Lifecycle Part 1: TikTok Creator Discovery (Finding Candidates at Scale)

This is the opening chapter of a five-part series walking the entire KOL lifecycle — from the first keyword you type, to the moment Claude runs the whole workflow without you. We start here because every downstream stage inherits the quality of this one. A campaign with the wrong discovery funnel cannot be saved by better vetting, a slicker CRM, or a smarter monitoring rule. Get the funnel right and everything after it compounds.

Why discovery is the load-bearing stage

Most influencer-marketing teams treat discovery as a chore — a morning of scrolling TikTok, screenshotting promising handles, and pasting them into a spreadsheet. It feels like the cheapest step because no money changes hands yet. In practice it is the most expensive step done badly, because every wrong creator that enters the funnel costs you time in Part 2 (vetting), space in Part 3 (the library), attention in Part 4 (monitoring), and credibility in every outreach reply rate report you ever ship to a stakeholder.

The math is brutal once you write it out. A typical mid-market brand runs one TikTok launch per month with ten paid creators. If the discovery funnel has a 1-in-50 conversion rate from candidate-to-signed (industry typical), you need 500 candidates per launch. Doing that by hand at five minutes per creator — open the profile, eyeball the engagement, click the bio link, look for an email, paste a row — is roughly 40 hours per launch. That is one full-time week, every month, before any creator is contacted.

The promise of this series is that the 40 hours collapses into roughly 10 minutes for the discovery stage, with a higher-quality funnel at the end. Not because the tooling magically makes the funnel perfect, but because the part of the work that should never have been manual stops being manual.

Keyword strategy — three categories that actually work

The single highest-leverage decision in discovery is the keyword. Pick wrong and the 200 creators in your result set are 200 of the wrong creators. Pick right and the funnel is doing your vetting for you before you even read the first row.

Three keyword categories consistently outperform everything else. They are not exotic — most teams just default to category words and never try them.

• Product nouns. phonecase, standing desk, dog leash, baby stroller. High-intent buyers post under these terms because they are reviewing or unboxing the actual product. Conversion rates from outreach to signed deal are typically 2-3x higher than category words.
• Routine / how-to phrases. skincare routine, morning routine, home gym setup. Creators who teach are creators who sell — the content format already trains the audience to act on recommendations.
• Niche community terms. vegan baking, web3 wallet, sober curious. Smaller candidate pools, but engagement rates are routinely 8-15% and contact rates approach 1-in-3 because creators inside niche communities expect to be approached.

What does not work: single-word category terms like beauty, fitness, or food. These return the most-followed accounts under the most generic umbrella, which means megacreators who already work with three competitors of yours and a long tail of creators tagging the category to chase impressions. The signal is bad and the price is worse.

The mechanics — wide entry, narrow exit

With the keyword picked, the second design decision is the shape of the funnel. The right shape is a wide entry and a narrow exit: let the result pool fill broadly, then tighten with filters downstream. The wrong shape — narrow entry and narrow exit — starves the funnel and produces 8 survivors that all look the same.

On /search the wide-entry pattern looks like this:

1. 1

   Type one product-noun or routine-phrase keyword. Leave follower range generous.

   A 10k-1M follower band is wide enough to catch the mid-tier sweet spot without sweeping in either dead micro accounts or mega-accounts you cannot afford. Set the result count to 200 (default) for a launch test, 500 for a campaign.
2. 2

   Optionally constrain country at the search step.

   Country filtering is the one search-stage filter that pays for itself, because a US brand that ships only to the US should not spend credits pulling Indonesian creators back. See Find TikTok creators by country for the country-filter recipe.
3. 3

   Let the live bio crawl run. Results return ranked by engagement rate.

   Email extraction happens in parallel — KOLens clicks through each creator's bio link (Linktree, Beacons, personal site) and pulls verified business emails. Most rows have a final email inside 60-90 seconds.
4. 4

   Apply tight exit filters at the table level.

   Engagement rate floor (4% for mid-tier, 6% for niche), posting cadence (last post inside 14 days), and account-role filter (drop reseller and brand accounts). The 200-creator pool now becomes the 30-60 candidates worth vetting in Part 2.

Niche segmentation — running three searches, not one

A common mistake at the discovery stage is to run one search per launch and treat the result as the entire candidate pool. The better pattern is to fan a launch across three to five sibling keywords inside the same niche and dedupe at the List stage.

For a skincare launch, that might look like: skincare routine, sensitive skin, face serum, korean skincare. Each pulls a different slice of the niche — the routine creators, the condition-specific creators, the product-category creators, and the regional/style creators. Combined, they produce a 600-800 creator pool with very different audience tones, and the right creators for your specific positioning fall out naturally during vetting.

The companion playbook here is Finding mid-tier TikTok creators — the same wide-entry / narrow-exit pattern with explicit follower band recipes.

Where KOLens enters the workflow

The KOLens surfaces relevant at the discovery stage are deliberately few — the product is built so the first ten minutes of a launch only require one screen.

• /search — the keyword discovery surface itself. One input, 200 ranked creators with engagement, audience country, posting cadence, and verified emails per row.
• /find-tiktok-influencers — niche landing pages with pre-seeded keyword recipes. Useful when you want a starting prompt for a vertical you have not worked in before (baby strollers, home fitness, K-beauty, etc).
• /tiktok-hashtag-analyzer — for when the keyword does not return enough volume, the hashtag view shows which adjacent tags the top creators are actually using. Often gives you the next two keywords to try.
• /lists — the saving surface. Discovery does not end with a CSV download; it ends with a named cohort in a KOL List that the rest of the lifecycle can refer to.

A worked example — phonecase launch, 10 minutes

Here is the discovery half of a real launch from a 4-person phonecase brand we have written about elsewhere (see the full case story). Numbers are exact; names are stylised.

The brand picked four keywords: phonecase, iphone case, aesthetic phone case, magsafe accessories. Each got a 200-creator search with the country filter set to US and the follower band 20k-300k. The four searches ran in parallel and completed in 4 minutes, returning 783 deduplicated creators with verified emails on 612 of them.

Three exit filters then ran at the table level: engagement rate above 5%, last post inside 14 days, and the account-role filter set to exclude reseller and brand accounts. The 783 collapsed to 134 candidates. The whole stage — from typing the first keyword to a named KOL List of 134 vetting-ready candidates — took 9 minutes and cost roughly 80 credits, or about four dollars.

The previous workflow, before the brand adopted KOLens, was a full Friday afternoon of TikTok scrolling and a 60-row Google Sheet. Half the emails in that sheet were wrong. The new sheet is bigger, fresher, and ready to hand off to vetting before lunch.

vs the alternatives

In the next chapter

Discovery hands you a 200-creator pool. Part 2 — Vetting is how you separate the 30 signal creators from the 170 noise creators inside that pool: engagement quality, audience geography, posting cadence, authenticity flags, and the 8-signal authenticity score that catches the inflated-800k-followers-at-2%-engagement pattern. Without that vetting layer, half the candidates you hand to outreach will burn replies and the other half will burn product samples.

Continue to Part 2 — Vetting TikTok Creators →

The full lifecycle series

1. Part 1 — Discovery (you are here): TikTok Creator Discovery — finding candidates at scale
2. Part 2 — Vetting: Separating signal from noise inside the funnel
3. Part 3 — Library: Turning finds into a durable asset
4. Part 4 — Monitoring: Tracking the picks you made
5. Part 5 — AI / MCP: Claude integration for natural-language research