What custom software actually costs a law firm

Ask what custom software costs a law firm and you will mostly get a shrug and a discovery call. The reason is not that the answer is unknowable. It is that publishing a number costs a vendor negotiating room, and almost nobody is willing to give that up.

So here are ours, and what actually moves them.

What custom legal software costs, in ranges

One workflow, built properly: $60,000 to $90,000. A workflow means a complete path — records arriving through to a cited chronology a paralegal can check, or an inquiry arriving through to a signed retainer. $75,000 is the common middle. Six weeks is a realistic shape, provided the scope stays at one workflow.

A narrow piece: around $6,500. A site and intake rebuild is not a workflow build; it is a much smaller job with a much smaller number, and for a firm whose problem is that inquiries go unanswered it is frequently the only thing that needs doing.

Assessment before either: $7,500 for a week. Whether this is worth paying for depends on whether you get something at the end. Ours ends with a working prototype on your own documents and a fixed quote. A week that ends in a slide deck is worth considerably less.

Keeping it working: 15 to 25 percent of the build per year. For a $75,000 build that is roughly $11,000 to $19,000 annually. This is the published benchmark for software maintenance generally and it is not specific to legal work.

That last one is the number most often missing from a quote, and its absence is informative. Software that reads documents does not fail cleanly — it degrades. The model underneath gets replaced by the vendor, a records provider changes an export format, and accuracy slides without anything visibly breaking.

What actually moves the number

Not the interesting parts. Three things, and they are all unglamorous.

Integration. Whether your case management system has an API you can write to, or whether it has a login and a support address. This single fact can move a project by weeks, and it is knowable in an afternoon. Any estimate produced without checking it is a guess wearing a suit.

Exceptions. Anyone can handle the clean record set. The cost is in the illegible one, the one with pages missing, the one naming a provider nobody recognizes. A quote built on the happy path is always lower and always moves.

How many people have to agree. A three-attorney firm decides. A forty-attorney firm decides, then revisits, then decides again with a different administrator in the room. This is not a criticism, but it is real time and someone pays for it.

Notably absent from that list: how clever the software is. The technology is rarely what makes these expensive.

The three-year number

A build quote is not what the thing costs. Over three years, a $75,000 build with support at the lower end of the benchmark runs to roughly $109,000. At the upper end, closer to $131,000. Add the assessment and you are between $116,000 and $139,000 for three years of a working system you own outright.

That is the number to compare against alternatives, and it is the comparison most firms never run — because the alternative is priced in a different unit.

A vendor charging per case at a few hundred dollars, across a caseload of a few hundred a year, reaches the same territory inside three years and leaves you owning nothing. Two paralegals whose week is substantially records work cost more than that annually, and are also doing something else you would rather they did.

None of which means building is right. It means the comparison is winnable in either direction and is worth an hour with a spreadsheet.

If your annual volume multiplied by the hours saved per case does not clear a build comfortably, buy something off the shelf. We will tell you that, and it costs us the engagement.

When you should not build

Genuinely, most firms should not.

Buy if a product fits. If an off-the-shelf tool does the job without you reshaping the firm around it, buy it. It is cheaper, it is supported by a company whose entire business is that product, and it improves without you paying for it again.

Do not build for a process you are about to change. Automating a workflow you are already unhappy with produces a faster version of something wrong.

Do not build without volume. These are fixed costs recovered over cases. A firm running a modest caseload usually cannot get there, and no amount of enthusiasm changes the arithmetic.

The case for building is narrower than vendors suggest, and it is specific: the value is in connecting systems you already run, and that is the thing no product company can sell you, because it is different in every building.

How to read a quote

Four questions, and the answers tell you more than the number does.

1. What happens to the exception? If the answer is about the happy path, the estimate is optimistic.
2. Where does it run, and whose keys? Your cloud tenant and your API keys, or theirs. This decides what happens if the relationship ends badly.
3. What is the ongoing number? If there is not one, ask why not.
4. What do I own at the end? Source code, in your repository, under a license permitting unrestricted use. Anything less and you are renting.

A vendor who answers those four cleanly is not necessarily the cheapest. But a vendor who cannot is expensive in a way that will not appear on the invoice until year two.