Methodology · Established 2019
The Performance-Score model, documented.
A proprietary scoring system — not AI magic — that vets 38,500 mid-tier creators across four dimensions and ships campaigns in 41 hours. Here is how the model works, what it rejects, and what it produces for clients.
Chapter 01 · Origin
Why we built the model in-house.
In 2019, off-the-shelf creator discovery tools — the platforms that sell reach by the thousand — couldn't tell us whether a creator's audience was real, whether the engagement was bought, or whether a given audience cohort would convert on a $42 skincare SKU. The data was vanity, not verdicts.
So we built the Performance-Score model to produce verdicts. It runs four independent scoring dimensions against every creator who applies to the WiiPals network, rejects the 71% who don't clear our thresholds, and leaves the brands we work with a short list of operators — not influencers — who have already demonstrated that their audience buys.
What follows is the documentation. It is long by design. Performance marketers we've worked with ask the same questions on every first call, so this page exists to answer them in writing, with citations, before the strategy call begins.
Chapter 02 · The Four Dimensions
Scored independently. Zeroed and recalculated quarterly.
Each creator in the index carries a 0–100 score on each of the four dimensions below. The Performance-Score is the weighted blend, recalibrated every quarter against client-side attribution data from Shopify, Klaviyo, Meta Ads, and TikTok Shop.
-
01
Authenticity
A composite measure of whether the audience is human, reachable, and not purchased. We score follower growth curves against platform baselines, flag engagement pods, run comment-lexicon analysis to detect bot signatures, and audit follower geography against declared content language.
-
02
Engagement Quality
Raw likes and views don't move revenue. We score the ratio of save, share, and dwell-time actions to passive impressions, weight comment depth over comment volume, and penalize creators whose engagement collapses on paid-boost posts versus organic ones.
-
03
Audience Fit
An audience is only valuable if it matches the brand's buyer. We score demographic overlap (age, geography, household income band) against the brand's historical customer file, then compare creator content categories against the product category's purchase-intent signals.
-
04
Conversion Likelihood
The most predictive dimension. We score a creator on their prior tracked conversion behavior — not just attributed revenue, but conversion rate per 1,000 reached and average order value of the buyers they send. This is the dimension that distinguishes a post from a purchase.
Operating Benchmarks · Calendar Year 2024
Creator rejection rate
71%
Source: WiiPals network intake, last 12 months
Median brief-to-launch
41h
Source: 1,800+ brand clients, 2024
Pairing activation rate
94%
Source: 12,400+ pairings since 2019
Hours saved per campaign
18h
Source: Average client survey, n=212
The numbers the methodology produces.
Chapter 03 · Data Posture
The only mid-tier creator platform with audited data handling.
WiiPals has held SOC 2 Type II certification continuously since 2022. Creator payout data, brand briefs, and attribution events flow through systems that are independently audited annually — not self-attested. For performance-marketing teams with a security review process, the audit report is available under NDA before the first call.
Compliance
SOC 2 Type II
Held continuously since 2022
Funding
$11.2M Series A
Led by Forerunner, Jan 2022
Team
47 people
Across 6 countries
Stack integrations — attribution and payouts in one tab.
- ShopifyOrder + LTV attribution
- KlaviyoFlow triggers on referred buyers
- Meta AdsWhitelisted creator ad sync
- TikTok ShopNative product-tag attribution
Chapter 04 · Buyer Objections
Questions performance marketers ask before booking the call.
Q1 What does the Performance-Score explicitly exclude?
The score is not a measure of a creator's aesthetic, production quality, or personal brand polish. It also does not predict virality, specific follower-growth figures, or whether a single piece of content will break out on the algorithm. What it does measure — and what it is calibrated against — is whether a creator reliably produces attributable revenue for the brands we work with.
Q2 How often is the model recalculated, and against what ground truth?
Every creator's individual dimension scores are recomputed weekly using the freshest platform data. The weighted blend — and the threshold cutoffs that determine the 71% rejection rate — are recalibrated quarterly against the client-side attribution data piped in from Shopify, Klaviyo, Meta Ads, and TikTok Shop. We do not ship a calibration change without re-running the last six quarters of pairings through it.
Q3 Which verticals has the model been validated in?
Audited ground truth exists for DTC, beauty, fitness, and home goods — verticals where we have at least twelve months of client-side attribution data feeding back into the blend. Within those, the median client ROI in the first 30 days is 4.8×. For verticals outside that set, we will not quote a performance figure on the strategy call — we will instead walk through what we would measure during a pilot.
Q4 Can a rejected creator appeal the decision?
Yes — and roughly 14% of rejected applicants do, mostly through a 90-day waiting period in which they accumulate fresh platform data that may move them across a threshold. We do not promise that an appeal will overturn a decision; the threshold is calibrated against ground-truth revenue outcomes, not the creator's narrative about their own work. We do promise that the score breakdown is returned with the rejection notice, so the creator knows which dimension to address.