Creator marketing runs on lists, and most teams build them by hand: scrolling feeds, opening profiles one by one, copying whatever contact detail happens to be visible. That process caps out at a few dozen prospects a week and decays immediately — creators rename, pivot niches, and change contact routes. A pipeline built on public profile data produces the same list continuously, with fields consistent enough to score and segment.
What belongs in an outreach row?
Define the row before collecting anything. A usable creator record needs more than a name and a follower count:
- platform and canonical profile URL;
- display name and handle;
- audience size and recent activity signal;
- niche indicators — bio keywords, hashtags, category;
- the business contact the creator chose to publish;
- collection date, because every one of these fields drifts.
The last field is the one hand-built lists always miss. Creator data ages fast; a list without timestamps cannot tell you which rows to refresh.
Where does discovery happen?
Start from content, not from directories. Paid influencer databases sell coverage, but their niche tagging is broad and their refresh cadence is opaque. Discovery from the platforms themselves keeps the niche definition in your hands:
- On Instagram, trending and hashtag surfaces expose which creators currently occupy a niche — Instagram Trending Scraper collects those surfaces, and the workflow in Instagram and TikTok Trend Research shows the research loop in detail.
- On TikTok, creator-level data is available through TikTok Creator API.
- For creator profiles on Instagram specifically, Instagram Creator Scraper returns the profile-level fields discovery surfaces do not.
Discovery produces candidates. The next two stages turn candidates into an outreach list.
How do you get from a profile to a business contact?
Collect only what the creator published for that purpose. Creators who want brand deals say so: a business email in the channel’s about section, a contact line in the bio, a link-in-bio page. That is the contact route to collect — it is public by intent, not by accident.
YouTube Creator Email Scraper works this seam on YouTube: given channels, it extracts the business contact details creators list publicly, alongside the channel fields needed for scoring. Because the source is the creator’s own published information, the resulting list is both cleaner and more defensible than scraped-and-guessed addresses — these are people who asked to be contacted about their work.
Two practices keep the list responsible:
- No pattern-guessing. If a creator publishes no contact route, the row stays contactless and outreach goes through the platform’s own messaging.
- Suppression on request. One opt-out removes the creator from every future refresh, which is only enforceable if rows carry stable profile IDs.
How should the list be scored?
Rank by fit and reachability, not audience size. Follower count is the least predictive field in the row. A workable score combines:
- Niche fit — bio and content keywords against the campaign’s terms;
- Activity — recent posting cadence, because a dormant channel converts nothing;
- Reachability — rows with a published business contact rank above rows without one;
- Audience band — segment micro from mid-tier rather than sorting descending; response rates run inverse to size.
Score in the storage layer, not in the collection layer. Rescoring a stored dataset is free; recollecting is not.
What does the refresh loop look like?
Treat the list as a dataset with a decay rate. A monthly scheduled run over the same discovery inputs keeps three things current: new entrants in the niche, contact changes on existing rows, and dormancy flags for creators who stopped posting. Diff each run against the last — the deltas are the work queue, the same schedule-and-diff pattern used across our operations guides, from price monitoring to keyword trends.
Validate the output the way any production dataset is validated — How to Validate Scraped Data Quality covers the checks that catch schema drift before it reaches an outreach tool.
Continue in the directory
Turn the guide into a real sample run.
Open the current AgentX contract, check pricing and fields, then validate a narrow output.