Social Media Audit: Four Questions, One Decision Sheet

Four questions, and the one-line decision each account has to get.

Frame the audit by the question, not the data

An audit that ends in a tidy spreadsheet and no decision was a waste of a week.

This page assumes a strategy already exists and tests whether your accounts deliver it. If the strategy itself is the problem, start with our social media marketing guide instead.

The columns are the easy part. Every template gives you followers, reach and engagement rate; almost none tells you what a bad number obliges you to do. So start from the questions. Four are worth answering, and the data exists only to answer them:

• Are we on the right platforms at all? Not "are we posting" — are these the places our buyers are.
• Is the content doing a job? Every post should educate, prove, sell or entertain. Posts with no job are the ones nobody remembers.
• Is anyone actually seeing it? Reach and distribution, separate from follower counts.
• Does any of it connect to the business? Traffic, enquiries, assisted conversions — or an honest "we cannot tell yet".

Step 1: Inventory every account that exists in your name

Before any metric, list every profile that exists in the brand's name, including the ones nobody remembers creating.

This is the step people skip, and it produces the commonest real finding in an audit: abandoned and duplicate profiles. They rank for your brand name, they catch customer messages nobody reads, and an old profile tied to a former employee's email is a security exposure, not just clutter.

1. Search each platform's own search box for your brand name, product names, spelling variants and abbreviations — and look past the first result.
2. Search the web for your brand name against each platform's domain — a site-restricted search on instagram.com, facebook.com, linkedin.com, youtube.com, x.com and pinterest.com surfaces what platform search buries.
3. List branch, city and campaign profiles. Multi-location brands accumulate these, and a launch handle from 2022 stays live until somebody closes it.
4. Check your own records — the shared password manager, and the marketing inbox for old signup and verification emails. That is usually where the forgotten handle is.
5. Mark each one owned, orphaned or impersonating. Orphaned means yours but unmanaged; impersonation goes to the platform, not into the audit.

Record the handle, URL, follower count, date of the last post and who holds access. An account nobody can log into is a decision in itself.

Step 2: Pull the same window for every platform

Pick one window and hold it across every account. Comparing a 7-day window on one platform to a 90-day window on another is how people conclude everything is broken.

Ninety days is the practical default — long enough to survive one viral post, short enough to describe the account you have now. Pull the preceding 90 days too, so you have a comparison rather than a snapshot.

One hard constraint shapes this: Instagram's platform documentation states that "User Metrics data is stored for up to 90 days" (Instagram Platform > Insights — no last-updated date shown; read 9 October 2026). Year-on-year account numbers only exist if you were already exporting them.

Export each pull to a dated file — months later it is the only version of the number you can still reproduce.

Step 3: Read reach, engagement rate and follower growth honestly

Reach is people, not deliveries. Instagram's documentation defines a post's reach as the "Total number of unique accounts that have seen the media object", which is why it sits below impressions. Compare reach to your follower count: if reach is a small fraction of followers, distribution is the problem, not the creative.

Engagement rate is engagements divided by a denominator — and the denominator is the whole argument. Engagements ÷ reach asks whether the people who saw it responded; ÷ followers asks how alive the audience is; ÷ impressions is what most platforms report. Three different questions, so state which one you used and use it everywhere. Our engagement rate calculator keeps the arithmetic consistent.

Platforms do not agree on the numerator either. LinkedIn's marketing documentation defines engagement as the "Number of organic clicks, likes, comments, and shares over impressions" (Organization Share Statistics, last updated 30 April 2026) — clicks included, impressions as the denominator. Instagram's documented engagement metric is the "Total number of likes and comments on the media object" — no clicks at all.

So a LinkedIn engagement rate and an Instagram one are not the same measurement. A number is only comparable to itself, over time, on the same platform — and a benchmark rate from another industry or follower band tells you nothing.

Follower growth is the weakest of the three and the one leadership asks about. Read it as a rate of change against the same window last quarter, and read it next to reach: followers climbing while reach falls means you are accumulating an audience that is not being served your posts.

And let fresh numbers settle: YouTube's help documentation says it "may temporarily slow down, freeze, or change your metric count, and discard low-quality playbacks" (How engagement metrics are counted — no last-updated date; read 9 October 2026).

Step 4: Group content by job, not by format

Here is where most audits go wrong: they sort the window's posts into carousels, Reels, statics and stories, find one format outperforms, and recommend more of it. Formats come and go. The job a post was doing does not.

Tag every post in the window with one of four jobs, then total the reach and the engagements each job earned:

• Educate — teaches something usable. Usually earns saves and shares.
• Prove — case studies, results, reviews, before-and-after. Usually earns profile visits and enquiries.
• Sell — the offer, the launch, the call to act. Usually earns clicks and fewer likes.
• Entertain — trend, humour, behind-the-scenes. Usually earns reach and follows.

Then read the mix. A feed dominated by entertain earns reach and no enquiries; one dominated by sell earns neither. And look for the job missing entirely — nearly always prove, because proof takes effort to produce.

If the mix is the problem rather than the posts, rebuild it with content pillars — the grouping above is a diagnostic, not a planning framework.

Step 5: The competitor pass, kept honest

You can see a competitor's public output and the public engagement on it. You cannot see their reach, ad spend, follower sources or results, so any audit reporting a competitor's "performance" is reporting a guess. Keep the pass to what is observable:

• Which platforms they maintain — and which they have quietly abandoned.
• What jobs their posts do, using the same four labels. The job they own is the gap you are fighting.
• Their publishing rhythm — a fact about them, not a target for you.
• What their comments say. Repeated questions under their posts are demand nobody is answering well — tracking that continuously is social listening, not auditing.

Write the section as observations, with "we cannot see this" stated plainly where it applies. It makes the rest of the audit credible.

What we find in almost every audit

The uncomfortable part of that second point: follower count is sunk cost. If a platform has produced no business outcome in a year of honest effort, the followers are not an asset — they are the reason you keep paying for the effort.

The question-to-decision table

The whole audit in one view.

Where this ends: the one-page decision sheet

The deliverable is one page, one row per account, and it is readable by someone who was not in the audit.

1. One row per account from the inventory, orphans included.
2. Six columns: account · reach and engagement rate for the window (denominator stated) · follower change · dominant content job · keep / fix / close · the one change to make next.
3. One verdict per row. Keep: it works, the plan stands. Fix: one named problem, one named change. Close: stop posting, archive the content, then delete the profile or leave it with a pinned post pointing at the live account.
4. One change only, per row — not a list of nine. "Move Reels to a three-second visual hook" is a change; "improve content quality" is not.
5. An owner and a review date against each row. An audit with no owner becomes a document instead of a decision.
6. A "what we could not see" line: missing access, metrics outside the retention window, competitor reach.

Expect the sheet to be uncomfortable. If every row says keep, you ran an inventory, not an audit. Then give the fixes time: changes to organic distribution and content mix read over weeks, not days, and the next audit compares the same 90-day window, measured the same way.