Business Club Varna2 years, 140+ members, zero ad spend
The 40 callers nobody rang back: 5 to 12 signed injury cases in 6 months
Rourke Injury Law in Wilmington, North Carolina was spending $3,000 a month on Google Ads and signing about one case a month from it, and in October 2025 Marcus Rourke emailed us to ask whether the ads had died. A year on, the firm signs 2 cases a month from the same campaigns on $2,250 a month. The Google Ads report for the first half of 2025 had counted 100 enquiries; when we matched them against the office phone log, 40 of those people had called the firm and never heard from anyone.
About the client
Rourke Injury Law is a 5-person personal injury practice off Market Street in Wilmington, North Carolina, opened by Marcus Rourke in 2017: car accidents, slip-and-fall and workplace injuries on contingency. Google Ads had run since 2020 at $3,000 a month paid to Google (every spend figure below means the same), set up by a freelancer and left alone, producing about 17 enquiries a month and 5 signed cases in the first half of 2025. Marcus did not know there was a problem with the phone; nobody had told him what happened to a call that rang out, and he had never asked.
Personal injury law
Wilmington, North Carolina
5 people
2017
Our analysis of the situation
The ad account was clean: 6 months of local search terms, a 5.2% click-through rate, cost per enquiry at $180, conversion tracking that fired on test, and one medical-malpractice ad group spending about $750 a month on cases the firm refers out, which did not explain 5 signed cases from 75 car-accident enquiries. The website was sound as well: 2.4 seconds on a phone, a working click-to-call, every form entry matched in the intake@ inbox, and once CallRail went on the lines the recordings showed callers who had crashed within the week, not the junk Marcus suspected.
What CallRail showed first was the answer rate, so we matched Google's 80 time-stamped call conversions for January to June 2025 against the carrier's log for the office line: 30 answered live, 50 to voicemail, and of those 50, 40 never called back. The 20 web forms waited a median of 6 hours 40 minutes in a shared inbox nobody owned.
That set the 4 numbers to work on: the live-answer rate, the callback rate, the median first reply on forms and the enquiry-to-signed rate. Cost per enquiry we left alone; the work was to rebuild what happens in the first 10 minutes after someone calls or writes.
Supporting research
Clio's Legal Trends Report 2024 puts the share of firms answering an incoming call at 40% and the share of missed calls that get a callback at 20%; Rourke was at 38% and 20% [1]. The same report puts a lead answered within 5 minutes at roughly 8 times the conversion of one answered after 30. Hennessey Digital's 2024 study of about 1,400 firms found Charlotte the slowest city it measured, with a median form response of 106 minutes; Rourke's forms waited nearly 4 times that [2].
Our strategy to solve the problem
The first decision was what to do about Ruby, the after-hours receptionist service the firm had paid for since 2023. We could rewrite its script or cancel it. Ruby answered the phone, and everything around it was broken. The script Marcus had handed over in 2023 was 4 lines long and said the firm "takes all injury cases", which it does not, and 2 matters had been mishandled on the strength of it; the call summaries went unread. A better script would have fixed the wording and left the summaries unread. We cancelled it in December 2025 and replaced it with an in-house rotation, which is the first thing we needed from the client: a named person on evenings and weekends who would take the callback bookings, because no software can make a firm willing to answer.
The second was where an enquiry should live. The cheap option was to keep the intake@ inbox and add a forwarding rule so both intake staff saw every form. That would have doubled the number of people who could reply and left the question of whose job it was exactly where it had been. On 8 December 2025 the inbox and the sticky notes were replaced with a Lawmatics pipeline in which every enquiry is a card with an owner and a timestamp, and a Lawmatics automation emails and texts any lead whose card is not marked "contacted" within 4 business hours and reminds the owner. The median first reply on forms fell from 6 hours 40 to 11 minutes, which put the delay down to ownership rather than capacity.
The third was the missed call itself, and this is where the result came from. The firm's phone vendor had quoted a call queue with hold music and overflow to mobiles, which would have kept people waiting on the line and still lost the ones who hung up. We built a text-back instead. A call on the CallRail number that is not answered within 4 rings fires a Zapier step into Lawmatics, which texts the caller within 2 minutes with a link to book a callback slot and pings whichever staffer is on rotation. The text says nothing about the claim; it schedules.
Because an automated text to a caller needs prior express consent under the TCPA, we needed a one-line consent disclosure from the firm on the CallRail greeting and beside the Fluent Forms phone field, and Marcus and the associate checked that wording, the text and the chat flow against North Carolina Rule of Professional Conduct 7.1 so that none of it said anything about the firm's results. That review pushed the launch from 22 December to 5 January 2026. The callback rate on missed calls went from 20% to 91%.
The fourth was what to do after hours once Ruby was gone: leave it to voicemail and the text-back, or add a chat. We added Tidio's widget to the WordPress site on 12 January 2026. In office hours a chat goes to whichever staffer is free; after hours a Tidio flow asks what kind of accident, when, and whether the person has been treated, then books a next-business-day callback. The questions came from a written intake script we asked Marcus and the associate for, so that the triage matched how the firm screens a case. The chat does not assess anything.
The last decision was the budget. We could have kept the $3,000 and let the new system catch more of it. Instead we paused the malpractice ad group on 29 December 2025, which took the spend to $2,250 a month from January. Nothing else in the account changed.
Before and after
6 months of the new intake system, January to June 2026, against the same 6 months a year earlier.
| Metric | Before (Jan–Jun 2025) | After (Jan–Jun 2026) | Change |
|---|---|---|---|
| Monthly ad spend | $3,000 | $2,250 | −25% |
| Ad spend, 6 months | $18,000 | $13,500 | −$4,500 |
| Enquiries (calls + forms) | 100 (80 calls, 20 forms) | 75 (60 calls, 15 forms) | −25% |
| Cost per enquiry | $180 | $180 | unchanged |
| Calls answered live | 38% (30 of 80) | 82% (49 of 60) | +44 pts |
| Missed calls returned | 20% (10 of 50) | 91% (10 of 11) | +71 pts |
| Median first reply, web forms | 6 hrs 40 min | 11 min | −98% |
| Enquiry-to-signed rate | 5.0% | 16.0% | +11 pts |
| Signed cases | 5 | 12 | 2.4× |
| Cost per signed case | $3,600 | $1,125 | −69% |
Cost per enquiry did not move; it did not need to. The campaigns were left alone apart from one paused ad group, and the flat row is what leaving them alone looks like. Case values and outcomes are not part of this study and are not implied by it.
Day to day, the before column meant about 17 enquiries a month, of which 7 or 8 went to voicemail and were never returned, while Marcus read the monthly Google Ads report and concluded the leads were bad. The after column means 12 or 13 enquiries a month, nearly all of them spoken to within minutes, and 2 signed cases a month from an account that is costing $750 less.
Summary
12 cases instead of 5 came from the same campaigns because a caller who now gets a text within 2 minutes, or a form that now has an owner within 11, is a person the firm can still sign.
Before that, half of the calls the ads produced went to voicemail and 4 in 5 of those were never returned, and the forms waited most of a working day in a shared inbox.
We found this by ruling out the ads and the site, then matching Google's time-stamped call conversions against the office phone log.
The fix was a Lawmatics pipeline with an owner on every card, a CallRail-to-Lawmatics text-back on any unanswered call, a Tidio flow after hours, an in-house evening rotation in place of a receptionist service, and one paused ad group that took the spend from $3,000 to $2,250 a month.
References
- Clio, Legal Trends Report 2024. clio.com
- Hennessey Digital, lead-form response time study of about 1,400 law firms (June 2024). hennessey.com
Final word from the client
The number was in the phone bill the whole time. I read that bill every month. I never once looked at the calls.