There is a common view that law firms are bad at building software. Plenty of partners have a story to back it up: the bespoke system that took two years, cost far more than planned, and was being worked around within months of going live. There’s a matching story about the subscription renewed every year because nobody checked who still logs in. We think both have the same cause, and it is decided before anyone signs anything. The firm chose without looking closely at how the work moves through the firm, or at what each option would cost to keep running.
There are four choices
The question is normally put as build or buy. That leaves out two options, and for a small or mid-sized firm they’re the two we would look at first.
Buying gets you a product: practice management, case management, accounts, time recording. It works on day one, someone else maintains it, and other firms have already found the bugs. You also take its idea of how a matter should run.
Extending keeps that core system and adds to it. A form that creates the matter for you. A report the product doesn’t offer. A link between two systems that currently share data through someone retyping it. This is possible where the vendor gives access. Clio, Actionstep and Microsoft 365, for example, offer API access we can connect to, and LEAP, Proclaim and InfoTrack can be connected where the vendor gives access (more on integrations).
A build gives you software of your own, made around one part of how your firm works.
Then there’s doing nothing, or stopping. Leaving a problem alone is a fair decision when it costs less than any fix. Stopping means cancelling a service because a feature of something you already own, or a simpler way of working, would do the job. Software sellers have no reason to suggest either.
Count what it costs to keep
Quotes tell you what it costs to get something. The decision should rest on what it costs to have it for the next several years, and each option hides its costs in a different place.
With a product, look at per-user fees as the firm grows, price changes at renewal, paid add-ons and the eventual cost of moving off it. With an extension, count the work plus keeping the connection going when the vendor changes its API, and ask what happens if the vendor changes its terms of access. A build carries hosting, security updates, support, documentation and change requests for as long as you run it. It also needs someone who understands it after the person who made it has moved on.
Stopping isn’t free either. Before cancelling, check the notice period, get your data out in a form you can use, and work out what anyone now has to do by hand. Everything except doing nothing costs staff time to change, and that belongs in the sums alongside the invoices.
The hardest comparison is a subscription against a small tool of your own. An in-house solution can be easier than a subscription, and simpler, because it does the one job you need and has none of the screens you don’t. Say a firm pays per user for a specialist forms service and only uses one form. A form on its own site that writes into case management could replace it, with one fewer login for everybody.
It can also be a false economy. Part of that subscription fee pays someone to keep the service running, fix faults, apply security patches and answer when it breaks. Bring it in-house and those jobs still exist. If the firm has no one to do them, you’re paying a supplier instead, and the fee you cancelled may turn out to have been the better deal.
We build software, so when our figures favour a build you should test them. Ask for the assumptions in writing and take them to someone else for a second price.
Where the money goes wrong
Once the choice is made, these are common ways to waste the money.
Time recording, legal accounts and client ledgers work much the same way at every firm, and the rules around client money are not something to rewrite for the sake of it. A firm that builds its own accounts system takes on years of upkeep for no advantage. The opposite mistake is buying a general product for the part of the work that is yours. A firm known for a particular way of handling, say, high-volume claims can spend years bending its process to fit the product’s screens, and get slower at the work its clients chose it for.
The rest come from how the project runs. Replacing case management, the document system and the accounts link in one go means paying for all of it before anyone sees a benefit. Requirements written in a partners’ meeting describe how the firm believes the work happens, and the fee earners who do the work only find the gap when they test it at the end. And a build with no named owner at the firm, and no arrangement for support and change, will fall behind as rules, staff and the systems next to it change.
How we would decide
We would follow how the work runs before anyone draws an ideal process. Pick a matter type that matters to the firm and trace one from enquiry to closing: who touches it, which system each step happens in, where someone copies information from one screen to another, and where work waits for someone to remember it. Alongside that, list what you pay for and who uses it.
Then go through the options in order of how little they ask of you. If the problem costs less than any fix, leave it. If something you already pay for covers it, turn that on and stop paying for the other thing. If the work is the same as at every other firm, buy. If a product handles the core and the pain sits in the gaps between systems or in a missing report, extend.
Build when three things are all true: the work is specific to how your firm operates, that difference matters to clients or to the firm’s margins, and the figures still come out ahead with the running costs included. Say a firm’s referral partners each send instructions in their own format, where the intake rules are the firm’s own and no product models them well.
We’d expect most firms to end up with a mix: a bought core, a few extensions, fewer subscriptions than they started with, and perhaps one thing built because nothing else fits.
If you do build
Start with the smallest piece that does useful work for a real team, and put it in front of the people who will use it within weeks. Keep the fee earners involved throughout. They’re the ones who will route around the software if it slows them down.
Make sure the firm can leave. Before you sign, find out from the developer what happens to your data if the relationship ends, what documentation you will hold, and whether another developer could pick the system up. You should ask us the same, and we intend our answers to be in the agreement before any work starts.
Agree before launch who at the firm owns the system, how changes are asked for, and who pays for them.
If this is your firm
If a renewal is coming up, or you’re weighing a new system against the one you have, our AI and technology review looks at all four options for your firm and writes down the running costs behind each. It may well tell you not to build anything.