Rules
Part of Discovery platforms: a clear guide with practical examples
Discovery platforms examples: lessons and useful context
Four documents that make discovery platform work reusable: a search spec, a shortlist row carrying evidence, a searchable rejection log, and a handover note.
Four documents, filled in far enough that you can copy them. A search spec written from a brief, a shortlist row, a rejection entry, and the handover note that ends the discovery job. No names and no figures appear in any of them, because the value is in the columns rather than the contents, and the columns are what most teams never settle.
What to take away
- A search spec separates what a filter can express from what a person has to judge. Keeping the two in one document stops the second half being forgotten.
- Every row on a shortlist needs a reason, and every rejection needs one too. Reasons are what make the list reusable.
- Discovery ends with a handover. Write it, or the person doing the booking will start the research again.
One: a brief translated into a search spec
The brief is prose. The spec is two lists. Turning one into the other is requirements analysis applied to people rather than to software. Everything that a tool can filter on sits on the left, everything a human has to look at sits on the right, and nothing gets to be in neither.
| Machine conditions | Human conditions |
|---|---|
| Markets and languages | Whether the account sounds like the brand could sit beside it |
| Platform, and the format the campaign needs | Whether the person already talks about this subject unprompted |
| Audience size band, stated as a range | Whether comments read like an audience or like traffic |
| Posting activity in the last few months | Whether recent partnerships are so frequent that another one disappears |
| Categories to include | Conflicts, including anything you cannot sit beside for legal reasons |
| Categories to exclude | Whether the person would plausibly say yes to this brief |
Underneath, three lines that stop the argument later: who wrote the spec, the date, and the sentence describing who the campaign is meant to reach. When the shortlist is challenged, this document is the answer.
Two: a shortlist row
One row per candidate, and the same columns every time. Anything longer than a screen per candidate is research, not a shortlist.
- Handle, and the link.
- Why they are here, in one sentence tied to a condition from the spec.
- What they make, described by format rather than by category.
- Evidence they can carry the message: a specific post, linked, not a general impression.
- The audience question mark, if there is one, stated plainly.
- Contact route, and whether it is representation or direct.
- Anything blocking, including conflicts and recent competitor work.
The evidence column is what makes a shortlist defensible. A named post that somebody can open changes the conversation from taste to fact. Audience figures do not do that job, because third-party breakdowns are modeled rather than measured, a distinction set out in the guide to what discovery data can and cannot show.
Three: a rejection entry
The list you throw away is the one worth keeping. Four fields.
| Field | Example of the content it holds |
|---|---|
| Who | Handle and link, so the same person is recognized next time |
| Why not | One of: conflict, out of budget, wrong format, unavailable, declined, audience mismatch |
| Detail | The specific fact, such as an existing arrangement in an adjacent category |
| Date and reviewer | Because a refusal from two years ago is not a refusal today |
The "why not" field should use a fixed short vocabulary. Free text produces a file nobody can search, and the whole point is that next quarter's search starts by removing these names automatically. A rejection log is records management rather than administration: it exists so a decision can be found again. Where the reason is a doubt about the account itself rather than the fit, that is a different judgment with its own evidence standard, covered in the guide to verification tooling.
Four: the handover note
Discovery stops at a shortlist and somebody else picks it up. The handover is five lines and it saves a repeated week.
- The spec, attached, with its date.
- The shortlist, in groups by role in the campaign rather than in rank order.
- Who has already been contacted, when, and what was said. This is the line most often missing.
- Known constraints per candidate: representation, blackout dates, categories they will not take.
- What the searcher would do next, in one sentence, including who they would approach first and why.
Send it to the person who will contract and brief, and keep a copy with the campaign. When the same brief comes round next year, this note is worth more than the shortlist itself. Where it lands matters too: a note in an inbox is gone, and a note attached to the campaign record survives, which is the seam described in the guide to campaign workflow tools.
What ties the four together
Each document forces a reason to exist at the moment it is cheap to write. Reasons written later are reconstructions, and reconstructions are how a team ends up shortlisting the same wrong person twice.
Where an agency produces them on your behalf, ask for the rejection log rather than the shortlist, since the rejections show you the standard being applied, and what agencies genuinely do well is separated in the guide to agency models. Where the shortlist is being bought rather than built, and the counterparty handles matching for you, the trade-offs are different, and those sit in the guide to creator marketplaces.
Common questions
Is a spreadsheet enough for all of this?
For a single team, yes, for a long time. The moment it stops being enough is when two people need to see the state of the same shortlist without asking each other, and that arrives sooner than the software budget usually does.
How many candidates should a shortlist hold?
Enough to survive refusals, unavailability and rates you cannot meet. Take your own conversion history rather than a rule, and if you have no history, keep the first list generous and start measuring.
Should the client see the shortlist?
Show them the grouped version with reasons, never the raw export. A raw export invites a conversation about follower counts, which is the least useful comparison available and the hardest to walk back.
What happens to these documents after the campaign?
The rejection log and the exclusion list carry forward. The spec gets filed with the campaign as a record of what was asked for. The shortlist itself expires quickly and is worth less than either.
