How to audit a social media account in five passes

The audit came back as forty slides. Follower growth by month, engagement by format, a word cloud, best time to post. Everybody nodded, nobody disagreed with anything, and three weeks later the account is doing exactly what it did before. The audit was thorough and it changed nothing, which means it was not an audit.

This is how to audit a social media account in about ninety minutes, on any public profile including one you do not own, ending in two decisions rather than forty charts.

TL;DR

Before you start: write the question

An audit with no question produces a description. Write one sentence at the top of the document before opening the profile, and every pass below either serves it or gets cut.

• Should we keep investing in this channel? The founder's question. The cadence and audience passes matter most.
• Why is this client's account not working? The agency's question. Distribution and themes matter most.
• What is this prospect doing? The pitch question. Cadence and positioning, and you can skip the rest.
• How do we compare to them? A benchmark rather than an audit, and a different procedure.

An audit that does not end in one thing to start and one thing to stop was a description with charts on it.

Pass one: cadence

Always first, because it explains more than anything else and it is the cheapest to read. List the last thirty posts with their dates. That is the whole pass.

1. Count per month. Not per week. Monthly counts show the shape without the noise.
2. Mark the gaps. Two silent weeks in a busy quarter is a resourcing fact, not a strategy.
3. Note when it started or stopped. A cadence that begins in September usually means somebody was hired.
4. Compare to what was promised, if anything was. This is the only pass with a target to miss.

A large share of accounts that look like they have an engagement problem have a publishing problem, and this pass finds it in fifteen minutes. Why consistency outperforms bursts is in social media posting consistency.

Pass two: distribution

Mark the top five and bottom five posts by public engagement. Then look at them next to each other and write down what separates them. Format, usually. Sometimes topic. Rarely posting time, whatever the tools say.

Use the median rather than the average when you summarise. One post that travelled ruins an average and tells you nothing about the next one, which is the number the client actually cares about.

Compute the rate consistently while you are here, because you will be comparing periods later. The formula and the trap that comes with it are in how to calculate engagement rate.

Pass three: themes, which no tool reads for you

Read the last thirty posts and write down the three subjects they keep returning to, in the account's own words. This is manual, it takes twenty minutes, and it is the pass that produces the finding people remember.

Software counts posts and cannot tell you that the account quietly switched from product features to customer stories in April. That switch is usually the most actionable thing in the whole audit, and it is invisible in every dashboard.

Pass four: who actually replies

Open the comments on the five best posts and look at who is commenting. Not how many. Who.

• Colleagues and friends. A pleasant account with no commercial audience yet.
• Peers in the same job. Good for reputation, slow for pipeline.
• People at target companies. This is the account working, and it is invisible in any follower count.
• Nobody, or bots. Engagement bought or borrowed, and worth saying out loud.

Fifteen minutes here beats a month of follower charts. It is also the pass that most often reverses the conclusion the numbers were pointing at.

Pass five: what a stranger concludes

Open the profile on a phone, logged out, and read only what is visible without scrolling twice. Then write one sentence describing what this person or brand does, using only what you just saw.

If that sentence is vague, the account has a positioning problem rather than a content problem, and no amount of posting frequency fixes it. This is the shortest pass and the one that most often produces the thing to stop.

The output: two decisions and a baseline

Close the audit with one thing to start, one thing to stop, and the numbers you read today written down with their date. The last part is what makes the next audit a comparison rather than another description.

Record follower count, posts in the last thirty days, and median engagement, with the date. In three months those three lines are worth more than the forty slides, because they are the only evidence that anything changed. What is readable from outside on each network, so you know which of the three you can capture for an account you do not own, is in public social media metrics.

Writing it up in one page

The audit document is one page, and the constraint is the point: forty slides exist because nobody decided what mattered, and a page forces the decision.

1. The question, restated. One line, the one you wrote before starting.
2. Three numbers with their date. Followers, posts in the last thirty days, median engagement.
3. What is working, with an example. The actual post, not a description of it.
4. What is not, with an example. Including this is what makes the rest believable.
5. One thing to start, one to stop. Named, and small enough to begin on Monday.

If a finding does not fit on the page it was not important enough to survive the meeting anyway. Keep the raw notes attached for anybody who wants them, and never lead with them.

Four ways an audit goes wrong

All four are common and all four produce a document that reads well and changes nothing.

1. Starting with engagement. It is the noisiest number and the least actionable. Cadence explains more and costs fifteen minutes.
2. Averaging instead of taking the median. One post that travelled makes an account look healthy for a year.
3. Auditing without a baseline to compare to. The first audit is a description by definition. Record the numbers so the second one is a comparison.
4. Confusing a content problem with a positioning problem. If a stranger cannot say what the account is for, posting more makes it worse rather than better.

The last one is the most expensive because it is invisible in every number. It only shows up in the fifth pass, which is the shortest and the one people skip when they are running out of time.

One last thing to keep from every audit: the date you ran it. An audit without a date is a claim about an account that nobody can check later, and the second audit is only worth doing if the first one left something to compare against. Three numbers and a date, in the same place every time, is the entire archive and it takes a minute to write.

And one note on who should run it. An audit written by the person who produced the content is a defence, however honest they try to be. Somebody one step removed finds the positioning problem in five minutes, because they are reading the profile the way a stranger does rather than the way its author remembers writing it.

How Groowth handles this

It does passes one and two continuously, and keeps the baseline for you. Paste a public handle and it reads the profile daily: what was published and when, the follower count on that day, and the public likes and comments on each post.

That turns the ninety minute audit into the three passes a person is genuinely better at, themes, audience and positioning, and removes the two that are just counting. It works on an account you do not own, which is the case for every pitch audit. There is also a free audit that reads one public profile with no account at all, and three accounts free with no card.

Frequently asked questions