How do you analyse your social performance?

Where to find the data, the weekly and monthly routine, the diagnostic method when a metric falls, and moving from analysis to decision

The essentials

  • Sources: the native statistics (Meta Business Suite, LinkedIn, TikTok Studio, YouTube Studio) for reach and engagement, GA4 for traffic and conversions, a consolidation tool to read it all in one place.
  • Routine: 15 minutes a week on the week’s posts, 1 hour a month on the trends, a quarterly review by pillar and by format.
  • Diagnostic method: isolate the variable that changed (pace, format, subject, time, algorithm) before concluding; compare with the 90-day median, not with the best post.
  • Expected output: three decisions a month at most, recorded with the date and the metric that will allow them to be judged.

Analysing your social performance does not mean opening the statistics and noting that “it went well” or not. It is a routine with defined sources, fixed questions and an output in the form of decisions. This page describes that routine for an SME or a provider, distinguishing what is read each week, each month and each quarter, then the method for diagnosing a fall without mistaking the cause. The choice of the metrics themselves is covered in choosing your KPIs by objective.

Where to find the data

SourceWhat you read thereDepth and exportLimit
Meta Business Suite (Facebook, Instagram)Reach, impressions, interactions per post, audience, stories, Reels90 days; CSV exportReach data not kept beyond 90 days in the interface
LinkedIn Analytics (page)Impressions, clicks, engagement per post, followers, visitors12 months; XLS exportNo reach in unique accounts per post
TikTok StudioViews, watch time, traffic sources, retention60 days in detailShort window; export every month
YouTube StudioViews, retention, sources, subscribers, thumbnail clicksUnlimitedShorts and long-form analysed separately
GA4Sessions, engagement on the site, conversions by source and UTM campaign14 months by defaultOnly sees the traffic going out to the site
Consolidation tool (Metricool, Agorapulse, GreenRed)All the networks and GA4 in a single view, history retainedDepends on the tool; cumulative historySubscription; check the plans

Two practical rules: export the native data every month to keep a history, and document the formula for each metric (“Meta” reach and the “reach” of a third-party tool are not always calculated the same way).

The weekly routine: 15 minutes

Each week, the reading covers only the posts of the last seven days:

  • the reach and engagement of each post set against the account’s median over the last 90 days (above, below, within the norm);
  • the comments and messages left unanswered;
  • the clicks to the site on the posts that contained a link;
  • a one-line note on each post that falls outside the norm, with the hypothesis (“carousel on pricing: saves x3”).

These notes feed the “result” column of the editorial calendar. Comparing with the median, rather than with the best post, avoids judging every piece of content against an exception.

The monthly routine: one hour

The monthly review looks at the trends and feeds the reporting. Five questions, in order:

  1. Was the planned pace kept up? A gap in pace explains most monthly reach variations on its own.
  2. How are the main KPIs changing compared with the previous month and with the same month last year, if seasonality plays a part?
  3. Which format and which pillar carried the month? Rank the posts by engagement by reach and by clicks; look at the top five and the bottom five.
  4. What does GA4 say? Social sessions, engagement rate of those sessions, attributed conversions, by network thanks to the UTMs; see social conversions in GA4.
  5. Which three decisions for next month? Recorded with the metric that will allow them to be judged.

The quarterly review adds a reading by content pillar and by network over 90 days, and a comparison with two competitors. It is at that scale that you decide to reinforce or drop a format or a network, not month by month.

Diagnosing a fall without mistaking the cause

Reach falling by 30% often triggers a hasty conclusion (“the algorithm has changed”). The method consists of eliminating the causes in order of how often they occur:

Possible causeHow to check itObserved frequency
Pace or consistency fallingNumber of posts per week over 8 weeksVery frequent
Change in the format mixShare of video and carousels, month against monthFrequent
Less engaging subjects (too many offers)Breakdown by pillar; engagement by pillarFrequent
Audience seasonalitySame period last year; holiday periodsRegular
Fall in early interactions (posting time, no replies to comments)Engagement in the first hour; response timeMedium
Platform changesComparison with similar accounts over the same period; official announcementsReal but rarely the only cause

The distribution mechanisms that explain these variations are described in the social media algorithm. A simultaneous fall across several comparable accounts signals a platform change; an isolated fall signals a change at your end.

Moving from analysis to decision

Analysis produces decisions, otherwise it serves no purpose. The typical decisions from a monthly review: increase the share of a format that outperforms, drop a type of post that pulls the median down, move a time slot, add a call to action on content that generates profile visits without clicks, fix a landing page whose conversion rate on social sessions is lower than that of the other channels. Three decisions a month is enough; each one is dated and tied to a control metric. The monthly report that follows runs to three to five pages; its content is detailed in the deliverables of an SMO provider and its general form in digital marketing reporting.

Our advice: keep a log of changes, one line per event: a new format, a change of time, the absence of a team member, a paid campaign. During a diagnosis, it is that log that lets you link a variation to its cause in two minutes, where the statistics alone leave room for guesswork.

How GreenRed helps

Rather than juggling several tools, GreenRed’s Social media tracking module brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.

Frequently asked questions

How often should you analyse your social media?

A quick 15-minute weekly reading of the week’s posts, a one-hour monthly review of the trends and the KPIs, a quarterly review by pillar, format and network. Analysing every day leads to reacting to statistical noise; analysing once a quarter loses the lessons of the posts.

Do you need a paid tool to analyse your social performance?

No for one or two networks: the native statistics and GA4 are enough, provided you export the data every month to keep the history. A consolidation tool becomes worthwhile from three networks, several people involved or reporting to be presented to a management team or a client.

How do you know whether a post worked well?

Compare its reach and its engagement by reach with the median of the last 90 days for the same account and the same format. A post above 1.5 times the median deserves to be analysed and adapted; a post below half the median deserves to be understood. The absolute number of likes says nothing on its own.

What should you measure in GA4 for social media?

Sessions by source and UTM campaign, the engagement rate and duration of those sessions, the landing pages and above all the key events (form, call, appointment) attributed to social. These figures are the only ones that link social activity to commercial results, provided you tag all outbound links.

Grow your social presence

GreenRed tracks your performance on Facebook, Instagram, LinkedIn and TikTok in a single dashboard.

Track my social media

Related articles