Influencer Marketing
The creator casting scorecard that forces better judgment
A scorecard cannot choose talent for you. It can expose why the team is seduced by reach, aesthetics, or one unusually polished post.
The founder loves candidate A. The social lead says candidate B feels more native. Performance wants candidate C because a database shows a higher average view rate. Everyone has a reason, but the reasons live at different levels: taste, channel craft, audience fit, and commercial evidence. The meeting ends with a compromise shortlist nobody fully believes in.
A casting scorecard is useful when it makes those reasons inspectable. It should not manufacture a decimal-point truth about people. It should force the team to define the role, review comparable evidence, identify risk, and record the judgment that a dashboard cannot supply.
Start every card with the campaign role
Write one sentence: 'We are considering this creator to…' The ending might be explain a technical mechanism, introduce the brand to a subculture, produce licensable demonstrations, host a live conversation, or make the launch feel unavoidable. If the sentence contains four unrelated jobs, split the evaluation.
Weight criteria by role. Audience authority may dominate an educator brief. Performance range and visual proof may dominate an asset brief. Reach may matter for a launch signal. A fixed universal weighting rewards the same creator profile regardless of what the campaign needs.
Attach evidence to every meaningful score
For audience fit, cite recurring topics, comment behavior, geography, language, and prior category response. For creative fit, link examples that show hooks, product handling, narrative control, and sponsorship integration. For reliability, use past delivery records or reference checks where available—not a vague impression from email tone.
A number without evidence creates false alignment. Two reviewers can both choose four out of five while imagining different standards. Short notes expose the gap and become useful later when performance confirms or contradicts the original casting logic.
Score the dimensions that survive a beautiful feed
Useful dimensions include role credibility, audience relevance, content behavior, performance range, product demonstration, brand integration, channel fluency, reliability, rights feasibility, disclosure history, reputational fit, cost, and portfolio overlap. Not every dimension needs a five-point scale. Some are facts, flags, or open questions.
Keep follower count and average performance as context rather than destiny. Review variance, recent trajectory, sponsored versus organic behavior, and whether the creator's best posts resemble the work being requested. A mean can hide one viral outlier or a format the brand cannot reproduce.
Preserve a written human judgment
After scoring, require the evaluator to write why this person, why now, and what could make the match fail. This catches candidates who score well across safe dimensions but lack a compelling reason to exist in the campaign. It also protects unconventional choices that a generic model would underrate.
Discuss disagreement instead of averaging it immediately. The social lead may see native format behavior the data team missed. Legal may recognize a claim pattern. A market lead may spot social meaning invisible to a global reviewer. The scorecard is a shared surface for expertise, not a machine for overruling it.
Calibrate the scorecard with outcomes
After each campaign, compare predicted strengths and risks with delivery, content quality, audience response, paid performance, rights friction, and relationship value. Update examples of what a strong or weak score looks like. Do not simply reward whoever generated the largest headline number.
Casting is a learning system. Over time, the team should become better at recognizing which signals predict success for different jobs. A useful scorecard preserves that institutional memory while leaving room for the one creator who is difficult to quantify and exactly right.
The scorecard should expose the judgment, not impersonate it.
