How to Find Creators Who Actually Move Product

The goal: build a shortlist of creators who have already shown they can sell

You are not hunting for famous faces. You are hunting for working proof. The creators you want already reach for products like yours without a brief, they post to an audience that can actually buy, and their following holds up when you check it closely.

Most teams do the opposite. They sort by follower count, they reuse an agency roster that already promotes two competitors, or they chase a viral post from six months ago. All three rank creators on appearance, not on behavior.

This playbook reverses that order. You will define a tight set of three to five direct competitors, pull every creator who mentioned that set across Instagram, TikTok, and YouTube in the past 180 days, keep only organic mentions, rank the survivors by Creator Value Score, calibrate each finalist against true peers, then check audience overlap so you pay for new buyers rather than your existing followers.

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

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

What you need before you start

• A Sentia monitor with crawl history. You need at least 90 days on Growth or at least 60 days on Scale, where crawl depth and backfill are higher. Go to Settings · Data coverage and confirm you see continuous daily bars, not gaps.
• Creator engine access. Growth and Scale include full access to affinity, calibration, and overlap. Starter shows the Creators tab as a 30 day read only preview. That preview is useful for learning the layout, but it is not enough to run this workflow.
• Three to five direct competitors you lose sales to today. Match on price and use case. If you sell a 34 dollar serum, your cohort is other 25 to 50 dollar serums with similar routines, not a 185 dollar luxury house and not a 9 dollar mass cleanser.
• A test budget range for context. Example: 6,000 to 15,000 dollars for 6 to 8 paid posts. You will need it in Step 4 to translate cost bands into a real test plan.
• One owner for outreach. Exported lists die in shared drives when three people think someone else sent the email.

If you lack history, pause here. Run social listening for two to four weeks to build a baseline, then return. Affinity without history is guesswork.

Step 1 · Define the cohort, not the creator

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

The cohort is the exam. Every later step searches only inside this universe. Precision here controls quality everywhere else.

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

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 healthy. The team screenshot for the brief was that card plus the five logos and the price ladder.

Expected output: a saved cohort plus a cohort card with total crawlable mentions per month. Targets for 2026: 40,000 to 120,000 per month for mid cap beauty and personal care, 15,000 to 60,000 for specialty food, beverage, supplements, and pet. If you see fewer than 5,000, add one credible competitor at the same price tier. If you see more than 250,000, remove the largest outlier and rebuild.

Screenshot as words: Settings · Cohort shows five rows with verified handles, a green Active pill on each row, and a timestamp that reads Footprint rebuilt today at 03:14 UTC.

Step 2 · Pull the cross platform affinity list

Go to Monitor · Creators · All mentions. Set the date control to Last 180 days and the brand control to Cohort: all brands. Confirm the table shows columns for Creator, Platform, Followers, Mentions, Affinity Score, and Last mention.

Click Sort · Affinity Score · High to low. Do not use mention count. Mention count rewards one viral spike. Affinity Score rewards repeated, recent, unscripted use.

How Affinity Score works in 2026:

1. Repeat use beats single trials. Four routine uses in 60 days outranks seven scattered mentions over two years.
2. Unscripted placement beats dedicated reviews. A serum visible on a shelf in three Get Ready With Me videos signals habit. A single 10 minute review often signals a one time trial.
3. Recency decays. A mention from the past 7 days carries full weight. Weight halves near day 90, and a mention from day 120 carries roughly one third weight. A creator who stopped posting about the category three months ago slides down automatically.

For mid cap beauty, expect 200 to 800 creators in this raw pool. For refillable fragrance, clinical haircare, electrolyte mixes, and high protein snacks, 60 to 160 is normal. Depth beats breadth. Sixty creators with habit beat 800 tourists.

You can click Export · CSV for offline scanning, but do the filtering inside Sentia where provenance tags and drawer evidence live.

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

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 and branded content disclosures, Instagram paid partnership labels, and caption disclosures such as gifted plus discount code plus tracking link. Any post 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 a roster everyone else already bought.

What survives is more valuable: the organic shoutout from a creator who reached for your category without a brief. Example rows that survived for that 34 dollar serum brand:

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

If a creator has both paid and organic mentions for the cohort, Sentia keeps the organic rows and hides the paid rows. Keep that creator in the pool. The hidden paid history is still useful at negotiation because it confirms they accept 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: 642 creators before the filter, 418 after. If your drop exceeds 65 percent, your category is heavily sponsored. Extend lookback to 365 days in Creators · Settings · Lookback to recover older organic history. Do not loosen the filter. Paid mentions do not predict future conversion.

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 after removing suspected inauthentic followers, engagement rate adjusted for follower size, brand affinity from Step 2, and partnership cost efficiency relative to 2026 category medians. It is geography aware. A creator concentrated where you can sell and ship scores higher for you than an identical creator concentrated where you cannot fulfill.

Open any row to see the drawer summary:

Two rules before you commit:

1. Check geography first. A score of 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. For United States only shipping, require United States 50 percent or higher.
2. Treat scores as fresh produce. CVS recalculates weekly. Re export within 48 hours of outreach so your offer matches what managers see. A score from last month can move 5 to 10 points after a viral spike or a quiet stretch.

Worked math: with a 10,000 dollar test budget and a 1,400 dollar average band, you can fund seven posts. Select seven across two follower tiers to learn what size converts, for example three at 50k to 150k and four at 150k to 400k. Reserve 1,500 dollars for boosting the two winners with paid amplification.

Price context for late 2026: United States micro routine creators at 50k to 150k followers typically quote 800 to 2,000 dollars per integrated TikTok or Reel, mid tier creators at 150k to 400k typically quote 1,800 to 4,500 dollars, and YouTube integrations run 1.5 to 2.5 times the short form rate for the same follower band. Use the Sentia band as your anchor, not a screenshot from a 2023 rate card.

Expected output: a ranked shortlist of 30 to 60 creators with CVS, cost band, primary geography, and last organic mention on one row. Export this version as cohort_shortlist_cvs_2026-10-12.csv.

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

Select your shortlist, then click Calibrate · Run calibration. Sentia compares each creator to peers matched on platform, follower band, posting category, and language, 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 3.6 percent near 50k followers, 2.0 percent near 500k, and 1.1 percent near 2M. A flat 3 percent rule would bless every small account and punish every large one. Calibration grades on a curve.

Calibration also catches what raw metrics hide:

• Green means consistent. Engagement, follower growth, and posting cadence track the peer median. Proceed to outreach.
• Amber means caution. Example reason: Engagement 38 percent above peer median with comment duplication across three posts. Proceed with a single video test, require manual comment review, and hold back paid usage rights until results clear.
• Red means cut. Example reasons: Follower spike plus low retention consistent with purchased followers, or reciprocal comment clusters consistent with engagement pods. Cut even if CVS is above 80. No score survives fake distribution.

Expect to lose 20 to 40 percent here. For the serum example, 52 ranked creators became 33 after calibration: 22 green, 11 amber, 19 red removed. That cut is the quiet work of the playbook. It is also where reach sorted spreadsheets and general listening tools fall short, because they surface who is large and loud without grading authenticity against peers. For a direct contrast on data depth, see Sentia versus Brandwatch.

Expected output: every survivor 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 Oct 2026.

Step 6 · 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, where connected.

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. Accept it only for conversion or retargeting where warm reach is the goal.
• Below 10 percent with high affinity needs a second look. It can signal a one off mention or an adjacent audience. A gaming creator who mentioned serum once is less valuable than a routine vlogger whose audience sits two clicks from checkout, even if both show low overlap.

The drawer states the Venn in words: Shared audience 22 percent, Creator only 68 percent, Brand only 10 percent. Paste 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, 33 calibrated creators became 19 in band, then 11 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 keeps provenance, badge reason, and overlap together 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 shelf use.
• Proof: two links to organic posts with date and format noted, for example TikTok routine posted Oct 3 and Instagram Story repurchase posted Sept 18.
• Value: CVS 76, cost band 1,200 to 1,800 dollars, United States 68 percent, ships to your core market.
• Risk check: Green calibrated, overlap 22 percent, creator only reach 68 percent.
• Opening offer: one integrated short form video plus 30 day paid usage, with a second video option if click to product page clears baseline. State usage, exclusivity for 30 days in category, and posting window in the same paragraph to avoid renegotiation.

That format turns a subjective pick into a defensible buy. Managers also respond faster because you priced against recent organic affinity, not against follower count.

File naming that survives handoffs: CreatorPacket_Brand_CreatorName_2026-10-12.pdf. Include the export date on page one. CVS and Affinity Score decay with time, and dating the packet prevents debates about stale numbers.

Troubleshooting

The Creators tab is empty or read only. Check Settings · Data coverage for 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. Do not run outreach on preview data.

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 volume does 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 confirm TikTok is included in the platform mix. TikTok often retains 20 to 30 percent more organic routine placement than Instagram in beauty, food, and pet. YouTube Shorts can add another 10 to 15 percent of routine use that long form search misses.

Every finalist shows overlap above 45 percent. You dominate your niche graph. 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 can 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. There is no discount deep enough to fix fake reach.

The partnerships team wants to add agency names to the list. Allow it as a labeled holdout group. 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 and click to product page rate, 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, especially for TikTok Shop and YouTube shopping integrations where the Federal Trade Commission expects clear disclosure.

Related

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