A renewal is decided in the three or four minutes somebody spends on your report. That is why the document is not a dump of metrics: it is the evidence that you know what is happening on the account and what you intend to do about it. A report that only shows numbers leaves the decision to price.
Why the series matters more than the month
A client renews when they can explain to somebody else why they are still paying, and that sentence never comes out of one good month. It comes out of having watched the same five or six numbers move for half a year in the same table, with the same definitions. A report that changes its metrics every month prevents exactly that: it turns each delivery into a new document instead of the next point on a line.
Two rules follow, and everything else rests on them. The first is that the set of metrics is decided once, at the start, and touched as little as possible after that; if one has to be added, it comes with its history recalculated backwards or it does not come at all. The second is that every number has to mean the same thing in January and in July, which sounds obvious until somebody changes how reach is worked out halfway through the year and the whole series stops being comparable with itself.
There is a third effect, discussed less often. A regular report is the only surface where the client sees the work that is invisible: the pieces that were redone, the campaign moved to another week, the question answered at eleven at night. If that is not written down anywhere, then as far as the person signing the invoice is concerned it did not happen.
The numbers a report cannot add up
Almost every badly built report is badly built for the same reason: somebody stored one value per day and then added thirty rows together. For some metrics that is exact, and for others it is wrong by an order of magnitude, and the difference is invisible on the sheet.
| Family | What it measures | How the period total is formed |
|---|---|---|
| Count | Likes, comments, shares, saves, clicks, impressions | Adding the days is exact |
| Running total | Total followers, videos published | The total is the last day value, never the sum |
| Unique | Reach, accounts engaged | Only the network can work it out; where it does not, the cell stays blank |
| Rate | Engagement rate, average view duration | Recomputed from its components, never averaged across days |
The case that breaks the most reports is reach. It counts distinct people, so somebody who saw you on Monday and again on Tuesday is one person and not two: thirty days of twelve hundred reached is not thirty six thousand reached. A report that sums that column shows growth that does not exist, and shows it every single month, so the error compounds until the client compares it against what their own app tells them. How to read it properly, and what to compare it with, is in the guide to reach and impressions.
The second case is the empty cell. When a network does not report a metric, the value is not zero, it is nothing. Printing a zero claims there were zero, and next to it appears a change of zero per cent that additionally claims nothing moved. In a monthly report that cell is the one that produces the awkward question, and there is no good answer to it once the document has already said the number was zero.
What each network reports, and what it does not
Six networks serve account level analytics, and no two of them report the same things. It is worth having this in front of you before promising a complete table in a proposal.
| Network | Not reported per post | When the figure is final |
|---|---|---|
| Clicks | Can lag up to 48 hours | |
| Saves | Can lag up to 48 hours | |
| TikTok | Impressions, reach, saves and clicks | No declared lag |
| YouTube | Impressions, reach, saves and clicks | Finalised two or three days late |
| Views and saves | No declared lag | |
| Google Business | A different vocabulary: it measures discovery | Can lag two or three days |
Two practical readings. First, LinkedIn does report reach per post, contrary to what is usually assumed, and what it has no notion of is a video view count on a company page. Second, Google Business does not fit that comparison and should not be forced into it, because what it measures is how many people found the listing and then called, asked for directions or opened the website. Putting those appearances in the same column as an Instagram post's impressions is a comparison that means nothing.
Audience demographics deserve the same honesty before they are sold: they exist on Instagram, with age, gender, country and city, and on YouTube, with age, gender and country and counting only signed in viewers. Everywhere else there is no chart to draw, and promising one in a proposal is an unpleasant conversation thirty days from now.
A weak month, and whose fault it is
Before writing the explanation, separate three causes that look alike on the sheet and are not alike at all. The first is provisional data: if you close the report on the 1st and one of the networks is still revising the last week, the month comes out short for a reason that fixes itself. The second is the platform: a format that stops being distributed, a metric whose definition changes, a shift that hits the whole sector at once. The third is the work, which also happens.
All three get told, and they get told differently. The provisional one is flagged inside the document and corrected in the next one. The platform one comes with what you did the moment you spotted it, because a client forgives the environment and does not forgive passivity. The third is said in one sentence, without decoration, followed by the decision, which is the only part the person paying actually wants to read.
What never works is the consoling average: moving the comparison window until you find one where the number goes up. It shows, and it costs more than the bad month did.
Everything in this article happens in one place in GoFeed.
Try it freeFrom numbers to a decision
The part that separates a report from a statement takes half a page and almost nobody writes it. It is the part that says, for each number that moved, what you conclude and what you are going to change. Without it the client has to do that work, and the work the client does is the work they stop paying you for.
It reads well as three short blocks. What stays, because it is working and should not be touched. What gets corrected, with the specific change and the date its effect will show. And what you propose, which is the one place in the document where it is fair to ask for something: budget, an access credential that is missing, a decision that has been waiting for weeks.
That last block turns the report into a commercial tool without looking like one. A client who has spent four months reading "we would change this if we had that" arrives at the renewal conversation with the proposal already digested. And when the report starts coming back without comments, or the monthly call is cancelled twice running, that is information rather than coincidence: the signals a client is leaving show up in the diary well before they show up in an email.
Who builds the report and who receives it
Two things get confused here and are worth keeping apart. Per network and per post metrics are in every plan, for all three audiences: if you run your own brands, you get the numbers just the same. The monthly report is something else, a document sent to a client, and that is why it lives in agency workspaces.
That report is prepared on its own from the month's metrics and goes out on the last working day, Madrid time. Each brand carries its own mode, and there are three: off, which generates nothing; automatic, which sends without anybody touching it; and review, which writes the draft, notifies you and sends nothing until somebody on the team says so. One client can receive it unattended while another always passes through you, with no need for two separate processes.
It reaches the client's people in their portal, plus any addresses you add, and with no recipients nothing is sent. Every send is stored and downloads as a PDF exactly as the client received it, which is the only way to answer "what did you send us in March?" inside a minute. What goes into each block is set out on the reporting page, and the full mechanics, with the three modes and each network's lag, on analytics and reports.