A request like "Find me creators similar to NikolaTravels" sounds simple, but defining what counts as similar is harder than it appears. Two travel creators can publish very different content for very different audiences, so a shared niche tag is not enough. Instead of comparing tags, we compare what each creator publishes in their bio, captions, hashtags, and video descriptions, then layer in a quality score, follower size, country, and account type. The full pipeline is below.
Selecting a Seed Creator
Lookalikes always starts with a creator that already works for you. You enter a creator's username. If they're already in the database, that profile is used. If not, it is added before running the search.
From One Reference Profile to a Set of Lookalikes
AI Extracts the Reference Creator's Specific Niche
Before we search, we enrich the reference creator's profile. The bio, captions and username are analyzed to extract the niche-specific hashtags this creator is known for. It groups them into broad, medium, and specific tags. We add those into the search query so we find creators with the same niche depth, not just the same broad category.
Niche Tags Extracted from "NikolaTravels"
Searching only for #travel would return millions of accounts. The specific tags are what separate "another travel creator" from "a travel creator based in Lisbon who posts about budget travel in Europe", which is the level of match most brands need.
We Compare the Reference Creator Against Every Creator in Our Index
The enriched reference profile is matched against our creator index using a similarity search. Each candidate gets a relevance score from 0% to 100% based on how much their bio, captions, hashtags, and (for video platforms) descriptions and topics overlap with the reference creator's. Per-platform tuning makes sure we weight the right signals for each network.
One Reference Creator, Compared Against the Whole Index
Each grey dot is a creator in our index. The reference creator NikolaTravels is checked against every one of them, and only those whose content signals overlap with the reference become lookalikes.
Per-Platform Signal Mix
Captions, biography, and hashtags carry the most weight. We boost matches on biography and full name by a factor of 10, because creators in the same niche tend to describe themselves in similar ways.
Video descriptions, topics, and hashtags lead. YouTube also brings in its topic categories with a smaller weight. Follower threshold to qualify for ranking bonuses is higher (30K+) because audiences there scale differently.
We Combine Content Similarity With Creator Quality
Content match alone isn't enough. A creator can talk about travel and still be a low-quality account. So we blend the relevance score with our Creator Quality Index (CQI), a precomputed score that captures how strong an account is overall. The final ranking weights content similarity at 70% and quality at 30%.
How the Final Score Builds Up
CQI is our Creator Quality Index. It's a score we pre-compute for every creator on every platform, capturing engagement quality, audience health, account legitimacy, and market fit. Without it, content match alone could surface low-quality accounts that used the right keywords.
Read how CQI works →Filtering Low-Quality Candidates
Before ranking, the pipeline removes candidates that would not be useful: fan accounts, reposts, unofficial pages, accounts that mention the reference creator (typically fans or spam), and accounts too small or inactive to be actionable. Whatever survives the filters is ranked by similarity score.
Filters Applied to Every Search
Countries We Cover
United States · United Kingdom · Canada · Australia · New Zealand
Norway · Germany · France · Ireland · Italy · Austria · Spain · Netherlands · Sweden · Japan · South Korea · Switzerland · Singapore · Denmark · Finland · Belgium
End-to-End Example
Below is the full pipeline running on an illustrative reference creator. NikolaTravels is submitted, AI extracts the niche, the index returns matches, CQI and country bonuses adjust the ranking, and the filters remove low-quality candidates. The example shows the final results and how to read each score.
The top two creators share Nikola's specific tags and fall in a similar follower band, so they receive both a strong content match and the full CQI quality bonus. CityHopperAnna shares the city and the niche but covers them differently, which lowers the relevance score. BorisLifeLog is the contrast case: travel words appear occasionally in his vlogs, but his primary niche is lifestyle and his audience differs. The sharp drop from 87% to 62% identifies the cutoff between a true lookalike and a loose tag match.
How We Validate Lookalikes 
Lookalikes is harder to validate than a binary match because there is no single "correct" answer. The same two-track validation used elsewhere applies here, adapted for ranked similarity. We verify whether the creators returned behave like the reference creator, not just share a tag with them.
Lookalikes Matched on Real Content Signals
By reading what creators publish, enriching the reference creator's niche with AI, and combining content match with our Creator Quality Index, we return lookalikes that share the same audience signal, not just the same broad category.