement or the sign off, and it should not be sold as if it does.
A retrieved answer from 2019 may rest on law, standards or circumstances that have since changed. The value is in getting an experienced person to the relevant material in four minutes instead of two hours, with the sources visible so they can see what they are relying on. The professional responsibility does not move and neither does the signature.
Any system that produces confident answers without showing where they came from is unusable in this sector, and we would not build one.
Sources
Statistics Canada, Analysis on artificial intelligence use by businesses in Canada, second quarter of 2025, released 16 June 2025.
AI Optimize makes eleven years of matter files answerable in plain language, with every source cited so a fee earner can check the reasoning. That work sits under Custom AI Integrations.
Related reading
What Happens When Your Best Person Leaves
Every small business has somebody who holds more than their job description. The exposure is not their workload, it is everything they know that was never written down.
Why Professional Services Firms Cap Out on Partner Hours
Every firm that bills for expertise eventually runs into the same ceiling. Revenue is bounded by the hours of the people clients actually want, and hiring more of them is slow and expensive.
WHAT WE BUILD
