How to Find Creators Who Actually Move Product

The goal: a shortlist that has already proven it can sell

You are not looking for popular creators. You are looking for creators who already reach for products like yours when no one is paying them, whose audience is the right size and shape to convert, and whose following is real.

That is a different search. Most teams fail because they sort a spreadsheet by follower count, or they accept an agency roster of ten familiar faces who are already promoting three competitors. Both methods rank creators on what they look like, not on what they have already done.

This playbook flips the order. You will build a tight cohort of three to five direct competitors, pull every creator who mentioned that cohort across Instagram, TikTok, and YouTube in the past 180 days, remove paid partnerships so you keep only organic affinity, rank survivors by Creator Value Score, calibrate each finalist against true peers, then check audience overlap so you pay for new reach, not for people who already follow you.

Plan on 75 minutes end to end in Sentia: 15 minutes to set the cohort, 20 minutes to filter and rank, 25 minutes to calibrate, 15 minutes to check overlap and export.

What you walk away with: 8 to 15 names with documented organic mentions, a Creator Value Score with cost band and geography, a green or amber calibration badge, and overlap in the 10 to 35 percent sweet spot. That packet is ready for outreach.

What you need before you start

• A Sentia monitor with at least 90 days of crawl history on Growth, or at least 60 days on Scale where crawl depth is higher. Affinity scoring needs a baseline. Check Settings · Data coverage to confirm.
• Creator engine access. Growth and Scale include full access. Starter shows the Creators tab as a 30 day read only preview, useful for learning the layout but not for running the full workflow.
• Three to five direct competitors you would lose a sale to tomorrow. Match on price tier and use case. If you sell a 34 dollar serum, your cohort is other 25 to 50 dollar serums, not a 180 dollar luxury house.
• A rough budget band for your first test. Example: 5,000 to 15,000 dollars for 8 paid posts. You will need it in Step 4 to interpret cost efficiency.
• One owner for outreach, so the exported list does not sit in a shared drive.

If you are missing crawl history, do not skip ahead. Run social listening for two to four weeks first, then return. Affinity without history is just a vibe check.

Step 1 · Define the cohort, not the creator

Open your brand monitor, then go to Settings · Cohort · Edit cohort. Add three to five direct competitors by handle or domain. Save. Sentia rebuilds the cohort footprint overnight, merging share of voice, mention velocity, and audience graphs across the combined set.

Think of the cohort as the exam. Every later step only searches inside this universe, so precision here controls quality everywhere else.

Too wide and you inherit noise. Add a mass market giant or a luxury house and you will surface mega creators who have never influenced a 34 dollar purchase. Too narrow and you starve the pool. Two niche clinical brands with low posting volume will return fewer than 50 creators, which is not enough to rank.

Worked example: a mid price serum brand at 34 dollars added four competitors priced at 28, 29, 38, and 42 dollars. The next morning the cohort card read:

That is a healthy pool. The team screenshot for the brief was that single card plus the five brand logos.

Expected output: a saved cohort and a cohort card showing total crawlable mentions per month. For mid cap beauty, target 40,000 to 120,000. For specialty food, beverage, or pet, target 15,000 to 60,000. If you see fewer than 5,000 mentions, widen by one credible competitor. If you see more than 250,000, tighten by removing the largest outlier.

Step 2 · Pull the cross platform affinity list

Go to Monitor · Creators · All mentions. Confirm the date control reads Last 180 days and the brand control reads Cohort: all 5 brands. You should see a table with columns for Creator, Platform, Followers, Mentions, Affinity Score, and Last mention.

Click Sort · Affinity Score · High to low. Do not use the default sort by mention count. Mention count rewards one off viral spikes. Affinity Score rewards repeated, recent, unscripted use.

How Affinity Score works in practice:

1. Repeat mentions count more than single mentions. Four routine uses in 60 days beats seven scattered mentions over two years.
2. Unscripted placement counts more than dedicated reviews. A serum visible on a shelf in three Get Ready With Me videos signals habit. A single 10 minute review signals a trial.
3. Recency decays. A mention from last week carries full weight. A mention from 120 days ago carries roughly one third weight, with weight halving near day 90.

For a mid cap beauty brand, this pool is typically 200 to 800 creators. For refillable fragrance, clinical haircare, or electrolyte mixes, 50 to 150 is normal. Depth beats breadth. Fifty creators with real habit is better than 800 tourists.

Click Export · CSV from the table header if your partnerships lead wants to scan offline, but do the filtering inside Sentia where provenance tags live.

Expected output: a prospect pool table sorted by Affinity Score, with platform mix visible. Sanity check: at least 60 percent of rows should show a last mention date within the past 60 days. If most last mentions are older than 120 days, your cohort may be stale. Swap one declining competitor for a rising one.

Step 3 · Strip out paid partnerships to keep only organic affinity

Stay in the Creators tab. Open Filter · Provenance and select Organic only. Click Apply.

Sentia tags paid provenance from five signals: Meta Ad Library matches, TikTok Commercial Content API flags, YouTube paid promotion flags, YouTube branded content disclosures, and Instagram paid partnership labels plus clear caption disclosures such as gifted with discount code plus tracking link. Anything matching at least one paid signal is removed from this view.

Expect the pool to shrink by 30 to 50 percent. That drop is the point. In most categories the highest mention counts belong to creators already on retainer somewhere. You do not want to rediscover the roster everyone else already bought.

What you keep is the organic shoutout · the creator who reached for your brand or a direct competitor without a brief. Example rows that survived for that serum brand:

• Creator A, TikTok, 84k followers, 5 organic mentions in 71 days, last mention 9 days ago, routine shelf placement in 4 videos
• Creator B, Instagram, 212k followers, 3 organic mentions in 44 days, last mention 12 days ago, repurchase story with empty bottle
• Creator C, YouTube, 46k subscribers, 4 organic mentions in 90 days, routine plus one comparison video versus two cohort brands

If a creator has both paid and organic mentions for the cohort, keep them. The filter retains the organic rows and hides the paid rows. The paid history is still useful later at negotiation because it confirms they take briefs and reveals their prior category rate.

Expected output: a filtered table where the Provenance column reads Organic on every row. For the example brand: 640 creators before the filter, 415 after. If your drop is larger than 65 percent, your category is heavily sponsored. Extend lookback to 365 days in Creators · Settings · Lookback to recover older organic history rather than loosening the filter.

Step 4 · Rank by Creator Value Score

With Organic only still applied, click Sort · Creator Value Score · High to low. Take the top quartile as your working shortlist, usually 30 to 60 names.

Creator Value Score, or CVS, is a 0 to 100 score that fuses four inputs: effective reach, engagement rate adjusted for follower size, brand affinity from Step 2, and partnership cost efficiency relative to category medians. It is geography aware. A creator whose audience concentrates where you can sell and ship scores higher for you than an identical creator concentrated where you cannot fulfill.

How to read the row. Open the creator drawer by clicking any name. You will see:

Two rules before you fall in love with a score:

1. Check geography first. A 78 concentrated in a market you cannot ship to is worth less than a 71 concentrated where you sell. Open Drawer · Audience · Top markets and confirm at least 50 percent in shippable markets.
2. Treat scores as fresh produce. CVS recalculates weekly. Re export within 48 hours of outreach so your offer matches what managers see on their side. A score from last month can be off by 5 to 10 points after a viral spike or a quiet month.

Worked math: with a 10,000 dollar test budget and an average band of 1,500 dollars per post, your top quartile of 51 creators gives you six to seven paid tests. Pick six across two follower tiers, for example three at 50k to 150k and three at 150k to 400k, to learn what size converts before you scale.

Expected output: a ranked shortlist of 30 to 60 creators with CVS, estimated cost band, primary geography, and last organic mention date on one row. Export this version as cohort_shortlist_cvs_date.csv.

Step 5 · Calibrate against true peers, not against the world

Select your shortlist, then open Calibrate · Run calibration. Sentia compares each creator to peers matched on platform, follower band, and category, then assigns a green, amber, or red badge with a one line reason.

Why this matters: engagement decays with scale. In Sentia 2026 beauty benchmarks, median engagement runs about 3.8 percent near 50k followers, 2.1 percent near 500k, and 1.2 percent near 2M. A flat 3 percent threshold would bless every small account and punish every large one. Calibration fixes that by grading on a curve.

Calibration also catches what raw numbers hide:

• Green means consistent. Engagement, growth, and posting cadence track the peer median. Proceed.
• Amber means caution. Example reason: Engagement 38 percent above peer median with comment duplication. Proceed with a smaller single video test, require original comments review, and hold back usage rights.
• Red means cut. Example reasons: Follower spike plus low retention consistent with purchased followers, or Engagement pod pattern with reciprocal comment clusters. Cut even if CVS is 80 plus. No score survives fake distribution.

Expect to lose 20 to 40 percent here. For the serum example, 51 ranked creators became 32 after calibration: 22 green, 10 amber, 19 red removed. That cut is the step doing the quiet work. It is also where reach sorted spreadsheets and listening tools such as Brandwatch or CreatorIQ most often fail, because they surface who is large and loud without grading authenticity against peers.

Expected output: every shortlisted creator carries a badge. Keep greens for core outreach, keep select ambers for low risk tests at 30 to 50 percent of your standard fee, remove all reds. Save the view as Calibrated shortlist.

Step 6 · Cross check audience overlap so you buy new reach

Select the calibrated survivors, then open Drawer · Audience overlap · Compare to my brand audience. Sentia compares each creator follower base against your known organic audience from owned social engagement and site behavior.

Use this band:

• 10 to 35 percent overlap is the sweet spot. Affinity is real and most impressions land on new potential buyers.
• Above 45 percent is redundant for awareness. You are paying to reach people who already follow you. Only accept this for conversion or retargeting campaigns where preaching to warm followers is the goal.
• Below 10 percent with high affinity needs a second look. It can mean a fluke mention or an adjacent audience. A gaming creator who mentioned your serum once is less valuable than a skincare routine creator whose audience sits two clicks from checkout, even if both show low overlap.

The drawer shows a simple Venn in words: Shared audience 22 percent, Creator only 68 percent, Brand only 10 percent. Screenshot that block into your outreach sheet. It becomes your defense when finance asks why Creator B at 212k followers beat Creator D at 900k followers. Answer: Creator B adds 68 percent new reach with proven taste. Creator D recycles your own followers at four times the fee.

For the serum brand, 32 calibrated creators became 18 in band, then 10 finalists after geography and budget fit. Six signed within three weeks. The first four live posts beat the brand median sponsored post engagement by 41 percent and drove a 2.3 times higher click to product page rate than the prior agency sourced cohort, because every partner had already demonstrated the exact behavior the brand paid to amplify.

Expected output: a final list of 8 to 15 creators with overlap percent, CVS, calibration badge, cost band, and two links to best organic mentions. Export from Creators · Export · Outreach packet CSV. That file includes provenance, badge reason, and overlap so context travels with the names.

What to send the partnerships team

Do not send a list of handles. Send a one page packet per creator:

• Why them in one sentence: 5 organic mentions in 71 days, last 9 days ago, routine use.
• Proof: two links to organic posts with date and format noted.
• Value: CVS 76, cost band 1,200 to 1,800 dollars, United States 68 percent.
• Risk check: Green calibrated, overlap 22 percent, creator only reach 68 percent.
• Opening offer: one integrated TikTok plus 30 day usage, with a second video option if click rate clears baseline.

That format turns a subjective pick into a defensible buy. It also speeds negotiation because managers see you priced against recent organic affinity, not against follower count.

Troubleshooting

The Creators tab is empty or read only. Check Settings · Data coverage for crawl history. You need 90 days on Growth or 60 days on Scale. On Starter you only get a 30 day preview. Confirm plan access with your admin, then return after history builds.

Prospect pool is under 50 creators after Step 2. Your cohort is too narrow or your category is too new. Add one adjacent but credible competitor at the same price tier, or extend lookback from 180 to 365 days in Creators · Settings. Do not drop the Organic only filter to inflate numbers. Paid mentions do not predict conversion.

Everything disappears after the Organic only filter. Your category is pay to play. Keep the filter on, extend to 365 days, and add TikTok to the platform mix if you started Instagram only. TikTok often retains 20 to 30 percent more organic routine placement in beauty, food, and pet.

Every finalist shows overlap above 45 percent. You dominate your niche. Either accept high overlap for a conversion push aimed at warm audiences, or add a competitor with a younger or adjacent audience to the cohort and rerun Steps 2 through 6 to pull in new graphs.

Scores shifted between exports. Normal. CVS recalculates weekly and Affinity Score decays with recency. A creator who went quiet for three weeks will slide 4 to 8 points. Re export within 48 hours of outreach and reference the export date in your offer sheet.

A high CVS creator is flagged red on calibration. Trust the badge. Purchased followers and pod engagement are exactly what follower count ranking misses. Cut the creator and move to the next green in band. Do not negotiate a discount on fake reach. There is no discount deep enough.

The partnerships team wants to add agency names to the list. Allow it as a holdout group, but keep them labeled separately. Run two to three agency names against six to eight affinity names with identical briefs and budgets. In most reruns the affinity group wins on cost per engaged view, which settles the debate with data.

Legal or compliance asks for disclosure history. Use the drawer provenance log. It lists paid flags by source with dates, which helps you confirm a creator discloses properly before you sign. Prefer creators with clean prior disclosure over those with missing labels.

Related

• Glossary: Creator affinity, CVS · Creator Value Score, EMV, Calibration
• Comparison: Sentia versus Brandwatch