I need to know if my marketing is working
You are paying for marketing and receiving reports, and you cannot tell from them whether the firm is better off. That is not a failure of attention; most marketing reporting is built to look busy rather than to answer the question. This page gives you the five numbers that answer it, the ones that only look like answers, how to set up measurement in a week, and what to do when the honest answer is no.
A managing partner should be able to read a marketing report in four minutes and know what it cost to win a file last month. If that is not possible from what you receive, the report is the first thing to fix.
Trusted by firms including
Five numbers, and only five
Everything a partner needs to know about marketing fits in five numbers a month. Anything else in a report is working, not results, and should be labelled as such.
| The number | Where it comes from | How to read it |
|---|---|---|
| Qualified enquiries | Calls and forms that were a matter the firm does, in a place it covers, worth taking. Counted from call tracking, form tracking and whoever logs enquiries, not from “leads” in a dashboard. | By source and by matter type. The split is the point: thirty enquiries is meaningless until you know that twelve came from one channel and that half of those were the wrong matter. |
| Instructions | Files opened that started as a marketing enquiry, confirmed from the case management system against the enquiry log. | The number the firm actually wanted. Lag it by a month, because an enquiry in March may instruct in April, and judge channels on the quarter rather than the month. |
| Cost per instruction | Everything spent on a channel in the period, media and fees, divided by the files it produced. | The only fair way to compare channels. A channel with expensive clicks and cheap instructions is better than one with cheap clicks and no instructions, and only this number shows it. |
| Fee income from marketed files | The estimated or billed fee on the files that marketing produced, from the matter type’s average or from the actual bill. | Set against the total marketing cost for the same period. This is the return. If the report cannot show it, nobody can say whether the marketing paid, and a conversation about budget is a conversation about feelings. |
| Direction against the plan | Each of the above against the number agreed when the work started, and against the same quarter last year. | Seasonality is real: conveyancing in August and family in January behave differently from the rest of the year. A month is noise; a quarter against the plan is a signal. |
If a marketing report cannot tell you what it cost to win a file last month, it is not reporting on marketing. It is reporting on activity, and activity is what you are paying for, not what you are paying to get.
The numbers that look like results and are not
None of these is useless. Each is a working number that explains why the five above moved. Reported on their own, as if they were the result, they are how a firm pays for a year of marketing and ends it no better off.
| The number | Why it is reported | Why it misleads on its own |
|---|---|---|
| Rankings | Easy to track, satisfying to watch go green. | A page one ranking for a term nobody who instructs searches is worth nothing. Rankings explain why enquiries moved; they are not the enquiries. |
| Traffic | Big, and it usually goes up. | Visitors from the wrong places, for the wrong questions, at the wrong stage. A firm can double its traffic and lose instructions in the same quarter. |
| Impressions | Even bigger. | How often the site appeared in a search result, including at position 80. Impressions without clicks are a description of invisibility. |
| “Leads” | Sounds like the result. | Undefined, it includes spam forms, recruitment agencies, students and people in Aberdeen. Ask what counts as a lead and whether anyone checked. |
| Return on ad spend | Standard in paid media. | Calculated on form submissions valued at an invented figure, it is fiction. Calculated on instructions and real fees, it is cost per instruction with a different name, and useful. |
| Followers, reach, engagement | The only numbers social platforms give you. | None of them is a client. Social media earns referrals and reputation, and should be judged on referral enquiries the partners can trace, not on applause. |
| Open rates | Standard in email. | Unreliable since Apple began pre-loading emails in 2021. Replies, calls and instructions from a send are what an email was for. |
| Domain authority and similar scores | A tidy single number from an SEO tool. | A third-party estimate of Google’s opinion, which Google does not use. It can rise for a year while enquiries fall. |
Measurement in a working week
The firm cannot judge its marketing until every enquiry has a source and an outcome. That takes five days to set up, most of it by whoever answers the phone, and it is the single highest-return thing on this page.
- Call tracking
A tracking number for each channel: one on the website, one in the ads, one on the Google Business Profile, one for print if you still use it. Calls recorded, with callers told. From now on a call has a source.
- Forms and analytics
Every form submission recorded as a conversion in GA4 with the page it came from, and the form sending to a named person rather than a shared inbox. Search Console linked, so impressions by matter can be read later.
- The enquiry log
A single list, in the case management system or a spreadsheet: date, source, matter, qualified or not, outcome, fee if instructed. Whoever takes new enquiries owns it. Ten minutes a day.
- The question
“How did you hear about us?” asked on every call and on the form, with the answer written in the log. Imperfect, and the only way referrals and brand searches get counted at all.
- The join
Enquiries matched to files opened, monthly, by someone with access to both. This is the step most firms skip, and it is the step that turns “we got 38 enquiries” into “we won 11 files at £410 each”.
How to read a report in four minutes
Whoever sends the report, your agency or your own marketing manager, ask the same questions of it every month. The questions are not hostile; a good report answers them before they are asked, and a report that cannot answer them is telling you something about the work behind it.
Start at the end: what did the files won this month cost, and what are they worth? Then work backwards. Which channel produced them, and did that channel’s cost per instruction rise or fall? What was changed this month, and why? What was stopped? And what, specifically, will be done next month as a result of these numbers?
A report that answers those five questions in a page is doing its job. One that opens with a traffic graph and closes with a list of tasks completed is a timesheet, however many pages it runs to. The checklist alongside is the quick version; it takes less time than reading the report.
- ✓Instructions and their cost are on the first page
- ✓Every enquiry can be traced to a source
- ✓Something was stopped this month, and the report says what
- ✓Next month’s actions follow from this month’s numbers
- ✓A bad number is named, with a reason and a fix
- ✕It opens with a traffic or rankings graph
- ✕“Leads” appear without a definition
- ✕It lists tasks completed rather than results produced
- ✕Every number is green
Attribution, honestly
No measurement system will tell you the whole truth about where a client came from, and firms that try to build one that does spend more on the measuring than on the marketing. Three things to accept, and the decision rule that follows from them.
A client who read a guide in March, saw a partner’s LinkedIn post in May and searched the firm’s name in June shows up as “direct” or “brand search”. The guide and the post did the work; the tracking credits the name. Brand searches rising is evidence that the invisible channels are working.
Almost every referral checks the firm online before calling. A referral that found a thin site and rang a competitor was lost by the marketing without ever appearing in it. “How did you hear about us” is how you catch the ones that came through.
Because attribution is imperfect, no single month proves anything. The rule: a channel’s cost per instruction, on the quarter, against the plan and against the same quarter last year. If it is falling, keep it. If it is rising for two quarters, change it. If it cannot be calculated, that is the finding.
When the answer is “it isn’t working”
Sometimes the five numbers come back and the marketing has not paid. The instinct is to cancel everything, and that is usually the second mistake, because it throws away the measurement and the working channels along with the failing one.
Separate the channels. One of them is often fine and being dragged down by another in the averages. Then look at where the loss is: not enough enquiries at all, enquiries that were the wrong people, or the right enquiries not becoming files. Those are three different problems with three different fixes, and only the first of them is about the marketing channel itself. The more leads page walks through locating it.
Finally, check the clock. Paid search that has not paid after three months has a problem. Organic search and content that have not paid after three months are on schedule; they are judged at six and nine. A programme judged on the wrong clock looks like a failure one month before it starts to work.
- Separate
- Cost per instruction by channel, not the blended figure
- Locate
- Too few enquiries, the wrong enquiries, or the right ones not instructing
- Check the clock
- Three months for paid search; six to nine for organic and content
- Keep the measurement
- Whatever else goes, the tracking and the log stay
- Change one thing
- Fix the located problem, give it a quarter, measure again
Questions partners ask about measuring marketing
-
Our agency reports rankings and traffic. Is that wrong?
Not wrong, incomplete. Ask them to add the five numbers above, starting with instructions and cost per instruction. A good agency will have the data or will set up the tracking to get it; one that resists is telling you the work does not stand up to it.
-
How long before we can judge?
A quarter for paid search, two to three for organic search and content, with the measurement in place from the first week so the judgement is on data rather than impression. Set the review date and the numbers it will be judged on before the work starts.
-
What is a good cost per instruction?
One that leaves the firm a margin on the fee for that matter type, which is why there is no single answer: £400 is excellent for a contested probate file and ruinous for a basic will. Work it out per matter from your own fees, and compare channels against each other rather than against a benchmark.
-
We cannot connect enquiries to files. Is that normal?
Common, and fixable in a week. It needs an enquiry log with a source, the case management system recording where a file came from, and one person matching the two monthly. Until that exists, every report is an estimate.
-
Should we use a dashboard?
Only if someone will look at it. A one-page monthly report that a partner reads beats a live dashboard nobody opens. The dashboard is useful for the people doing the work; the report is for the people paying for it.
-
Can you review our existing reports?
Yes. Send three months of whatever you receive and we will tell you what they show, what they hide, and which of the five numbers can be produced from the data that already exists. We will say if your current arrangement is working, because sometimes it is.
Send us your last three reports
We will read them the way a partner should be able to, tell you what they actually say about instructions and cost, and show you what the one-page version would look like for your firm. If your marketing is working, we will say so.
- Three months of reports, and read access to your analytics if you can give it.
- A written reading within a week: what the numbers show, what is missing, and the five-number version.
- A conversation about whether anything should change. Often the answer is the report, not the marketing.