Find smaller creators by following four routes: people mentioning your brand, contributors in specialist communities, collaborators of relevant creators, and recommendations from customers or existing partners. Record where each lead came from, merge duplicates, and stop a route when it stops producing new candidates who meet your brief. Broad hashtag searches can remain one input, without becoming your entire search plan.
Set a working size band
Write down the platform and follower range before you start. Sprout Social uses 10,000 to 100,000 followers for micro-influencers. Treat that as an industry convention. Size bands vary across briefs and providers, so keep the actual count alongside the label.
For a hypothetical sewing-tool campaign, you might search accounts with 5,000 to 50,000 followers and accept candidates outside that range for review. That is a campaign choice, not a platform rule or a claim that every account in the range qualifies as a micro-influencer.
Add two content requirements. In this example, candidates must publish garment-repair tutorials and show the steps clearly enough for a beginner to follow. Record the required platform separately. A strong blog or community contributor may have no suitable video account.
If you cannot describe the required content, first turn the customer's problem into search terms. The routes below need a defined subject to follow.
Four routes and what to save
The limits below are suggested research controls. They are not tested benchmarks or platform requirements. Adjust them to the time you have, then keep them fixed during a search pass.
| Route | Starting point | Evidence to save | Suggested stop rule |
|---|---|---|---|
| Brand mentions | Public posts mentioning your product or account | Post URL, date, account URL and what the person demonstrated | Finish a declared date window, such as the previous 30 days |
| Specialist communities | Public tutorials, event programs, community discussions and contributor pages | Contribution URL plus the creator-owned profile it links to | Review two relevant communities before reassessing the topic |
| Relevant collaborations | Guest videos, co-created posts and credited contributors | Collaboration URL, creator role and topic connection | Follow one step outward from each starting creator |
| Customer and partner recommendations | Names volunteered by customers or creators you already work with | Recommendation date, reason given and verified public profile | Review one batch of ten recommendations, then measure new qualifying leads |
Brand mentions
Start with people who show your product in use or discuss the problem it solves. Open their profiles and inspect their own content before adding them. A customer complaint, giveaway entry or repost can mention your brand without showing any relevant creator work.
TikTok's mention documentation explains that mentions can appear in comments, Stories, posts and captions. Mentioned accounts receive notifications, and users control who may mention them. Your notifications therefore make one starting point; they cannot establish a complete inventory of discussion about your brand.
Keep the evidence literal. Write "demonstrated a repair using our thread" when that is visible. Do not turn it into "loyal customer" or "unpaid fan" without evidence.
Modash's discovery guide also recommends reviewing brand-related content and existing followers. Its broader set of methods is useful when one source of candidates runs out. The four-route log here keeps those leads in a single review process.
Specialist communities
Search where people teach the task your product supports. For the hypothetical sewing campaign, try public mending discussions, repair-workshop programs and tutorial contributors. Follow a contributor's own published link to their social profile where possible.
Look for repeated useful contributions. One popular comment gives you a lead to inspect; it does not establish a publishing history. Separate a knowledgeable participant from someone who regularly makes the kind of content your campaign needs.
Use public material and respect each community's participation rules. Do not join a private group solely to copy its membership into a prospect list. If you need recommendations there, ask a moderator whether a clearly identified brand request is welcome.
Relevant collaborations
Choose a starting creator whose subject and format fit the brief. Inspect credited guests, shared tutorials and co-created projects. A guest who explains a narrow technique may be worth reviewing even when the host's channel is too large for your campaign.
Record the person's role. Being credited as a camera operator does not establish that they publish tutorials themselves. Open their own account and check.
Stop after one connection in the first pass. Otherwise, a search for repair tutorials can drift through collaborators into unrelated lifestyle accounts. Keep a separate note for promising second-step leads so you can return if the first pass is thin.
A visible collaboration also says nothing conclusive about current availability or commercial restrictions. Use the sponsorship-history ledger when a past brand relationship needs closer review.
Customer and partner recommendations
Ask a question tied to a task: "Whose garment-repair videos would you send to a beginner, and which video would you start with?" Requesting one example gives you something to inspect.
Record the recommendation as a lead, even when it comes from a trusted customer. Verify the account and check the same content requirements you use elsewhere.
If your brand has an active YouTube channel, its own analytics can supply another starting point. YouTube's Audience report includes other channels your viewers watched over the previous 28 days. YouTube identifies collaboration discovery as a use for that report. Availability can be limited, and this requires access to your channel's Studio analytics.
Keep one candidate record and several discovery records
Use two spreadsheet tabs. This avoids counting the same creator as several new prospects while preserving evidence that multiple routes found them.
| Candidate tab field | What to enter |
|---|---|
| Candidate ID | A stable internal ID such as C001 |
| Platform and canonical profile URL | The actual profile, without tracking parameters |
| Handle and checked date | Current handle plus the date you opened it |
| Displayed follower or subscriber count | The displayed value and date; keep rounded values rounded |
| Topic evidence | Links to relevant original posts |
| Review status | New, review, reject or hold, with a short reason |
| Linked accounts | Other profiles confirmed through creator-owned links |
The discovery tab needs candidate ID, route, starting page or person, evidence URL, discovery date and one sentence explaining the connection. Add a duplicate flag and the existing candidate ID when a route finds someone already recorded.
Deduplicate by platform and profile URL first. Similar names or matching profile photos alone are weak reasons to merge accounts. When identity remains uncertain, leave both records on hold rather than silently joining them.
Public content supports a subject-fit judgment. It does not establish the audience's age or location. YouTube, for example, puts demographic reports inside Studio and defines top geographies by watch time. Keep unavailable audience fields marked unknown; evaluate missing audience data before making a final selection.
Decide whether to continue a route
Imagine one hypothetical search pass produces these results:
| Route | Profiles reviewed | New candidates meeting content requirements |
|---|---|---|
| Brand mentions | 12 | 3 |
| Specialist communities | 10 | 4 |
| Collaborations | 10 | 2 |
| Recommendations | 8 | 0 |
These are illustrative counts, not expected results. "New" excludes duplicates found through earlier routes. Record the route order because later routes may look less productive after earlier routes found the same people.
If the recommendation batch contains good candidates already in the log, record that overlap. If it contains unrelated accounts, rewrite the question before asking another batch. If specialist communities produce several relevant creators, review another community with the same content requirements.
Start your next search by writing the size band, two content requirements and each route's limit at the top of the log. Then work one pass across all four routes before expanding whichever route produced new, relevant candidates.
