A social report that changes what you film

The report lands on the second of the month. Followers, reach, impressions, likes, a graph that goes gently up and to the right. You read it in about forty seconds, feel vaguely reassured or vaguely worried, and file it.

Then you plan next month exactly as you would have planned it if the report had never arrived.

A report is not a record of what happened. It is an instruction for what to film next, and it needs three columns to be that: what happened, what it means, and what we will do differently. Most reports contain only the first, which is why most reports are read once.

What is a report for?

To inform a decision, including a well-supported decision to keep doing what works.

A monthly review can provide a useful planning rhythm. Some results need a longer window, and the report should say which figures are still settling.

So the question the report answers is not how did we do. It is what should we make more of, and what should we stop.

The three columns

Write it as a table if that helps, or as three short paragraphs per finding. The shape matters more than the format.

What happened

The observation, in one sentence, with the number attached. No adjectives. "The two posts filmed in the workshop were the two most saved posts of the month" is a what-happened row. "Engagement was strong" is not, because it is a judgement, not a number.

What it means

Your best reading of why, labelled as a reading. This column is allowed to be wrong. It is not allowed to be absent, because a number with no interpretation attached gets interpreted anyway, silently, by whoever reads it next.

What we will do differently

The row everybody skips. It names a change to next month's filming or posting, specifically enough that somebody could act on it without asking a question. "Film two workshop pieces on the batch day, drop the desk-based explainer to once a fortnight." That is the row. Everything above it exists to justify it.

Which numbers earn a place?

The ones tied to a decision you can make.

Saves and shares, because they tell you a piece was worth keeping or worth passing on, which is closer to intent than a like has ever been. How far people watched, because it tells you where the piece lost them and that location is a filming instruction. Profile visits and follows from a specific piece, because that is the moment somebody decided you were worth returning to. Comments, read rather than counted, because they contain the next month's subjects in the audience's own words.

Which numbers should come out?

Use impressions when distribution or awareness matters, and explain what they cannot establish. Seeing a post does not show that someone understood it or took action.

Follower count as a headline, because it is a slow-moving vanity number that flatters you in good months and frightens you in bad ones without either being informative. Track it quietly, do not lead with it.

Anything you have never once acted on. Look at last quarter's reports and find the row that has appeared three times and never caused a change. That row is decoration. Cut it and the report gets read, which is the trade you are making.

How do you write the middle column without guessing?

By comparing pieces that differ in one thing, which means you have to have made pieces that differ in one thing.

If several things changed, treat the explanation cautiously. Even two posts with one planned difference may have reached different audiences or appeared in different conditions. Use comparisons to form a hypothesis and design a stronger test where the decision warrants it.

Where you do not know, write that you do not know and name the test that would tell you. An admitted gap is more useful than a confident invention, and it survives contact with next month.

The handover test

Could somebody who was not there plan from this? Apply it literally.

Hand the report to somebody who did not do the filming, did not read the comments and was not in any of the conversations. Ask them what they would shoot next month. If they can answer, the document works. If they come back with questions, the answers to those questions were the rows you left out.

Do that once, this month, before you write the next one. It takes ten minutes and it will embarrass you exactly once.

Sources and evidence

This is our reporting method. Observations, possible explanations and actions are kept separate; the strength of any conclusion depends on the data and comparison behind it.

Our method. Writing what happened, what it means and what will be done differently as three separate columns, dropping numbers nobody has ever acted on, and testing a report by handing it to somebody who was not there, is how this studio builds a monthly report. It is a working practice, not an external rule.

References.

• YouTube. YouTube performance FAQ and troubleshooting. support.google.com. Publish time is not known to affect long-term performance; use appropriate video length. Official platform statement, not proof for other platforms.