Amari Law Firm was getting calls. What nobody could tell you was which ads produced them, or which of them turned into clients. Fixing the measurement — before touching the campaigns — is what changed the numbers.
A Birmingham criminal defense firm had steady call volume and no way to connect it to spend. The conversion tracking was broken in three separate places, so the account had been optimizing against bad data for months. I rebuilt the measurement first, then restructured the campaigns around the searches that actually produced signed cases. Over the following month calls rose 33.9%, forms 33.3%, and cost per call fell 21% — on a budget increase of $300.
The situationSteady Volume, Zero Visibility
When Johnny brought me in, the firm wasn’t starving for leads. The phone rang. Forms came in. On the surface the account looked like it was doing its job.
The problem was that nobody could answer the only question that matters: which of this is producing clients, and what is each one costing?
Three things were in the way.
| Problem | What it meant in practice |
|---|---|
| No line from click to signed case | Ad spend and case outcomes lived in separate systems that never spoke. Budget decisions were being made on instinct. |
| Broken GA4 and Tag Manager setup | Phone call conversions misfiring, form submissions not registering, duplicate events inflating the numbers that did come through. |
| Leads that didn’t become cases | Call and form volume held steady while the sign rate didn’t follow. Volume was being mistaken for performance. |
An account optimizing against broken conversion data isn’t neutral. It’s actively learning the wrong lesson, every day, with your money.
That last point is the one most firms underestimate. Google’s Smart Bidding chases whatever you tell it a conversion is. If duplicate events are inflating some campaigns and missed events are hiding others, the algorithm doesn’t sit still — it spends more where the data looks good, which is not the same as where the cases are.
What I foundThree Tracking Failures, Not One
The audit came before any campaign changes. This is what turned up.
The most common way a firm arrives at the wrong conclusion. Calls were the primary way clients reached the firm and the primary thing not being counted correctly.
Real inquiries arriving in the inbox with no corresponding conversion in the account. From Google’s point of view, the campaigns producing them were underperforming.
Single actions counted more than once, which made certain campaigns look stronger than they were and pulled budget toward them.
Not a uniform error that washes out in the averages. The distortion ran in opposite directions across the account, which is why the reported numbers looked plausible enough that nobody had questioned them.
The workMeasurement First, Then Campaigns
I ran this over three months in a deliberate order. Every phase had to hold before the next one started.
Phase 1 — Rebuild the measurement
GA4 and Tag Manager were rebuilt from the ground up rather than patched. Call conversions were re-implemented, form events corrected, duplicate triggers eliminated. Only after the numbers reconciled did I touch a campaign.
This phase produces nothing a client can see, which is exactly why it gets skipped. It’s also the reason everything after it worked.
Phase 2 — Rebuild the account around intent
I mapped the competitive landscape across Birmingham and the surrounding markets — competitor ad copy, keyword intent, how people in this practice area actually search — and restructured the account around it. Campaigns were separated by service so that budget could follow case type rather than being averaged across everything.
Search term data was then used to cut spend on searches that never produced clients: out-of-area queries, case types the firm doesn’t take, and informational searches from people who were never going to hire anyone.
Phase 3 — Feed real outcomes back in
With clean tracking in place, the firm could see which ad groups and which spend levels sat behind actual signed cases, across paid, organic and referral. That’s what allowed the campaigns to be optimized toward client acquisition instead of raw inquiry volume.
The resultsWhat Changed, Month Over Month
September is the first month with trustworthy data — the baseline after the tracking rebuild. October is the following month.
| Metric | September | October | Change |
|---|---|---|---|
| Ad spend | $5,500 | $5,800 | +$300 |
| Phone calls | 62 | 83 | +33.9% |
| Form fills | 15 | 20 | +33.3% |
| Total conversions | 77 | 103 | +33.8% |
| Cost per call | $88 | $70 | −21.2% |
| Cost per conversion, blended | $71 | $56 | −21.2% |
| Impression share | 34% | 45% | +11 pts |
| Conversion rate | 17% | 21% | +4 pts |
The September return, measured against signed case value in that month:
Signed case value … $30,000
Return ……………….. $30,000 ÷ $5,500 = 5.4x
The efficiency gain matters more than the volume gain. Getting 33% more conversions by spending 33% more is arithmetic. Getting 33% more conversions on a 5% budget increase is the campaign working better — and that came from impression share rising while cost per call fell, which is the combination that indicates ad relevance improved rather than bids simply going up.
Straight talkWhat These Numbers Do and Don’t Prove
If you’ve read a lot of agency case studies, you’ve noticed they never include a section like this. Here’s mine, because you should be skeptical of the ones that don’t have one.
- This is two months of data from one firm. Real, verified, and not a large enough sample to call a trend. Criminal defense has seasonal and event-driven swings that a two-month window won’t separate from campaign performance.
- There’s no control group. I can’t prove with certainty that no portion of the lift would have happened anyway. What I can say is that the tracking rebuild preceded the campaign changes, and the changes were specific and documented.
- Signed case value is not collected revenue. Criminal defense cases sign, then get paid over time, and some engagements don’t collect in full. The 5.4x figure reflects the value of matters signed in that month, not cash in the bank that month.
- Some attribution is permanently unrecoverable. People search on one device and call from another, or call a number they saved days earlier. A share of every firm’s cases can’t be traced, and any consultant claiming otherwise is overselling.
None of that undercuts the result. It’s context, and I’d rather you have it before a call than discover it after.
What it tookThe Part Nobody Advertises
Two things were required from the firm, and both were small but non-negotiable.
Intake had to tag consistently. Somebody at the firm had to log where each inquiry came from, in a fixed set of categories, every time. Without that, the outcome data never connects back to the ad data and the whole system stops working. This is maybe fifteen seconds per call.
The account had to be left alone during relearning. After structural changes, Smart Bidding needs an uninterrupted period to recalibrate. Touching it mid-cycle resets the clock. That means a stretch where the numbers look uncertain and the correct action is to do nothing — which is harder than it sounds when it’s your money.
If you’re not willing to have your intake team log sources consistently, don’t hire anyone for this — including me. The tracking infrastructure is only as good as the twenty seconds of human input at the front of it.
TakeawaysWhat Transfers to Your Firm
- Fix measurement before campaigns. Optimizing on broken data doesn’t just waste the effort — it moves budget in the wrong direction while feeling productive.
- Volume is not performance. Steady call counts hid a sign-rate problem for months. The metric that matters is what a signed case costs.
- Efficiency beats spend. Impression share up and cost per call down at the same time is the signal worth chasing. More budget can’t manufacture that combination.
- Intake is part of the ad campaign. The most sophisticated tracking setup in the world breaks at the point where a human doesn’t log the source.
QuestionsWhat Firms Usually Ask Next
How long before we’d see something?
The tracking rebuild takes days to weeks depending on how much is broken. Campaign results take longer, because Smart Bidding needs an uninterrupted learning period after structural changes. Here, the first fully trustworthy month was the one following the rebuild.
Does this work at a smaller budget?
The tracking work applies at any spend level, and it’s arguably more important at lower budgets because there’s less margin for waste. Campaign results scale differently — less spend means less conversion data, which means Smart Bidding learns more slowly and takes longer to show a readable result.
Our tracking was set up by our last agency. Is it likely fine?
It might be. It also might not be, and there’s no way to know without looking. Broken tracking rarely announces itself — it produces numbers that look plausible. In this account, three separate failures had gone unnoticed while monthly reports were being delivered on top of them.
Do you only work with criminal defense firms?
No, though the approach fits practice areas where a single case is worth enough to justify measuring properly and where clients contact you by phone. Criminal defense, family law, and personal injury all qualify.
What would you look at first in our account?
Whether your conversions are counting what you think they’re counting. Before structure, before keywords, before budget. It’s usually a short conversation and it usually tells you whether the rest is worth discussing.
Find out what your account is actually telling you
A short call, no pitch deck. Bring your Google Ads account and roughly what a signed case is worth to you, and I’ll tell you whether your conversion data can be trusted and what I’d look at first. If your setup is already solid, I’ll say so.
Talk to Henry directly 15 minutes · Henry Bell, not a sales rep
