Social Media Benchmarks: Build Your Own Baseline Instead

June 10, 2026 · 8 min read · by the Kadenzo team

Every few months someone asks us what a good engagement rate is, and the honest answer never changes: good compared to what? Industry benchmark tables answer "how is everyone doing" — a population average that blends every niche, audience size, and content mix into one number that describes nobody in particular. The bar your account should actually clear is its own recent history: a trailing 90-day baseline, recalculated monthly. This guide is the exact method — which metrics to baseline, the 30-minute setup, a worked example, and how to turn the baseline into targets you can defend in a planning meeting.

Why industry benchmarks keep misleading you

Benchmark reports aren't wrong; they're just answering a different question. Three things break when you treat a population average as a personal target:

Niche. A B2B compliance-software page and a streetwear account live on the same platform and different planets. The blended "average engagement rate" sits somewhere between them, reachable by neither as a meaningful goal.
Audience size. Small accounts run structurally higher rates — a 400-follower page where 60 regulars engage will post percentages a 400,000-follower brand can never touch. Most tables average across the whole range.
Methodology. The table's rate and your rate are often different formulas wearing the same name — by followers here, by reach there, saves counted or not. Comparing them is a category error before it's an accuracy problem (our engagement rate calculator prints its formula under every result for exactly this reason).

The result: accounts chase a number that was never theirs to chase, and report meetings argue about the bar instead of the work.

The baseline: trailing 90 days, refreshed every 30

A personal baseline is simple to state: for each metric you care about, the average of your own posts over the last 90 days. Two numbers in that sentence do the heavy lifting:

90 days back is long enough to smooth luck — algorithm wobbles, one viral outlier, a slow holiday week — without averaging in an account you no longer are.
Recalculated every 30 days keeps the bar moving with the account. Last quarter's baseline judging this quarter's work is how growth gets invisible.

Baseline four to six metrics, no more. A workable default set:

Engagement rate per post — mean of per-post rates, by reach where the platform reports it.
Reach per post — use the median, not the mean; one viral post drags a mean for a quarter.
High-intent interactions per post — saves + shares (or bookmarks + reposts), the signals that predict distribution.
Link clicks or tracked sessions per post — the business bridge.
Follower delta per month — kept last, as context, never as the headline.

Build it in thirty minutes

Export the last 90 days of posts from each platform's analytics (every platform exports per-post reach/impressions and interactions).
Compute the per-post engagement rate — paste each post's numbers into the calculator or apply the same formula in your sheet, one denominator everywhere.
Take the mean (rates, interactions, clicks) or median (reach) per metric.
Write the baseline row — one row per platform, dated. This row is now "normal."
Annotate outliers instead of deleting them. A post that reached 5× your median stays in the export with a note. If it distorts a mean badly, show the metric with and without it — both numbers, labeled.
Repeat on the first of every month and keep the old rows. Twelve baseline rows later, you own the only benchmark report that describes your account.

A worked example — a B2B LinkedIn page, baseline computed April 1 from 26 posts:

Baseline (Jan–Mar): engagement rate 3.4% by impressions · median impressions 4,100 · reactions+comments+reposts 19 per post · link clicks 31 per post · +180 followers/month.

April actuals: 3.9% (+0.5 pt) · median impressions 3,800 (−7%) · 24 interactions per post (+26%) · 44 clicks (+42%).

The reading: resonance and traffic are up while raw distribution dipped — the content is working harder than the algorithm is. Double down on the formats driving it; don't panic about reach. No industry table can produce that sentence.

What industry tables are still good for

Three legitimate uses, all orientation rather than judgment:

Day one. A brand-new account has no history; a public report like Rival IQ's annual benchmark study tells you which neighborhood to expect while your first baseline accumulates. The ranges on our calculator page serve the same purpose.
Client education. When a client arrives expecting 10% engagement on a 200k-follower account, a published industry median is a neutral way to reset expectations before your baseline takes over.
Platform weather. When your reach drops and every report says the platform's median dropped too, that's a platform story, not a performance story — worth one annotating sentence in the monthly report.

From baseline to targets

A baseline describes; a target commits. The discipline is to make every target a baseline plus a believable delta tied to a named change: "engagement rate 3.4% → 3.8% this quarter, because we're shifting two posts a week to the carousel formats that out-performed in March." A target with no attached change is a wish with a deadline.

Two framings keep targets honest. Use a beat rate — "beat the baseline in at least three weeks of four" — rather than demanding every single post clear the bar; content has variance, and a beat rate absorbs it. And re-baseline after step changes: a press mention that adds 5,000 followers overnight resets what normal means — date the epoch, start the new baseline, and stop comparing across it.

Keep it honest

The baseline method fails in the same three ways every measurement system fails, and they're all preventable:

Silently dropped outliers. Annotate, never delete — the moment a stakeholder discovers a quiet exclusion, every past number is suspect.
Moving windows. 90/30, every month, no exceptions; a window that flexes to flatter is not a baseline.
Mixed methodologies. One denominator, one formula, written down — the same discipline that makes a monthly report defensible makes a baseline comparable across months.

Run it for a quarter and the change is hard to unsee: charts stop being verdicts handed down by someone else's average and become trend lines against your own. That's the entire point of measuring anything.