Measure a character beyond follower count.

Editorial analysis

Followers are stored attention, not delivered value

A follower count describes an account relationship accumulated over time. It does not show how many people noticed this release, understood its point, visited the intended destination or completed the action the operator wanted. It can still be useful context, but it should not be the denominator for every decision.

Choose one job for each campaign: introduce the character, deepen a recurring audience, collect qualified interest, sell an item or help a sponsor explain a product. The scorecard should follow that job. Otherwise a large reach number can make a weak commerce test look successful, while a small specialist audience can look weak despite producing useful inquiries.

Source record

Read platform metrics in context

YouTube's current guidance says impressions and click-through rate should not be read alone. Traffic sources carry different viewer intent, and even strong click-through and average-view-duration performance may not produce continuing impression growth. That makes a universal good CTR target a poor editorial shortcut.

Use platform measures as stages: eligible exposure, a view definition, watch depth, repeat or regular audience, and meaningful responses. Record the platform's definition and export date. Do not merge views from unlike formats unless the comparison explains those differences.

Evidence: YouTube Help [s1]

Source record

Give every outbound link an identity

Google Analytics documents campaign parameters including source, medium, campaign and content. It also warns that values are case-sensitive, so inconsistent naming can split one effort across several report rows. Establish the naming table before publication and preserve it with the brief.

For example, use `utm_source=youtube`, `utm_medium=organic_social`, `utm_campaign=atelier_launch` and `utm_content=episode_01_endcard`. A paid reuse should receive a different medium and creative value. This does not solve attribution; it prevents an avoidable naming failure.

Evidence: Google Analytics Help [s2]

Practical application

Work a small example from exposure to contribution

Assume a fictional episode records 80,000 eligible impressions, 4,000 views under the platform's definition, 1,000 viewers reaching the product mention, 240 tagged site sessions, 36 email sign-ups and 9 paid preorders. The derived rates are 5% views per impression, 25% product-mention reach per view, 24% sessions per reached viewer, 15% sign-ups per session and 25% preorders per sign-up.

Now add economics. If nine orders contribute an estimated $18 each after product, payment and variable fulfillment costs, modeled contribution is $162 before fixed production cost. If the episode cost $900 to make, this isolated sales window did not recover production cost. That does not prove the character failed: awareness, later purchases and sponsor value are unmeasured. It does show why nine orders cannot be turned into a profitable claim without a defined window and cost boundary.

Practical application

Run the review in three columns

Column one is observed: exported counts, dates and definitions. Column two is calculated: formulas applied to those counts. Column three is interpreted: the team's explanation and next test. Flag missing consent, unattributed traffic, refunds and delayed conversions. Compare like formats and similar windows, then change one major variable at a time.

End with a decision: continue unchanged, revise the creative, change the landing page, narrow the audience, or stop. A measurement meeting that produces no decision is a reporting ritual. A modest scorecard that identifies the next reversible test is more valuable than a dashboard built to make the operator profile look large.