Cost per lead is the most popular number in law firm marketing and one of the least useful. It rewards whoever produces the most contacts, whether or not anyone signs. A firm that manages to cost per lead will reliably buy cheap leads from channels that never close, and then wonder why the partners don’t feel the growth. Cost per signed case fixes that, but only if it’s calculated honestly: from your CRM, with its basis stated, after cohorts mature, and never on two cases.
The definition, in one sentence
Cost per signed case is the money spent on a channel or campaign in a period, divided by the signed cases that can be credited to it, where a signed case is an engagement agreement executed by the client and recorded in the firm’s CRM. For non-injury practices, read “signed case” as “retained client”.
That definition does three things cost per lead doesn’t. It counts the outcome the firm is paid for. It uses the firm’s own system as the source of truth, not the ad platform. And it forces a decision about what “credited” means, which is where most reports quietly cheat.
Few firms have it at all. In one survey of more than 4,800 firms, only 37% had systems to track marketing ROI (Best Law Firms), and 61% of respondents in a 2026 vendor survey had less than moderate confidence measuring ROI across channels (Martindale-Avvo). Agencies rarely publish it either: only 6% of 50 law firm marketing case studies reviewed in 2026 disclosed a cost per signed case (Silver Atlas).
Why cost per lead misleads: a worked example
Two channels, same budget, same month, same firm.
Worked example: a fictional injury firm
IllustrativeThe channel with the cheapest leads produced the most expensive cases
| Channel A | Channel B | |
|---|---|---|
| Media spend | $20,000 | $20,000 |
| Leads | 200 | 80 |
| Cost per lead | $100 | $250 |
| Qualified leads | 30 (15% of leads) | 40 (50% of leads) |
| Signed cases | 8 | 16 |
| Cost per signed case | $2,500 | $1,250 |
Managed by cost per lead, this firm would move budget from B to A, and sign fewer cases for the same money. The pattern isn’t hypothetical in kind. In one 2026 lawsuit, a PI firm alleges it rejected 204 of 265 leads from a vendor (Rankings.io, secondary, allegations unproven). Junk leads are cheap per lead and ruinous per case.
The formulas
Under The Probative Standard, four cost metrics sit side by side:
| Metric | Formula | How to use it |
|---|---|---|
| Cost per lead | Media ÷ leads | A diagnostic only, never a headline |
| Cost per qualified lead | Media ÷ leads intake accepted as potential cases | The first cost to manage to, because it arrives fast |
| Media cost per signed case | Media for the cut ÷ signed cases credited to the cut | For any cut below the account: channel, campaign, practice, language |
| All-in cost per signed case | (Media + agency fees + measurement and intake tooling) ÷ signed cases from paid channels | Account level only; the number partners should see |
What counts as a lead matters as much as the formula. A lead should be a real conversation (a call with at least 30 seconds of connected talk and a caller’s voice, or a form, chat or text with working contact details) about a new matter, in a practice and area you take, not spam, a job seeker or a vendor. Set your ad platforms’ minimum call length to the same threshold, so every count starts from the same calls (Google).
The two bases: cohort and signing date
Every cost per signed case needs a label saying which of two bases it uses, because they answer different questions.
- Cohort basis (for decisions). Signed cases from leads created in the period, divided by the period’s cost. It ties each case to the spend that produced it. It is provisional until the cohort matures.
- Signing-date basis (for reconciliation). Cases signed in the period, whatever month their lead arrived. This is what your CRM counts, so it is how you check the report against the firm’s books.
A rolling 12-week figure on the signing-date basis makes a steady headline. Budget decisions should use the cohort basis, because that is the only way to know what this month’s spend bought.
Maturity: why recent months always look expensive
Cases sign late. An injury lead from the last week of the month may sign in three weeks; a complex case in three months. So a cohort’s cost per signed case falls as it ages, and a report that shows last month’s cohort as final is showing a number that will improve.
Worked example: one month's cohort as it ages
IllustrativeThe same $20,000 cohort looks more than 80% more expensive at two weeks than at twelve
| Cohort age | Share of eventual signings landed (typical) | Signed so far | Cost per signed case so far | Label |
|---|---|---|---|---|
| 2 weeks | 55% | 6 | $3,333 | Provisional |
| 4 weeks | 75% | 8 | $2,500 | Provisional |
| 8 weeks | 90% | 10 | $2,000 | Mature |
| 12 weeks | 100% | 11 | $1,818 | Mature |
Three rules keep maturity from fooling anyone:
- Show the label. “Provisional” below 90% of usual signings; “mature” at 90% or more, using your own trailing 12-month lag curve.
- Restate, never hide. As late cases land, past months update. A monthly close restates the trailing 12 weeks; a quarter freezes at its review, and later changes appear in a restatement table.
- Don’t pad public numbers. Lag-adjusted estimates are useful internally for forecasting. They never belong in published results.
The 5-case rule
Small numbers swing. With 2 signed cases, one more or one fewer moves cost per signed case by a third or a half. So: no cost per signed case for any cut with fewer than 5 credited signed cases in the period. Show the spend and the counts instead. A report that says “too early to tell” is more useful than one that celebrates a $900 cost per case built on two signings.
Credit: which channel gets the case
A cost per signed case is only as honest as its crediting rule. A sound default, and the one the ad platforms bid on: credit the last paid click or tracked paid call in the 30 days before the lead’s first qualifying contact (Google’s default click-through window is 30 days, Google). If there was none, credit the last non-paid tracked source (organic search, direct, a referring site, an AI assistant). Failing that, the self-reported source. Failing that, “source unknown”, which stays on its own line and is never spread across channels.
Three habits protect the number:
- One credit per case. Channel totals plus “source unknown” equal total signed cases.
- Brand search on its own line, with a note that brand demand is partly created elsewhere.
- TV, radio and billboards never get click-based credit; they show up through their own tracking numbers, the self-reported source and geo tests.
Attributed is not caused
A channel can have a low cost per signed case because it caused cases, or because it sat in front of people who would have signed anyway: a referral who searched the firm’s name, a TV viewer who clicked an ad. The only way to tell is a test: switch a channel off or step it down in some geographic areas and not others, then compare signed cases. Meta’s open-source GeoLift guidance suggests a year of history and tests of several weeks (GeoLift). Until a test exists, the right word is “attributed”, and every channel line should show its test status.
What can you afford per signed case?
Work backward from what a case is worth. The average auto bodily-injury liability claim was $28,278 in 2024 (Insurance Information Institute). At a one-third fee, that is about $9,400 of fee before costs. A firm willing to spend 25% to 40% of expected net fee on acquiring the case could pay roughly $2,350 to $3,760 per signed case at that average, and more for higher-value case types (our arithmetic, an estimate, not a benchmark). One agency dataset puts actual PI cost per signed case at $2,500 to $12,000 (Foundry CRO).
Your own math is better than any benchmark: average net fee by case type, from your resolved cases over the last two to three years.
The tracker: what to record each week
Free template layout
IllustrativeThe cost-per-signed-case tracker: one row per channel, language and week
| Week of | Channel | Language | Media | Leads | Qualified | Signed (cohort, to date) | Cohort maturity | Media cost per signed case | Signed (signing date) | Test status |
|---|---|---|---|---|---|---|---|---|---|---|
| Sep 7 | Google search, non-brand | English | $6,200 | 41 | 18 | 6 | Mature | $1,033 | 5 | Not tested |
| Sep 7 | Google search, non-brand | Spanish | $1,900 | 14 | 7 | 2 | Mature | Under 5 cases | 3 | Not tested |
| Sep 7 | Local Services | English | $3,100 | 22 | 9 | 3 | Mature | Under 5 cases | 2 | Not tested |
| Sep 7 | Brand search | Both | $900 | 19 | 11 | 5 | Mature | $180, see note | 4 | Holdout planned |
| Oct 12 | Google search, non-brand | English | $6,400 | 44 | 17 | 2 | Provisional | Provisional | 6 | Not tested |
Two details make the tracker trustworthy. First, the brand line carries a note every week: people searching your name may have found you elsewhere. Second, the “under 5 cases” cells are left blank on purpose. Pool weeks into months or quarters to reach 5.
Common mistakes
- Dividing by leads and calling it cost per case.
- Counting platform-reported conversions instead of CRM signings. Platforms count clicks that converted on their terms; your CRM counts clients.
- Blending bases, such as this month’s spend over this month’s signings when most signings came from last month’s leads, and then calling it a cohort figure.
- Spreading “source unknown” across channels in proportion to their credited cases. It hides how much you don’t know.
- Excluding fees at the account level. The partners pay the agency too.
- Celebrating immature or tiny numbers.
What this means for your firm
- Pick one definition of a signed case and write it down, including how you count multiple clients from one incident.
- Make intake record the source and the stage on every lead in the CRM, with a reason code for every decline.
- Ask for both bases, labeled, and for maturity labels on recent cohorts.
- Refuse figures on fewer than 5 cases.
- Ask what’s been tested. If nothing has, every channel’s number is attributed, not proven.
Probative is AI-first SEO and ads for law firms, and cost per signed case is the one number in every Monday report. See how we report, the ILLUSTRATIVE sample report, and signed-case reporting. If you already have an agency, Independent Measurement checks its numbers against your CRM.
Marketing information, not legal or accounting advice. What client information leaves the firm is your responsible attorney’s decision.