How do I track which marketing channel produced a criminal defense case?

Give each channel its own tracked phone number, log every call with a timestamp and a source, and write the matter number back onto the call record once the case is signed. That last step is the one most firms skip, and without it you have a call report and a revenue report that never meet. Because the fee is known at signing, the join gives you real cost per signed case rather than an estimate.

Should a criminal defense firm use call tracking or form tracking?

Both, but weight them correctly. The phone carries the urgent inquiries in this practice area and the form carries the researchers, so a report built around form submissions will describe the smaller and slower half of your demand. Treat the call as the primary conversion and the form as a secondary one, and make sure a click on a phone link is not being counted as a conversation.

The first caller is usually a family member, not the defendant. Does that break attribution?

Only if your model is keyed to a person. Attribute to the matter rather than to the individual who called, and record the caller's relationship to the defendant as a field on the inquiry. Identity-based and device-based attribution genuinely fail here, because the person who searched and the person who signs the engagement letter are frequently different people on different devices.

How do I calculate cost per signed case for a flat fee criminal practice?

Divide the month's marketing spend by the matters signed in that month, segmented by charge class. Flat fees make this unusually clean: the revenue side resolves within days, so you are dividing two known numbers instead of forecasting one of them. Segment first, because a channel producing many small misdemeanours will look cheaper than a channel producing a handful of felonies.

Do missed after-hours calls count as lost leads?

They count as lost cases, which is worse. A family calling at eleven at night moves to the next firm within the hour, so an unanswered call is not a delayed inquiry that can be recovered on Monday. It also never enters your attribution data, which means a channel can be performing well and reporting badly purely because its calls arrive when nobody picks up.

Is last-click attribution good enough for criminal defense?

It is a more defensible default here than in most practice areas, because the decision window is usually hours rather than weeks and the models converge when the path is short. It still under-credits the research session that happened the night before, often on a different device. Run last click as the operating number and review a multi-touch view quarterly to see how much it is hiding.

Part of: Marketing attribution

Criminal Defense Marketing Attribution

By Mohammad Kashif, Chief Technology OfficerLast updated September 21, 2026

Criminal defense is the easiest practice area to attribute properly, if you capture calls. Inquiries arrive by phone, at night, often from a relative, and flat fees collected up front make revenue per case knowable within days.

A flat fee is known the day it is quoted, so the numerator and the denominator land in the same month. As an illustration rather than a result to expect: divide a Starter plan at $599 a month by four signed matters and the marketing cost is about $150 per case, calculable before the first hearing rather than after a settlement.

Why criminal defense inquiries never show up in a form fill report

Standard marketing reporting for a defense practice counts sessions, keyword positions and form submissions. In criminal defense the form is the smallest and slowest of the inquiry channels. The arrest already happened. Somebody is in custody or has just posted bond, and the person trying to fix it is dialling numbers, not filling in fields.

That mismatch means a report can be entirely accurate and still describe a minority of the demand. Two channels can produce identical form counts while one of them drives several times the call volume, and a default analytics view will not say so. Event tracking in Google Analytics 4 gets closer by recording clicks on the phone link, but a click on a tel link is not the same event as a conversation that actually happened, and on desktop it is frequently neither.

The correction is structural rather than clever. Make the call the primary conversion, demote the form to a secondary one, and require that every call carries a source when it lands in the log.

A distinct tracked number per channel, so organic, paid, directory listings and referral pages stay separable at the call log rather than merging into one office line.
Ring duration recorded, not just call count. A call that rang out is a lost case, and lost cases appear nowhere in a conversion report.
Timestamp on every call, kept at hour granularity. The night and weekend share of inquiries is the number that justifies answering coverage.
A repeat-caller flag, so one family calling back three times is one inquiry rather than three.
The eventual matter number written back onto the call record, because that link is the only thing that turns a call into attributed revenue.

How do you attribute a call that came in at 11pm from the client's mother?

The caller is frequently not the client. A parent, spouse or sibling makes the first contact, and the person whose name ends up on the engagement letter may never have touched your website. Any model keyed to a person, a device or a logged-in identity loses that case entirely. Attribute the inquiry to the matter, not to the individual who dialled.

The after-hours problem is larger than the identity problem. A defense inquiry that arrives at eleven at night does not sit in a queue until Monday, because the family is calling the next firm on the list within the hour. Whether the call was picked up is therefore the single biggest variable in the whole measurement, and it sits upstream of every channel comparison. A missed call does not become a lead with a bad source. It becomes nothing, and nothing is invisible to attribution.

Self-reported source still helps, but treat it as corroboration rather than ground truth. Ask how they found you, record the answer verbatim, and reconcile it against the tracked number. A caller whose son was arrested four hours ago is not a reliable narrator of their own search history, and the two signals disagreeing is itself useful information about which channels get remembered.

Attribute to the matter, not the caller, and record the caller's relationship to the defendant.
Log calls that arrive outside business hours in the same place as the rest, so the answered rate is comparable by hour.
Track answered, missed and voicemail separately. Rolling them into one call count hides the expensive category.
Keep the self-reported source as a second field alongside the tracked source, and review the disagreements monthly.

What flat fees let you measure that contingency practices cannot

Most of the difficulty in legal marketing attribution is lag. A personal injury firm cannot tell you what a signed case was worth until it resolves, which can be two years, so cost per signed case is a forecast held together by assumptions. A flat fee practice does not have that problem. The fee is quoted at the consultation and usually collected before the arraignment, so revenue per case is a fact within days of the inquiry.

That makes the arithmetic honest rather than modelled. Whatever the monthly marketing line is, a Starter plan at $599 a month or an agency retainer in the $4,000–$10,000 a month range, the denominator is fixed and the numerator is knowable the same month. Divide and you have a real cost per signed case, not a projection with a confidence interval attached to it.

The concession: fee variation will lie to you if you average it. A misdemeanour possession and a felony assault are both one signed case, and a channel producing ten of the first is not outperforming a channel producing three of the second. Segment by charge class before comparing anything, or the cheapest channel will always be the one bringing in the smallest matters.

Record the quoted fee, the collected fee and the charge class on every signed matter.
Report cost per signed case segmented by charge class, and only then in aggregate.
Count declined and referred-out inquiries separately. A channel with a high decline rate is a targeting problem, not an intake problem.
Track collection, not just signature. A flat fee signed on a payment plan and never completed is not revenue and should not be attributed as such.

How to measure a decision window that closes in hours, not weeks

In most practice areas the choice between last-click and multi-touch attribution changes the answer substantially, because the buyer takes weeks and touches five things. In criminal defense the window is usually hours, so the models converge and the argument matters less than it does elsewhere. Last click is a defensible default here in a way it is not for an estate planning practice.

It is not zero, though, and the honest version of this point concedes the gap. Families often search the night before the arraignment, read two or three firm pages, sleep on it and call in the morning from a different device. That first session is real and last click will hand its credit to a branded search. The practical compromise is to run last click as the operating number and review a multi-touch view quarterly to see how much it is hiding, rather than running two models in parallel and arguing about them weekly.

Cadence matters more than model choice. A broken tracked number costs a contingency practice some reporting noise; it costs a defense practice a month of untraceable signed matters that have already been billed and closed. Check that every published number rings and logs at the start of each week, which takes about five minutes and is the highest-return habit in this whole exercise.

What criminal defense attribution still cannot tell you

The measurable share of this practice area is large but not total, and claiming otherwise is how attribution reporting loses credibility internally. A meaningful volume of defense work arrives through channels that leave no digital trace at all, and those matters will show up as direct or unattributed no matter how carefully the tracking is built.

The right response is to size the unmeasurable bucket and hold it constant rather than to quietly redistribute it across the channels you can see. Ask the source question at intake for every matter, tally the referral answers, and report them as their own line. A firm that knows a third of its work comes from bondsmen and prior clients can read the remaining two thirds accurately. A firm that silently allocates that third to organic search cannot read anything.

Referrals from bail bondsmen, prior clients and other attorneys, which are a phone call with no preceding search.
Jail and courthouse word of mouth, including recommendations passed between defendants.
Public defender conflict referrals and appointed-counsel overflow.
Calls to a mobile number a partner published years ago on a profile you no longer control.

What a flat fee practice can know, and when, compared with contingency work

	Criminal defense, flat fee	Contingency practice
Revenue per case known	At the fee quote, before work begins	At settlement or verdict, months or years later
Decision window	Hours to a few days	Weeks to months
First contact channel	Phone, weighted to nights and weekends	Phone and web form, mostly business hours
Who makes first contact	Often a parent, spouse or sibling	Usually the injured person
What breaks attribution	Calls that ring out or arrive untagged	The lag between first touch and revenue
Cost per signed case	Calculable within the same month	Calculable only once matters resolve
Safe default model	Last click, reviewed quarterly	Multi-touch, because the path is long

Common questions

How do I track which marketing channel produced a criminal defense case?: Give each channel its own tracked phone number, log every call with a timestamp and a source, and write the matter number back onto the call record once the case is signed. That last step is the one most firms skip, and without it you have a call report and a revenue report that never meet. Because the fee is known at signing, the join gives you real cost per signed case rather than an estimate.
Should a criminal defense firm use call tracking or form tracking?: Both, but weight them correctly. The phone carries the urgent inquiries in this practice area and the form carries the researchers, so a report built around form submissions will describe the smaller and slower half of your demand. Treat the call as the primary conversion and the form as a secondary one, and make sure a click on a phone link is not being counted as a conversation.
The first caller is usually a family member, not the defendant. Does that break attribution?: Only if your model is keyed to a person. Attribute to the matter rather than to the individual who called, and record the caller's relationship to the defendant as a field on the inquiry. Identity-based and device-based attribution genuinely fail here, because the person who searched and the person who signs the engagement letter are frequently different people on different devices.
How do I calculate cost per signed case for a flat fee criminal practice?: Divide the month's marketing spend by the matters signed in that month, segmented by charge class. Flat fees make this unusually clean: the revenue side resolves within days, so you are dividing two known numbers instead of forecasting one of them. Segment first, because a channel producing many small misdemeanours will look cheaper than a channel producing a handful of felonies.
Do missed after-hours calls count as lost leads?: They count as lost cases, which is worse. A family calling at eleven at night moves to the next firm within the hour, so an unanswered call is not a delayed inquiry that can be recovered on Monday. It also never enters your attribution data, which means a channel can be performing well and reporting badly purely because its calls arrive when nobody picks up.
Is last-click attribution good enough for criminal defense?: It is a more defensible default here than in most practice areas, because the decision window is usually hours rather than weeks and the models converge when the path is short. It still under-credits the research session that happened the night before, often on a different device. Run last click as the operating number and review a multi-touch view quarterly to see how much it is hiding.

Find out what your signed cases actually cost

We trace your last quarter of signed cases back to the channel that produced them, and show you the cost per case next to each one.

Get an attribution audit