Brief Discovery

What happens after a brand says what it wants, and how the shortlist it sees is chosen, scored and priced

What it is

A brand tells Zooly what it wants in chat: the product, who it pictures, when, and the budget. Once the chat knows enough about the talent, Zooly turns those answers into search rules and replies with a ranked shortlist.

Every name comes with a score, a few plain reasons, and a price line. The brand can take the whole list or pick from it. Nothing is sent to anyone at this point.

How the brief becomes search rules

The answers are read as the brand wrote them. Zooly looks for:

  • The niche — from the brand's own words, ignoring words that only name the kind of deal ("video", "audio ad", "AI"). If the request names no niche, what the brand is promoting stands in.
  • Audience size — "about 100K followers" is read as a band around that number, "50K+" as a floor, "under 50K" as a ceiling, and "50K to 150K" as a range. Podcasts and voice read downloads or listeners when the brief names them.
  • Platforms — TikTok, Instagram, YouTube, Twitch, X (when written as "Twitter" or "x.com"), LinkedIn, Snapchat, Facebook, Spotify, Pinterest. A voice audio ad ignores platforms.
  • Engagement — a floor such as "3% or more", for the deals where buyers filter on it.
  • Where the talent is based — "US", "U.S.", "USA" and "UK", "U.K." are read as the same places. For brand use, "US only" is read as where the ads run, not where the talent lives; only "based in" or "from" makes it the talent's home.
  • Language — spoken for voice, podcast and brand video; written for the rest.
  • Voice, accent, sport — only for the deals where those are part of what is bought.
  • Exclusions — "never worked with Sephora" or "no competitors" becomes a hard rule, not a score.

Almost everything is a preference: it moves the ranking but removes nobody. A few things are hard rules. For a post on the talent's own account, the platform the brand names is required. A named competitor, a live exclusive, or a red line on either side removes the pair, and so does a talent who blocked the brand.

Protected traits are never used. Race, ethnicity, religion, disability, sexual orientation and similar traits are stripped before any rule is built, and they are never shown in a reason.

The budget is read from the budget answer, or from a dollar amount written in the brief itself. "$5k" is read as $5,000. In the brief text only an amount marked with "$", "USD" or "dollars" counts, so "100,000 followers" is never taken for a budget. When the brand gives a total rather than a per-person number, Zooly treats it as the most any one talent can cost.

The six deal types

A brief becomes one of six kinds of deal. These are the shapes a brand can ask for:

What the brand buysThe deal
A podcast sponsorshipA host reads the ad inside one episode
A photo post on the talent's own accountThe talent posts a photo with the product
The talent's photo in the brand's adsA license to use one photo in the brand's channels
A video post on the talent's own accountThe talent films and posts a short video
The talent's video in the brand's adsA license to use one short video in the brand's channels
A voice audio adA recorded read in the talent's own voice

Five of the six go through the ranking on this page today. Podcast sponsorship is the exception, explained below.

What each result shows

Zooly shows up to five people, best first. For each one the reply gives:

  • A match score out of 100 — the two sides combined, held back by the weaker one. See How the scores are built.
  • Why it fits — up to three short reasons in plain words, such as "Uses the brand (they told us)" or "Asks $2,500, inside the $3,000 budget". Internal notes to double-check a fact are kept out of the message.
  • A price line — what the talent asks from, and whether that sits inside or above the brief's budget. With no budget in the brief it gives the asking price alone. With no price on the talent's side it gives none.
  • A lower-ranked pick, sometimes — about one slot in ten goes to someone further down who probably fits, so newer or thinly profiled talent are not shut out.

Talent who break a hard rule never appear, however well the rest fits.

The price line in full

The line depends on what is known:

  • Inside the budget — "Asks from $2,500, inside your $3,000 budget".
  • Above the budget — "Asks from $4,000, above your $3,000 budget".
  • No budget in the brief — "Asks from $2,500".
  • No price on the talent's side — no price line; Zooly describes who they are without one.

A thin brief still searches

Some briefs describe a feeling rather than a niche ("someone funny who gets outdoorsy people"). When the answers say little about the talent, Zooly searches on the brief's own words and their meaning, and adds nothing the brand did not say.

When the search finds nobody

If the ranking fails, or finds nobody, the older search runs in its place: the closest profiles by meaning, then one pass that picks the best of those. The brand still gets names and the same question about which to approach.

Only if that also finds nobody does Zooly say plainly that it found no match, and ask the brand to name the people it wants.

Podcast sponsorship stays on the previous path for now, because the matcher does not yet read downloads per episode — the filter that matters most for a podcast buy. The brand still gets names, each with a short line on who they are.

After the shortlist

The brand can say "all of them" or pick the ones it likes. Zooly then repeats the brief back and asks whether to send it. Nothing leaves Zooly until that yes.

A shortlist is not an approach. Nobody on it has been told anything, and no deal exists. The people named only hear about the brief once the brand agrees to send it.

What it does not do

  • It does not invent prices. A price line only appears when the talent has a minimum on file.
  • It does not search protected traits, even when a brief asks for them.
  • It does not send anything. Picking people only prepares the approach.
  • It does not use timing. The dates answer is kept for the deal; it does not change who is found today.
  • It does not adjust itself from what brands pick. The starting weights come from research and are still adjusted by hand.

What comes next

  • Podcast sponsorship moving onto the same path once downloads per episode are read.
  • A screen where a brand can browse its shortlist, instead of only reading it in chat.
  • Deal seeding — how Zooly proposes a deal to a matched pair
  • Outside judge — the two simulated readers that score a proposal
  • Automation — what Zooly may do without the team
  • Matching — the three scores and the rules behind them
  • Deal record — the story of each deal after both sides say yes