Measurement guide
Reading a marketing report, including the parts designed not to be read
A forty page report is not more thorough than a two page one. It is usually a two page report with thirty-eight pages of cover, and the difference matters because you are meant to judge the work from it.
9 min read · 7 steps · By Aman Mishra · Written August 2026
What you end up with
A short list of the numbers you will actually judge the work on, and six questions that change what your next report contains.
Have these ready
- The last two reports you were sent
- Your own count of enquiries for the same period
- Ten minutes and a willingness to ask
What it takes
9 min to read
7 steps, in the order below. Skip to the one you are stuck on.
Reports are how you judge work you cannot see. That makes the format worth understanding, because a report can be entirely accurate and still leave you knowing nothing.
None of what follows assumes bad faith. Most padded reports are produced by people who are busy and using a template, and the template was designed to look thorough.
-
Look at the first number on the first page
What a report opens with tells you what it wants you to think about. If it opens with impressions, note that and keep reading with that in mind.
Whatever a report opens with is what it wants you to think about.
If the first thing is enquiries, or cost per enquiry, you are looking at a report written to be judged. If the first thing is impressions, sessions or reach, you are looking at a report written to be presented.
That is not proof of anything on its own. It is a useful thing to notice before you read the other thirty pages, because it tells you what the author considers the headline.
-
Separate branded traffic from everything else
People searching for your company name were coming anyway. Counted together with the rest, they can make a flat quarter look like growth.
Branded searches are people typing your company name. They were coming to you regardless of any marketing work.
Included in a total traffic figure, they distort everything. A business that gets a mention in the press, runs a radio ad, or simply grows by word of mouth will see branded search rise, and a report showing total organic traffic will present that as the result of search work.
Ask for the split. Search Console does it in two clicks with a query filter. Any report on search work should already contain it, and the fact that most do not is worth noting.
-
Check whether the leads number is a real lead
Conversions is a configurable word. Find out exactly which actions are being counted before you judge the total.
Conversions is a word that means whatever somebody configured it to mean.
Ask exactly which actions are counted. It is common to find newsletter signups, PDF downloads, contact page views and clicks on a phone number all included in one total labelled leads.
Some of those are worth tracking. None of them is somebody asking you to sell them something, and a total that mixes them cannot be compared with anything, including itself over time if the definition changed.
-
Compare against the right period
Month on month comparisons in a seasonal business are noise. Year on year is usually the only comparison that means anything.
Month on month is the default comparison and it is usually the least informative one.
In any business with seasonality, and that is most of them, comparing March with February measures the calendar. Year on year is the comparison that means something, because it holds the season constant.
Be similarly careful with percentages on small bases. Enquiries up 200 per cent reads impressively and can mean three became nine.
-
Find what the report does not contain
Absences are informative. A report with no cost per lead, no mention of anything that went badly, and no plan is telling you something.
Absences carry information.
No cost per lead. No comparison to the same period last year. Nothing that went badly. No mention of anything tried that did not work. No plan for next month more specific than continue to optimise.
A report where everything went well every month for a year is not describing marketing. Real work includes tests that failed and months where a competitor did something, and a report that never says so is either not looking or not telling.
-
Ask the six questions
Six specific questions change what arrives next month, because a report has to be written differently to answer them.
Send these once and next month’s report will be different, because it has to be written differently to answer them.
How many enquiries did we get, and what did each one cost.
How does that compare with the same month last year.
How much of our traffic is people searching our own name.
Which specific actions are being counted as a conversion.
What did you try that did not work.
What are you doing next month, and what do you expect it to change.
None of these is aggressive and all of them are answerable by anyone doing the work properly. The response to being asked tells you as much as the answers.
-
Agree what a good month looks like in advance
Judging performance after the fact against no agreed standard is how everybody ends up arguing about which metric to use.
The underlying problem with most reporting arguments is that nobody agreed the standard beforehand.
Set it at the start. Two or three numbers, the direction they should move, and over what period. Write it down where both sides can see it.
Then the monthly report has one job: show those numbers against that standard, and explain the gap either way. Everything else in the document is context, and context is fine as long as it is not doing the job of the result.
When you have done all 7
A short list of the numbers you will actually judge the work on, and six questions that change what your next report contains.