Which Social Media Metrics Are Worth Reporting
Most social media reports are long because length feels like effort. They carry impressions, follower growth and a screenshot of the best post, and they change nothing. A useful report is short and uncomfortable. It answers whether the work is producing customers, and it names the decisions the numbers now demand.
The only test a metric has to pass
A metric earns its place if a plausible change in it would change what you do next. That is the whole test. If reach falls twelve per cent and your response is identical either way, reach is not a reporting metric for you. It might still be a diagnostic you check inside the platform, but it does not belong on the page a director reads.
Apply that test honestly and most monthly reports shrink by two thirds. Follower count usually goes first. It moves slowly, it responds to things you cannot control, and no sensible decision hangs on it. Impressions usually go second, for the same reason: the number is large, it looks like progress, and it is mostly a function of how much you posted.
Engagement rate is not one number
Engagement rate is the metric people most often assume is standard. It is not. Every platform defines both the numerator and the denominator differently, which means a figure from one platform cannot be compared with a figure from another, and a figure from a third party tool may match neither.
X states in its post activity dashboard documentation that engagement rate is "the number of engagements divided by the number of impressions", and defines engagements very broadly: "Clicks anywhere on the post, including reposts, replies, follows, likes, links, embedded media, username, profile photo, or post expansion." A profile photo click counts.
LinkedIn's Marketing API documentation defines its engagement field as "Number of organic clicks, likes, comments, and shares over impressions." Clicks are in, but the click definition is narrower than X's, and the field returns a decimal rather than a percentage.
Meta's Instagram Platform documentation for insights defines engagement on a media object as "Total number of likes and comments on the media object", with reach as the count of unique accounts that saw it and impressions as total views. Likes and comments only, in that particular definition. Saves and shares are counted separately.
Google Analytics 4 uses the same words for something else entirely. Google's Analytics Help defines engagement rate as "the percentage of engaged sessions" on a site or app, and an engaged session as one that lasts longer than ten seconds, has a key event, or has two or more screen or page views. Bounce rate is simply the inverse.
Four platforms, four meanings, one word.
| Platform | What counts as engagement | Divided by |
|---|---|---|
| X | Clicks anywhere on the post, including reposts, replies, follows, likes, links, media, username, profile photo, post expansion | Impressions |
| LinkedIn (API) | Organic clicks, likes, comments, shares | Impressions |
| Instagram (Platform insights) | Likes and comments on the media object | Not divided; reported as a count |
| Google Analytics 4 | Sessions over ten seconds, or with a key event, or with two or more page views | All sessions |
The practical consequence is simple. Pick one definition per platform, write it down at the top of your report, and never change it mid year. A benchmark you set yourself and hold steady is worth more than an industry average calculated by a method you cannot see.
Which metric answers which business question
The useful reframe is to start from the question rather than the dashboard. If you cannot name the question a metric answers, cut it.
| Business question | Metric that answers it | Metric that does not |
|---|---|---|
| Are we reaching people who do not already know us? | Reach from non followers, new accounts reached | Total impressions |
| Is the content actually holding attention? | Average watch time, retention curve, saves | Likes |
| Are we building an audience we can address again? | Follower to reach ratio, email or WhatsApp signups from social | Follower count alone |
| Is social sending qualified traffic? | Sessions from social with engagement rate and key events, in GA4 | Link clicks reported in platform |
| Is social producing revenue or enquiries? | Assisted and last click conversions, enquiry volume tagged to source | Engagement rate |
| Which creative should we make more of? | Saves and shares per post, watch through rate by format | Post with the most likes |
| Are we posting enough to matter? | Reach per post held steady as frequency rises | Number of posts published |
Saves and shares deserve particular attention. They are the two actions that most reliably indicate a person found something worth keeping or worth being seen endorsing, and they are the actions most closely tied to distribution. Instagram's own explanation of ranking lists likes, comments, shares and saves among the popularity signals it reads from a post, alongside how often people have interacted with that account recently.
The best time to post on Instagram, honestly
This is the most searched question in social media and the least useful answer in most reports.
Instagram has published a plain account of how ranking works, and it makes two things clear. There is no single algorithm: "Instagram doesn't have a singular algorithm that oversees what people do and don't see on the app." And for Feed, the four signals that matter most are your activity, information about the post, information about the person who posted, and your history of interacting with them. Posting time appears as one input among the post information signals, not as a controlling factor.
That matters because it means a universal best time cannot exist. Ranking is personalised per viewer. What a global average can tell you is when large numbers of people happen to be active, which is a weaker claim than most articles make for it.
Sprout Social's analysis, published on 31 March 2026, is one of the few with a stated method: nearly two billion engagements across roughly 307,000 social profiles between 27 November 2025 and 27 February 2026. It reports peak Instagram windows on Mondays from 2pm to 4pm, Tuesdays from 1pm to 7pm, Wednesdays from 12pm to 9pm and Thursdays from 12pm to 2pm. The same analysis says plainly that your own audience dictates the outcome and recommends finding custom times from your own data.
So use the global figure as a starting hypothesis for a new account with no history, then replace it within six weeks. Your own account data beats any average because it is the only data drawn from the exact people you are trying to reach.
Falling reach is not automatically your failure
Before you rebuild a content strategy on a declining engagement rate, check whether the denominator moved for reasons outside your account.
Ofcom research published on 02 April 2026 found that around half of adult social media users in the UK, 49 per cent, now actively post, share or comment, down from 61 per cent in 2024. The proportion exploring new websites fell from 70 per cent to 56 per cent over the same period. Eighty nine per cent of adult internet users still use at least one platform, so the audience has not left. It has gone quieter.
If a smaller share of the same audience takes any public action at all, engagement rates fall across the board without any change in content quality. That is a reason to weight your reporting towards reach, watch time, saves and downstream site behaviour, and to stop treating comment volume as a proxy for interest.
Ofcom's wider media use research, covering 2025, put UK adults at around four and a half hours online per day, with YouTube reaching 94 per cent of adults and Facebook or Messenger 93 per cent. Attention is still there. Visible interaction is what has thinned.
Reporting cadence & the one page rule
Three rules keep a social report honest.
- One page, monthly. Anything longer stops being read. Put the three metrics tied to commercial outcomes at the top, the diagnostic detail below, and the raw export in an appendix nobody has to open.
- Always show the comparison. A number without a prior period, a target, or a benchmark is decoration. Show the same month last year where seasonality matters.
- End with decisions, not observations. "Reels retention drops at four seconds, so we are cutting the intro" is a report. "Reels performed well" is not.
Set the review cadence to match the decision cycle. Weekly numbers on organic social are mostly random variation. Monthly is usually right for content decisions, quarterly for strategy.
What to do next
- Open last month's social report and delete every metric that would not change a decision. Note what survives. If fewer than four metrics remain, that is a healthy result, not a problem.
- Write down the exact engagement rate formula you use for each platform, using that platform's own definition, and fix it for the next twelve months.
- Check in Google Analytics 4 whether social traffic is tagged properly and whether social sessions hit your key events. If the tagging is broken, no amount of platform reporting will tell you whether social produces revenue.
- Pull your own top performing posting times from native insights and run a six week test against the global averages. Keep whichever wins on reach from non followers.
- Add one line to the top of the report naming the single decision you made because of the data. If you cannot fill it in, the report is not working yet.
Good reporting is mostly subtraction. The value comes from deciding what not to measure, then holding that decision long enough for trends to mean something. If you would like a second opinion on what your social reporting is actually telling you, our social media and content work usually starts by cutting the report in half.
Sources
- Instagram, Instagram Ranking Explained
- Meta for Developers, Instagram Platform Insights
- X Business, Post and Video Activity Dashboards
- Microsoft Learn, LinkedIn Organization Share Statistics
- Google Analytics Help, Engagement rate and bounce rate
- Sprout Social, Best Times to Post on Instagram
- Ofcom, Passive social media use, AI companionship and online side hustles
- Ofcom, From apps to AI search: how the UK goes online in 2025
