/influencer-discovery-brief

Name

influencer-discovery-brief

Description

Define creator sourcing criteria and produce the discovery brief used to shortlist candidate influencers and creators - weighted criteria for audience fit, engagement quality, niche consistency and brand safety, knockout screens, and a ranked shortlist scoring template with evidence per score. Covers B2C brand campaigns and B2B creator and thought-leader programs. Use whenever the user mentions finding creators, vetting influencers, audience fit, creator shortlists, or deciding which creators to approach, even if they never say discovery. Runs before any contact. Do NOT use for writing the pitch message itself - use mbfinotti/partnerships-skills@influencer-outreach instead.

License

MIT

author

Maya-Beth Finotti

version

1.1.1

# Influencer Discovery Brief

Define the sourcing criteria and produce the discovery brief a partnership manager uses to shortlist candidate creators - who qualifies, by what weighted criteria, and with what evidence.

The "3 Rs" (Reach, Relevance, Resonance) is a genuinely recurring industry convention for judging influence, but research could not trace it to a canonical originator - treat it as a floating mnemonic, not a citable method. A "4 Rs" variant adding "Reference" is one influencer-analytics vendor's framework, not an industry standard. This skill therefore works from explicit weighted criteria the manager approves, never from a named model presented as authoritative.

## Interview

Ask before sourcing anything. One question per message; offer multiple-choice answers when possible. Skip any question the user already answered.

- Is this a B2C brand campaign or a B2B creator/thought-leader program? (Or both - then weight criteria separately per track.)
- What is the brand/product, and what is the value prop a creator's _audience_ would actually care about?
- What is the campaign objective: awareness, conversion, content-for-paid, or trust-building? The objective changes the criteria weights, so answer before any scoring.
- Who is the target _customer_ - demographics, role, buying context? (Not the target creator; the creator is a channel to this person.)
- Which platform(s) does the campaign run on?
- Which geographies/markets actually convert for this product? (Audience anywhere else is reach you pay for and can't use.)
- What is the budget tier, and is compensation cash, product, or a mix?
- How many creators will actually be contracted from the shortlist?
- What are the brand-safety redlines, and is the product in a regulated category (alcohol, health, finance, gambling, kids)?
- Any competitor or category-exclusivity conflicts to enforce? Who is already on the roster? (Needed for audience-overlap checks.)
- What audience data can you actually obtain: creator discovery platform exports, audience-insights exports, media kits, or public profiles only? (A licensed discovery platform moves its export pass from the slowest sourcing option to the fastest. Public profiles only deletes the two-stage weighting gate from the menu.)
- By what date must the shortlist land, and what is waiting on it? (A date inside a week promotes the roster pull, the customer-mention pass, and the floor-gate weighting. A quarter makes social listening and full objective-led scoring affordable.)
- One-off campaign or a creator program you will run again? (A one-off stops at a shortlist. A repeating program justifies the full objective-led scoring pass, whose approved weights are the only output the post-campaign fit-rate KPI can tune.)
- What is the effort ceiling - your own hours, whose data you can request, and whether you can commit to a standing monitoring job? (A solo ceiling deletes social listening. No budget owner for a platform licence deletes discovery-platform exports.)

Carry these last three answers into every ranking below. They are the only inputs that reorder the defaults for this manager.

Every ordering in this skill is a default, not a law. It ranks by qualified candidates returned per unit of effort, which is never the same ordering as cheapest-first, and it shifts with context and with who runs the pass.

Re-rank against what you already know about this manager before recommending anything. Say which answer moved which pass, so the reordering is auditable rather than a preference you asserted.

- A discovery platform already licensed moves its export pass to the front.
- An existing creator roster with performance history makes the roster pull the strongest pass rather than merely the cheapest.
- A niche too small for platform databases deletes the export pass outright and promotes customer mentions and social listening in its place.

## Workflow

1. Run the Interview; collect every answer before looking at a single candidate.
2. Translate the objective into a draft weighted-criteria set - awareness, conversion, content-for-paid, and trust-building each weight differently. See [references/sourcing-criteria-and-weighting.md](references/sourcing-criteria-and-weighting.md).
3. Enter brainstorming mode: present the weighting approaches in ranked order, lead with the recommendation, and give each one's trade-off in the manager's own vocabulary (criteria, floors, knockouts, shortlist). Full trade-offs in [references/sourcing-criteria-and-weighting.md](references/sourcing-criteria-and-weighting.md).

   - efficiency: two-stage audience-fit gate > objective-led fixed weights > floor-gate plus equal weights, wherever one audience-fit figure can be obtained per candidate; floor-gate leads and the other two are unaffordable where the only data is public profiles.

   Default to the two-stage gate. Drop to the floor-gate when the data answer was public profiles only, or the date answer leaves under a week. Pay the full objective-led scoring pass at least once when the answer was a repeating program - it is the only approach that produces tunable weights.

4. Ask brainstorming questions one at a time; score nothing until the manager explicitly picks a weighting approach.
5. Build the raw candidate pool through sourcing passes. Six passes exist; treat them as a budget, not a checklist. Run them in this order and stop once the pool is deep enough to survive the knockouts.

   - **Existing roster and CRM** - an hour, no tool access. The only pass returning performance history with _your_ brand: who already posted, what it did, who the manager would sign again. It returns no net-new names, and step 9's overlap check needs it regardless.
   - **Customer and community mentions** - a day pulling support and community threads, review text, and a "which creators do you follow" question into a survey already going out. Names arrive with fit evidence attached, because your buyers demonstrably watch them, and this is the only pass reaching podcasts, newsletters, and closed communities no platform indexes.
   - **Discovery-platform exports** - near-zero with a licence in hand: one filtered query returns breadth plus the audience-demographic export step 8 scores against. Without one, a quarter and a procurement conversation. Covers only what the vendor indexed - thin in narrow B2B niches and on non-social formats.
   - **Competitor collaborations** - an hour browsing competitors' tagged and disclosed posts. Pre-qualified: they take sponsorships, they are in category, and their sponsored-versus-organic engagement is visible before you make contact. Every competitor works the same list, and the exclusivity knockout removes part of it.
   - **Niche keyword and hashtag search** - an hour that keeps being an hour, no tool access. It mostly re-derives by hand what the export returns, at a low qualified-per-hour rate; it earns its place only where no platform covers the niche.
   - **Social listening** - a week to design queries and a standing job to maintain. It surfaces creators already talking about the category unprompted, the strongest unique-name yield of any pass, but it is a monitoring system whose discovery output is a by-product.

   - effort: social listening > discovery-platform exports without a licence > customer and community mentions > competitor collaborations == hashtag search (tied: both an hour of manual browsing, no tool access, no one else's permission) > roster and CRM == discovery-platform exports with a licence (tied: both an hour spent querying data you already hold)
   - compliance cost: customer and community mentions (a privacy review before repurposing support, community, or survey data into a list you build a campaign from) > hashtag search run by scraper (a platform-ToS breach that gets the account banned - near-zero read by hand) > discovery-platform exports (the vendor licence governs what you may re-export and retain) > roster and CRM (past contracts can carry exclusivity or re-approach clauses) == competitor collaborations == social listening (no exposure)
   - value, as qualified creators the other passes would have missed: customer and community mentions > social listening > discovery-platform exports > competitor collaborations > roster and CRM > hashtag search
   - efficiency: roster and CRM > customer and community mentions > discovery-platform exports > competitor collaborations > social listening > hashtag search

   **Always run:**

   - The roster pull and the customer-mention pass together - they cost a day and return the two kinds of name nothing else does.

   **Add next, in this order:**

   - The discovery-platform exports third; move it to first whenever the licence already exists - the same query then costs an hour and also supplies the scoring data.
   - Competitor collaborations, when the pool is still thin after three passes.
   - Social listening, the moment the answers said a repeating program and someone will own the standing job.

   **Delete rather than demote:**

   - The scraper variant of hashtag search: it breaks platform terms and the banned account takes the manual pass down with it.
   - The export pass entirely, when the niche has no platform coverage and no licence exists.

   Social listening is what this order starves - second on value, first on effort, so it loses every efficiency round and a manager following the ratio alone never builds the one pass that surfaces creators nobody else has found.

6. If you can browse the web, verify each candidate's public profile yourself; otherwise ask the manager to paste discovery-platform or audience-insights exports. Never score from memory or assumption.
7. Screen every candidate against the knockout checklist _before_ any scoring; a knockout failure goes straight to the rejected list with its reason. See [references/brand-safety-screening.md](references/brand-safety-screening.md).
8. Score survivors per criterion with the approved weights, attaching one evidence source per score. Signals in [references/audience-fit-and-engagement-signals.md](references/audience-fit-and-engagement-signals.md); tier and niche mapping in [references/creator-tiers-and-niche-mapping.md](references/creator-tiers-and-niche-mapping.md).
9. Rank by weighted total; check audience overlap across the shortlist and against the existing roster; demote candidates duplicating an audience already paid for.
10. Self-check the brief against the Quality gate below; iterate until every item passes.
11. Present the brief section by section - header, criteria, knockouts, shortlist, rejected list - validating each with the manager before finalizing.
12. If your harness has persistent memory, memorize the approved criteria weights, redlines, and target-audience definition - later runs skip straight past the interview.

Stop at the ranked shortlist. Outreach messages and the creative brief handed to a chosen creator belong to sibling skills, not this one.

## B2C vs B2B - where criteria diverge

- B2C: weight audience demographics, geography, and aesthetic/content fit; follower tiers carry real meaning (nano/micro for engagement and conversion, macro/mega for reach).
- B2B: weight the creator's job title and employer credibility, who is in the comments (do commenters match the ICP?), speaking and publication history, and newsletter subscriber quality. Treat raw follower counts as near-worthless. The guidance above is practitioner reasoning: state that caveat in the brief when a B2B manager asks where the weights come from.
- Identical for both: audience-fit-over-creator-fit, geography verification, brand-safety screening, audience-overlap checks across the roster, and the discipline of agreeing weights _before_ looking at candidates.

## Expected output

Deliver one structured artifact (full worked example in [references/examples.md](references/examples.md)):

```
DISCOVERY BRIEF - <brand>, <date>
Header    : brand, objective, target customer, converting markets, platforms
Criteria  : table - criterion / weight / how measured / pass floor
Knockouts : screening checklist applied before scoring, with redlines
Shortlist : table - candidate / score per criterion / weighted total / rank
            / evidence link per score
Rejected  : candidate + one-line reason (knockout failed or floor missed)
```

Every score cites its evidence; every rejection states its reason. A brief the manager can't audit line by line is not done.

## Quality gate

Check the finished brief against every item below, verified for every shortlisted candidate. Pass threshold: 7/7. Iterate - rescore, re-source, re-screen - until nothing fails.

These thresholds are working conventions for auditable shortlists, not measured industry standards.

1. Weights were explicitly approved by the manager before the first candidate was scored.
2. Every knockout criterion was applied to every candidate before scoring began.
3. Every scored cell carries a cited evidence source (profile link, export, or pasted data) - no memory-scored cells.
4. No candidate was scored on follower count alone; every shortlisted candidate has audience-fit and engagement evidence.
5. Audience geography is verified for every shortlisted candidate against the converting markets.
6. Audience overlap is checked across the shortlist and against the existing roster.
7. Every rejected candidate has a stated, specific reason.

## KPIs - shortlist quality, not campaign performance

Measure the discovery process itself. Campaign KPIs belong to the measurement skill.

No reliable published industry benchmarks exist for these ratios. Don't invent targets: establish your own baseline and track the trend.

- Screen-to-shortlist ratio: candidates screened per candidate shortlisted. Falling over time means sourcing passes are getting better targeted.
- Shortlist-to-contracted ratio: shortlisted candidates the manager actually signs. Persistently low means criteria don't match what the manager really wants - revisit weights.
- Criteria-weight stability: how much approved weights change between discovery cycles. Stabilizing weights mean the criteria are maturing; churning weights mean the objective is unclear.
- Post-campaign fit rate: share of shortlisted-then-contracted creators later judged a good fit. Feed it back into the weights.

- effort: post-campaign fit rate (a quarter of waiting plus a judgment collected per contracted creator) > screen-to-shortlist == shortlist-to-contracted == criteria-weight stability (near-zero - all three are counts the brief already produced)
- value: post-campaign fit rate (the only one saying whether the criteria themselves were right) > shortlist-to-contracted > screen-to-shortlist == criteria-weight stability (tied because neither judges the shortlist, only the pass that produced it, and neither reads at all until a second cycle exists)
- efficiency: shortlist-to-contracted > screen-to-shortlist == criteria-weight stability > post-campaign fit rate

Track the three free ratios from cycle one. Start collecting the fit rate in cycle one anyway - its lag is precisely why it never gets started, and it is the only signal that corrects a wrong criteria set.

## Common failure modes

| Failure                                                  | Fix                                                                                                       |
| -------------------------------------------------------- | --------------------------------------------------------------------------------------------------------- |
| Follower-count worship                                   | Weight audience fit and engagement first; cap or zero reach weight per the approved criteria              |
| Matching the creator's persona instead of their audience | Score who follows and comments, not who posts; sample the comment section against the target customer     |
| Audience-geography mismatch                              | Verify geo split for every shortlisted candidate; audience outside converting markets is unusable reach   |
| Engagement-rate gaming and pods                          | Apply the shallow red-flag checks; park unverifiable candidates as "authenticity unverified"              |
| Niche dilution (unrelated content categories)            | Categorize recent posts; enforce the on-niche share floor                                                 |
| Over-indexing on aesthetics/production quality           | Keep aesthetic fit a minor weighted criterion (B2C only), never a proxy for audience fit                  |
| Ignoring audience overlap across the roster              | Run the overlap check before ranking; demote duplicated audiences - never pay twice for the same eyeballs |
| Brand-safety screening after outreach                    | Knockouts run before scoring, which runs before any contact                                               |
| Scoring before weights are agreed                        | The brainstorming gate blocks scoring until the manager explicitly picks the weighting approach           |
| One platform's engagement formula applied to another     | Use each platform's own formula; never compare cross-platform rates directly                              |

## Invocation Examples

- "Build the discovery brief for a B2C skincare launch - we need 8 micro-creators, budget $40K, conversion objective."
- "We need B2B thought-leaders for a developer-tooling campaign. Define the vetting criteria and the knockout screens."
- "Score these 14 candidate creators against our brand-safety and engagement-quality criteria, with evidence per score."

## Reference

- See [references/sourcing-criteria-and-weighting.md](references/sourcing-criteria-and-weighting.md) when translating the objective into criteria - the criteria menu, objective-to-weight mapping, the ranked weighting approaches for the brainstorming gate, and pass floors.
- See [references/audience-fit-and-engagement-signals.md](references/audience-fit-and-engagement-signals.md) when scoring candidates - audience-fit tests, per-platform engagement formulas with benchmark reliability flags, shallow authenticity red flags, geography and overlap checks.
- See [references/creator-tiers-and-niche-mapping.md](references/creator-tiers-and-niche-mapping.md) when mapping tiers and niches - the verified tier taxonomy with its provenance, B2C tier use, B2B credibility signals, niche-consistency checks.
- See [references/brand-safety-screening.md](references/brand-safety-screening.md) when screening - the knockout checklist, regulated-category constraints, competitor-conflict handling.
- See [references/examples.md](references/examples.md) before writing the brief - a full worked B2C brief with scored shortlist, a B2B criteria-weighting variant, and an annotated negative example.
- See `mbfinotti/partnerships-skills@influencer-campaign-brief` for the creative brief handed to a creator after selection.
- See `mbfinotti/partnerships-skills@influencer-measurement-framework` for campaign KPIs; this skill measures only shortlist quality.
- See `mbfinotti/partnerships-skills@affiliate-disclosure-compliance` for the disclosure rules any contracted creator must follow.