Same Revenue, Different Offer
Picture two legal-tech shops, each doing roughly $6M, each serving AmLaw 200 firms and a handful of corporate legal departments. On paper they look like twins. One gets a term sheet at a multiple that makes the founder do a small dance in the parking lot. The other gets a number that feels like an insult, plus a long earnout. The revenue was identical. The difference was what happened inside the diligence room.
The first firm sells eDiscovery as a managed process: ingestion, processing, hosting, review support, productions — a repeatable workflow that runs the same way whether the matter is a 40-custodian employment dispute or a multi-terabyte second request. The second firm sells "we do litigation support and also help with their Relativity instance and also patch their servers." That second story reads as a bundle of favors a few senior people happen to know how to do. Buyers do not underwrite favors.
The thing that converts a favor into an asset is a documented, defensible workflow — and in eDiscovery, defensibility is not a marketing word, it is the whole game. The Sedona Conference commentary on achieving quality in the e-discovery process makes the point a buyer will be checking against: quality comes from a managed, measurable process, not from a clever tool or a heroic reviewer. If your value lives in three people's heads, you have a staffing agency with good margins, not an eDiscovery platform business.
The Four Questions a Diligence Team Actually Asks
When a buyer's analyst opens your data room, they are not reading your pitch deck. They are quietly answering four questions, and your multiple moves with the answers:
Is this revenue going to repeat? Project-based litigation matters are lumpy by nature — but a firm that can show a base of clients who route every new matter through it, with a predictable per-matter or per-GB economic shape, looks far more like recurring revenue than one chasing one-off productions. Show the cohort: how many corporate clients sent you a second, third, fourth matter, and what that retention curve looks like over 24 months.
Does the work survive the founder leaving? This is where most legal-tech founders get quietly downgraded. If the lead litigation-support partner personally scopes every matter, sets every search-term strategy, and is the named contact on the AmLaw relationships, the buyer is pricing in the risk that those relationships and that judgment walk out the door. Documented intake, scoping templates, QC checklists, and named secondary contacts are not bureaucracy — they are the difference between selling a business and selling yourself into an earnout.
Will the security story survive enterprise procurement? Your clients' clients — the corporations behind the law firms — increasingly run their own vendor security reviews on you. RelativityOne's security and trust documentation is a useful mirror here: it shows the depth of evidence enterprise eDiscovery buyers now expect — access management, encryption, incident response, certifications. If you host privileged and PII-laden data, a buyer needs to see that you can pass the same scrutiny without a fire drill.
If you use AI, can you defend it? Technology-assisted review is old news; generative classification and summarization are the new pressure point. A buyer will ask exactly where AI touches privileged material and what the human review checkpoints are. The NIST AI Risk Management Framework gives you the vocabulary to answer cleanly — mapping, measuring, and governing where the model operates and where a human owns the decision. "We use AI to speed up review" is a liability. "Here is the boundary, here is the human checkpoint, here is how we govern it" is a premium.
If you have reusable workflow logic, search-term libraries, or classification models that travel across matters, that IP is worth pricing deliberately — see valuing proprietary data assets in tech acquisitions.
What To Build Before You Take a Call
You do not need a year. You need to convert what you already do into evidence a stranger can verify. Start with three things you can assemble in a quarter.
A matter-flow runbook. One document that walks a new analyst from data receipt to production: ingestion steps, processing settings, QC gates, privilege-screen checkpoints, escalation rules. If a competent operator could run a typical matter from this document without calling the founder, you have proven the workflow is the asset.
A security evidence room. Not a promise — a folder. Access controls and who has them, encryption at rest and in transit, incident-response plan with a real runbook, data-retention and disposal policy, and any audit reports. Because legal data is some of the most sensitive a vendor handles, the controls have to exist before any data touches an AI-assisted step, not after; the CISA best practices for securing data used to train and operate AI systems are a sensible checklist to grade yourself against.
A retention exhibit. A simple cohort view: clients by start date, matters per client over time, revenue per client. This single chart often moves the multiple more than any deck slide, because it answers the repeatability question with data instead of assertion.
Build those three, and the conversation shifts from "what do you do" to "what is it worth." If you want a second set of eyes on which gaps a buyer will hit first — and a plan to close them before you go to market — talk through the plan with us.
