Five things, in this order: what it cost, what it produced, what changed and why it changed, what happens next month, and what went wrong. Everything else in a monthly report is decoration.
A report that lists activity without connecting it to booked work is a timesheet. A report you cannot act on is a receipt. Both are common, and both are usually longer than the version that would have been useful.
Here is what each of the five should contain, what a report should never do, and the two things that belong in it for reasons that have nothing to do with marketing.
The five things, and why the fifth is the tell
The list is short deliberately. A monthly report that cannot be read in ten minutes will not be read at all, and a report nobody reads is a cost with no benefit attached.
| What | What good looks like |
|---|---|
| What it cost | Ad spend and fee, separated, reconcilable against what actually left your bank |
| What it produced | Enquiries counted once each, and — where it can be recorded — how many became work |
| What changed and why | A short list of the changes made, each with the reason, dated |
| What happens next | What will be done next month and what it is expected to move |
| What went wrong | What was tried that did not work, and what was learned from it |
The fifth row is the diagnostic. Every month contains something that did not work, because that is what testing means. A supplier who reports a good month every month is either not testing anything or not telling you about it, and both of those are worse than a month with a failure in it honestly described.
The fifth row is also the one that predicts the relationship. A supplier who volunteers a mistake in month two will tell you about the serious one in month fourteen. One who has never reported a failure will not start when it matters.
Money in, work out: the two numbers an owner has to hold
An owner does not need to hold twenty metrics. Two are enough to run the relationship, and everything else is the supplier's working.
The first is total cost — spend plus fee, together, because the fee is part of what the enquiry cost you and reports that exclude it are flattering themselves. The second is booked work attributable to it, or as close as your systems can get.
Cost per enquiry is the supplier's number. Cost per booked job is yours.
Where the second number genuinely cannot be produced, say so in the report rather than substituting a proxy without comment. A report that quietly presents form submissions as though they were customers is not lying, exactly, but it is letting you believe something specific that it has not checked.
The gap between those two numbers is where nearly every unproductive agency argument lives. Enquiries are up and revenue is flat means the enquiries got worse, or the follow-up did, and the report should be able to tell you which. That requires the outcome of each enquiry to travel back into the measurement setup — which is a job somebody has to own, and it is usually the job nobody has been given.
A change log beats a dashboard
Live dashboards are useful and they are not a report. A dashboard shows the current state. A report explains how it got there, and only one of those helps you decide anything.
What makes the difference is a dated list of changes with reasons attached. Six lines is usually plenty. With it, a movement in the numbers can be traced to a decision, and next month's decision can be made on evidence. Without it, every conversation restarts from the beginning, and both sides are reduced to arguing about impressions of the last four weeks.
What a change entry needs
- The date, so it can be lined up against the numbers.
- What was changed, in a sentence a non-specialist can follow.
- Why, referring to something observed rather than to a general principle.
- What it was expected to do, written before the result was known.
That fourth item is what separates a change log from a diary. A prediction made in advance and reviewed the following month is the mechanism by which an account gets better. Recorded afterwards, it is just narration, and narration always agrees with whatever happened.
Cost per enquiry is the supplier's number. Cost per booked job is yours.
What a good report never does
Some things appear in reports so consistently that their presence has stopped registering. They are worth noticing again.
- Leading with impressions. Impressions are the cheapest number available and the least connected to money. A report that opens with them has chosen its opening for a reason.
- Comparing mismatched windows. A five-week month against a four-week one, or a period containing a public holiday against one without, produces a difference that means nothing.
- Changing the headline measure. If last month's report led with cost per enquiry and this month's leads with click-through rate, the measure was changed because the first one moved the wrong way.
- Screenshots without commentary. Ten platform screenshots are ten pieces of evidence for an argument nobody made.
- Language you cannot repeat. If you cannot re-explain the report to your business partner, it has not been written for you.
The last one is the most reliable of the five, and it needs no expertise to apply. A person who understands what they did can explain it plainly. Jargon in a client report is either a habit or a hiding place, and the two are hard to tell apart from the outside — which is the argument for treating an unexplainable engagement as a decision rather than a mystery.
The records you are expected to be able to produce
Two obligations sit underneath the monthly report and neither is a marketing matter, which is exactly why they get missed.
The first is record-keeping. The ATO requires business records supporting claimed expenses to be kept, generally for five years, and advertising is a claimed expense. If your only record of what you spent is a supplier's dashboard you lose access to when the relationship ends, you have a problem that will not present itself until an inconvenient moment. Invoices and platform billing you can retrieve yourself are the answer, and the monthly report is the natural place to reconcile them.
The second is substantiation. The ACCC publishes guidance on substantiating claims made in advertising, and it applies to the claims in your own marketing — the ones your supplier writes on your behalf. If an ad says fastest, cheapest, or number one, somebody has to be able to support it, and that somebody is the business whose name is on the ad. A monthly report is a reasonable place for a supplier to note what claims are running, and a reasonable place for you to ask.
The same standard is worth turning around. When a supplier tells you what they achieved for someone else, ask what window, what spend and what was being counted. Our own written-up engagements publish those alongside the result for exactly that reason — a figure without its window and its definition is not evidence, it is a decoration.
A report that lists activity without connecting it to booked work is a timesheet.
Who owns the account, and why that belongs in the report
It looks like an odd thing to put in a monthly report, and it is there because it is the item most likely to be discovered at the worst possible time.
Google Ads distinguishes between administrative ownership of an account and access granted to somebody else. If a supplier holds the ownership, then the history the platform has learned from, the record of what has already been tried and the measurement built around it can all leave with them. Businesses find this out during a handover, when there is no leverage left.
One line is enough: who owns the advertising account, who owns the measurement setup, and who owns the data. Repeated monthly, it stops being a negotiation and becomes a fact both sides have already agreed to.
The same logic applies to the reporting itself. A report you can reproduce from your own access is a check on the work; a report only the supplier can produce is a claim about the work. The difference costs nothing while the relationship is healthy and everything when it is not.
If you want the version of this written as an operating agreement rather than a checklist, our ninety-day method sets out the gates and the evidence expected at each one, and the terms we engage under state what stays with the business.
Common questions
How long should a monthly report be?
Short enough to read in ten minutes. The five things — cost, output, changes and reasons, next month, and what went wrong — fit comfortably in two or three pages. Length is usually a substitute for a conclusion rather than evidence of work.
Is a live dashboard enough on its own?
No. A dashboard shows the current state; a report explains how it got there and what will be done about it. Keep the dashboard for the numbers and expect a written explanation of the changes, the reasons and the expected effect alongside it.
My report only shows clicks and impressions. Is that a problem?
It is a signal. Those are inputs, and a report built on them is describing effort rather than outcome. Ask for enquiries counted once each, the cost including the fee, and — where your systems allow it — how many enquiries became paid work.
What should I ask for if the report never mentions failures?
Ask what was tried last month that did not work. Every month contains something, because that is what testing means. A supplier who volunteers a small mistake early will tell you about a serious one later; one who never reports a failure will not start when it matters.
Questions and answers
What should I send the agency before the reporting meeting?
Share changes in availability, accepted work, repeated customer questions and significant delivery constraints. These help interpret results that advertising data alone cannot explain.
How should a report handle an experiment that did not help?
Record the original question, the change and what was learned. Explain whether it was reversed, adjusted or needs more observation, with a reason for the next decision.