Measuring social — and sizing up a platform you've never used

The number Ana watched for four years

Open Ostara's funnel baseline from Module 1: 6,500 sessions, 91 orders, 19% returning. Now look for the number Ana checked most often over four years — the follower count. It isn't there. It was never there. 14,200 followers appears in no row of the arithmetic that pays the studio's bills.

That is the whole lesson, in one observation. Social produces a torrent of numbers, and most of them exist because platforms like showing you numbers that go up. Your job is to keep the few that connect to the funnel you baselined in Module 1 and let the rest go.

The test a metric must pass

A metric earns a place on your scorecard only if a different value would change what you do next week. That's it. Run every candidate through it:

• Follower count — if it were 12,000 instead of 14,200, what would Mateus do differently on Monday? Nothing. It fails. (It has one honest use — a rough proxy for permission-to-audition, checked quarterly, never optimised.)
• Tagged link clicks — if clicks from social halved, he would change the calls-to-action and the content mix that week. Passes.
• Saves and shares per post against his own rolling average — a post far above average is a format to repeat and a paid candidate under Lena's 7.3 rule; far below, a format to retire. Passes.
• Email signups sourced from social — the entire canonical job M5.L3 gave Ostara's Instagram. If this stalls, the strategy is failing whatever the likes say. Passes.

Vanity metrics are not evil; they are undirected. They go up without telling you what to do. Decision metrics come with instructions attached.

Two definitions, one formula

Reach is unique people who saw a post; impressions are total displays — impressions ÷ reach is frequency, a number that matters far more in paid (Module 9) than organic. Engagement rate is interactions ÷ reach × 100, per post. The only benchmark that matters for it is your own rolling average — cross-account comparisons founder on audience size, category and platform accounting differences, and the "industry benchmark" tables floating around rarely survive a question about method. You are not trying to beat ceramics-industry Instagram. You are trying to beat last month's Ostara.

The click bridge

Here is the mechanical heart of the lesson: without tagged links, social is a faith-based channel. A click from a feed to your site arrives, by default, as barely labelled traffic. Add a source tag to every link you control — the ?src= move Storkflow made canonical in Module 4, or full UTM parameters — and social becomes rows in your analytics: sessions, then conversions, then euros.

Mateus's first full month of tagging is the canonical demonstration, and it re-priced everything. Of a typical month's 6,500 sessions, roughly 2,900 arrive from organic Instagram — the bio link, stories, the email pushes — and about 1,400 from Meta ads, which mostly retarget people Instagram already brought. Numbers Ana half-suspected but had never seen: some 44% of sessions, feeding the sitewide 1.4% conversion rate, all through one rented room. The dependency the whole Program has circled is now a measured fact on a dashboard, not a feeling — and diversifying it has a denominator. Nobody had to lecture anyone.

The bridge also settles smaller questions honestly. The Émilie collaboration from 7.4: 312 tagged sessions, 9 orders (€702 at the €78 AOV), 41 email signups in its first fortnight. Set against the €370 cost: at Ostara's 58% margin, roughly €407 of gross margin — about breakeven on first orders, before counting 41 owned addresses, a licensed Reel, and any repeat purchases. That is what an honest creator evaluation looks like: not "it went well", but arithmetic, with the unmeasurable parts named as unmeasurable rather than claimed.

Leading and lagging

Saves, shares and link clicks move within hours. Enquiries, orders and bookings lag by days or weeks — a considered gift buyer saves in March and buys in May. So run two clocks: judge content weekly on leading metrics (repeat what's working while it's working), and judge the channel quarterly on lagging ones (sessions, conversions, email growth from social). Judging content on lagging metrics makes you slow; judging the channel on leading ones makes you delusional. Both mistakes are common, and they are usually made by the same person in the same meeting.

Your scorecard, then: five metrics at most, each passing the decision test, each with a numeric target and a review cadence. Fewer is better. A scorecard you check weekly beats a dashboard you admire monthly.

Sizing up a platform you've never used

Now the skill the module has been building towards, because the platform that matters most to your business in three years may not exist today — and when it launches, there will be no lesson about it anywhere.

You met the New Platform Evaluation framework in 7.1. Here it runs at full length, on a platform we invented so that the exercise cannot rot: Loom.social — fictional, video-first, sound-off, 20-second cap, popular with under-25s in creative professions. Ana runs the five questions:

1. Audience & mindset. Two hours of reading finds interior-design students and young renters showing flats — ICP-adjacent, but the gift-buying mindset her beachhead runs on is unverified. Honest answer: plausible, unproven.
2. Reach window. Young platform, content supply scarce relative to attention — the one structural moment when organic reach is generous (7.3's economics, running in reverse). A real argument to test early rather than late.
3. Format fit. Her packing format survives a 20-second sound-off grammar with captions. No new production burden. Passes.
4. True cost. Two hours a week at minimum cadence, inside Mateus's twelve. Survivable.
5. Measurability & risk. No outbound links. The only bridge is a code shown on screen. Measurement nearly blind; platform longevity unknown.

Verdict: a contained test — six weeks, two posts a week of repurposed material, judged on code redemptions and profile-search traffic, with a kill criterion written before the first post: stop if fewer than 10 redemptions by week six. Lena runs the identical framework for Storkflow and declines in one paragraph: her ICP is not there in any mindset, and no reach window compensates for an absent audience. Same tool, opposite verdicts, both correct — which is exactly what a framework is for. When the real next platform launches, you will not need a course about it. You will need two hours of reading and these five questions.

---