Quality Score
Every product on Teach Pair Share is scored across 7 independent signals that measure real quality, from teacher reviews to content originality. Here's how it works.
Products scoring 40+ qualify for organic Pair Credit. Scores help teachers discover stronger listings and support trust badge eligibility.
The Formula
7 signals, weighted by how much they reflect genuine quality. Nearly half comes from real teacher activity.
Quality Score = sum of each signal score multiplied by its weight
Each signal scores 0 to 100. The final score is the weighted total.
The 7 Signals
Review each signal to see how it is calculated and how we keep it fair.
What verified teachers think
Ratings from verified purchases, adjusted so new resources are not judged too harshly. Written reviews count more than star-only ratings because they give teachers more useful context.
How it's calculated
- Only verified teachers who completed a purchase can rate
- Written reviews (with body text) are weighted 1.5× compared to star-only reviews
- Bayesian adjustment pulls toward a 3.5 marketplace average so one early 5-star review does not overstate quality
- New products start at 62/100 (the Bayesian prior) and converge to their true rating as reviews accumulate
Fairness check
Ratings are locked to verified purchases. Multiple reviews from the same teacher on the same resource are prevented. The adjustment rewards consistent quality across many reviews.
What happens after teachers buy
Looks at whether a resource appears useful after purchase: purchase interest, download follow-through, and low support friction.
How it's calculated
- View-to-purchase rate (40% weight) - teachers who choose the resource after viewing it
- Support/refund ratio (40% weight) - low issue rates suggest the resource matches expectations
- Download follow-through (20% weight) - teachers who download are likely preparing to use the resource
- Only activates after sufficient data (10+ views, 5+ sales)
Fairness check
These signals come from real teacher actions, not creator inputs, so the score reflects how the resource performs after purchase.
How complete and helpful your listing is
A checklist of the information teachers need before they buy. Curriculum resources are checked for standards alignment and grade levels; visual products are checked for print dimensions and preview images.
How it's calculated
- Description quality (0–2 points) — measured by vocabulary diversity, sentence structure, formatting, product-relevant terms, and length
- Preview image uploaded
- Standards alignment (curriculum) or print dimensions (visual)
- Grade levels (curriculum) or colour/theme info (visual)
- At least 3 tags for discoverability
- File format specified
- Multiple preview images ≥ 3 (visual) or file uploaded (curriculum)
Fairness check
Description quality is scored on vocabulary diversity, sentence variation, structure markers, and product-relevant vocabulary — not just character count. Copy-pasting the same sentence over and over scores poorly because unique word ratio drops.
How accessible and well-formatted your product is
Rewards products that provide multiple file formats, quality descriptions, and appropriate metadata. The more accessible your product is to different teaching environments, the higher this signal.
How it's calculated
- File format diversity — offering PDF + Google Slides + editable format scores highest
- Preview images present
- Description quality (structure, vocabulary, usefulness)
- Grade levels specified (curriculum) or print dimensions (visual)
- Standards alignment (curriculum) or multiple preview images (visual)
Fairness check
Format diversity is verified from actual uploaded files. Description quality uses the multi-signal scorer — structure markers and vocabulary diversity can't be faked with padding.
How original your content is
Every product submitted for review goes through automated originality analysis. We compare your listing against all other products in the same category using domain-aware NLP — stripping out shared marketplace vocabulary so "worksheet" and "Common Core" don't cause false flags.
How it's calculated
- SHA-256 file hashing catches exact file duplicates instantly
- Text hashing catches copy-pasted descriptions
- 5-word phrase shingling + MinHash fingerprinting screens for structural similarity
- TF-IDF cosine similarity provides precise comparison for candidates
- Per-category adaptive thresholds — each category has its own similarity baseline, so a high similarity is only flagged when it's significantly above the category norm
- Domain-aware stop words strip 150+ common educational terms before comparison
Fairness check
Exact duplicates score 0. High-similarity products are flagged for admin review with full context. The per-category adaptive threshold means you're compared to the variation within your specific niche, not the entire marketplace.
Community response and product traction
Measures whether teachers are finding the resource useful through helpful review votes and steady sales.
How it's calculated
- Helpful votes on reviews (60% weight) - when educators mark a review as helpful, it gives other teachers better context
- Sales volume (40% weight) - sustained sales over time indicate consistent value
- 10+ helpful votes is considered excellent for most products
Fairness check
Helpful votes come from verified teachers, not the creator. Sales are tied to real transactions.
How recently your product was updated
Products updated within the last 3 months score 100%. After that, a gradual decay encourages creators to keep content current — especially important for standards-aligned materials that may change with curriculum updates.
How it's calculated
- 0–3 months since last update: 100% (full freshness)
- 3–24 months: linear decay from 100% to 0%
- 24+ months without update: 0% (considered stale)
- Any update resets the freshness clock
Fairness check
The decay is gradual and the window is generous. Making a trivial edit does reset the clock, but this is by design — we want creators to at least review their content periodically.
Example: "5th Grade Fractions Unit — Common Core Aligned"
Here's how a well-crafted curriculum product might score across all 7 signals:
4.2 average from 18 reviews (12 written), pulled slightly toward 3.5 prior
6% view-to-purchase rate, 2% refund rate, strong download follow-through
6/7 checklist items met — description quality adequate but not yet detailed tier
PDF + Google Slides, preview images, 3 grade levels, 4 standards
Below category similarity baseline — unique approach to fractions instruction
3 helpful votes, 18 sales — growing but still building community traction
Updated 2 weeks ago with new practice problems
This product qualifies for organic Pair Credit and is eligible for the Highly Original badge.
How Originality Detection Works
Every product submitted for review goes through our automated originality pipeline. Here's what happens behind the scenes:
File Hash Check
We compute a SHA-256 hash of your uploaded file and check it against every other file in the marketplace. Exact file duplicates are caught instantly — score: 0.
Text Hash Check
Your product description is normalised (lowercased, whitespace collapsed) and hashed. Identical descriptions are caught even if formatting differs.
Domain-Aware Preprocessing
We strip 150+ common educational marketplace terms — "worksheet", "lesson plan", "Common Core", "printable", etc. — so shared vocabulary doesn't cause false positives. Two products about fractions will naturally share domain words; we only care about unique phrasing.
Phrase-Level Fingerprinting
Your description is broken into 5-word sequences (shingles) and compressed into a MinHash fingerprint. This lets us quickly estimate how much structural overlap exists with other products in your category.
Precise Similarity Analysis
Candidate products that pass the fingerprint screening are compared using TF-IDF cosine similarity — a standard NLP technique that weighs rare, distinctive words more heavily than common ones.
Adaptive Threshold
Each category has its own similarity baseline. A product is only flagged when its similarity is significantly above the category norm (2+ standard deviations). This prevents false flags in categories where products naturally share more vocabulary.
Designed To Stay Fair
45% from real teacher activity
Nearly half the score comes from what teachers actually do: ratings, reviews, purchase interest, and support outcomes.
Description quality, not quantity
We score vocabulary diversity, sentence structure, formatting, and product-relevant terms - not just character count. Repeating the same sentence does not make a listing more helpful.
Domain-aware originality
We strip shared marketplace vocabulary before comparing products, so legitimate resources on similar topics are not treated as duplicates. Each category has its own comparison baseline.
Bayesian adjustment
A single 5-star review doesn't give you a perfect rating score. The Bayesian prior pulls toward the marketplace average until you have enough reviews for statistical confidence.
Multiple independent signals
Improving one signal, such as writing a longer description, cannot compensate for poor teacher reviews or repeated support issues.
Frequently Asked Questions
Ready to create resources teachers trust?
Use the Listing Quality Preview when creating your resource to see what is clear, complete, and ready to improve. The AI Listing Assistant can help draft helpful, specific descriptions.