VerdictsGuidesMCP

Start free

TickClip currently earns no affiliate commission from retailer links. It does not sell placement or let sellers pay to change rankings, scores, or verdicts.

support@tickclip.ai

Explore

  • How it works
  • Features
  • Deals
  • Verdicts
  • Guides
  • Try TickClip AI

Trust

  • Methodology
  • Independence

Company

  • About
  • MCP
  • API & feeds

© 2026 TickClip. All rights reserved.

PrivacyTerms
VerdictsGuidesMCP

Start free

OverviewCategoriesBrandsRetailersReview integrityReportsSellersDealsPromo codesMethodology

How the Shopping Index is computed

Every metric below has one definition, one eligibility rule and one minimum sample size, and the index computes nothing of its own. A page, a report, the API and a download for the same release read the same frozen rows.

Methodology 1.0 · Schema 1

Metric definitions

Verdicts

verdict.count

How many products TickClip has published a Tick, Clip or Skip for, inside the selected filter. One product counts once no matter how many listings or sellers it has.

Numerator

Eligible published verdicts

Denominator

n/a

Eligibility

Depends on the audience, and the audience is ALWAYS reported in the result's `filters.visibility`. `public` (every shopper-facing surface) counts only catalog-eligible rows: a real ASIN, renderable product data, and a passed publication gate. `all` (admin monitoring) counts everything analysed, including rows the catalog rejected. A number without its visibility is not this metric.

Counted over

product

Minimum sample

None. A count of zero is a fact rather than a small sample.

Currency treatment

Not a money metric.

Products

catalog.product_count

How many distinct products are in the catalog. One product sold by five sellers counts ONCE. Never show this under the word 'listings'.

Numerator

Distinct eligible products

Denominator

n/a

Eligibility

Same as verdict.count.

Counted over

product

Minimum sample

None. A count of zero is a fact rather than a small sample.

Currency treatment

Not a money metric.

Listings

catalog.listing_count

How many distinct offers are in the catalog. One product sold by five sellers counts FIVE. Never show this under the word 'products'.

Numerator

Distinct eligible listings

Denominator

n/a

Eligibility

Listings attached to an eligible product. Each listing counts once even if it appears under several variants.

Counted over

listing

Minimum sample

None. A count of zero is a fact rather than a small sample.

Currency treatment

Not a money metric.

Sellers

seller.count

How many DISTINCT sellers TickClip has recorded operating on this retailer. One seller running two storefronts counts ONCE. This is not the storefront count and must never be shown under that word.

Numerator

Distinct sellers behind the recorded storefronts

Denominator

n/a

Eligibility

Sellers with at least one storefront recorded on this retailer. It counts what TickClip has MODELLED, not what exists on the retailer, and a surface showing it has to say so.

Counted over

seller

Minimum sample

None. A count of zero is a fact rather than a small sample.

Currency treatment

Not a money metric.

Storefronts

store.count

How many storefronts TickClip has recorded on this retailer. A storefront is one seller's shop ON a retailer, so two storefronts can belong to one seller. This is not the seller count.

Numerator

Recorded storefronts on this retailer

Denominator

n/a

Eligibility

Storefronts attached to this retailer. Like seller.count, it reflects what TickClip has modelled rather than what exists.

Counted over

store

Minimum sample

None. A count of zero is a fact rather than a small sample.

Currency treatment

Not a money metric.

Brands

brand.count

How many DISTINCT brands appear across the eligible listings in the selected filter. A brand with forty listings counts once. Unbranded listings are not counted as a brand.

Numerator

Distinct brand keys on eligible listings

Denominator

n/a

Eligibility

Eligible listings that carry a brand key. Two spellings of one brand normalise to one key and count once.

Counted over

product

Minimum sample

None. A count of zero is a fact rather than a small sample.

Currency treatment

Not a money metric.

Tick rate

verdict.tick_rate

The share of eligible verdicts that say Tick (buy now). Tick, Clip and Skip rates always add up to 100%.

Numerator

Eligible verdicts with action = tick

Denominator

All eligible verdicts in the same filter

Eligibility

Same as verdict.count.

Counted over

product

Minimum sample

5

Currency treatment

Not a money metric.

Clip rate

verdict.clip_rate

The share of eligible verdicts that say Clip (wait for a better price). Tick, Clip and Skip rates always add up to 100%.

Numerator

Eligible verdicts with action = clip

Denominator

All eligible verdicts in the same filter

Eligibility

Same as verdict.count.

Counted over

product

Minimum sample

5

Currency treatment

Not a money metric.

Skip rate

verdict.skip_rate

The share of eligible verdicts that say Skip (do not buy this). Tick, Clip and Skip rates always add up to 100%.

Numerator

Eligible verdicts with action = skip

Denominator

All eligible verdicts in the same filter

Eligibility

Same as verdict.count.

Counted over

product

Minimum sample

5

Currency treatment

Not a money metric.

Clean rate

integrity.clean_rate

The share of a brand's analysed products where TickClip found no pricing-manipulation signals. This is about the listing's pricing behaviour, not about product quality.

Numerator

Eligible verdicts with no manipulation flags

Denominator

All eligible verdicts for the brand

Eligibility

Same as verdict.count.

Counted over

product

Minimum sample

5

Currency treatment

Not a money metric.

Flag rate

integrity.flag_rate

The share of a brand's analysed products where TickClip found at least one pricing-manipulation signal. Exactly the inverse of the clean rate, over the same records. Internal framing; the public surface shows the clean rate.

Numerator

Eligible verdicts with one or more manipulation flags

Denominator

All eligible verdicts for the brand

Eligibility

Same as verdict.count.

Counted over

product

Minimum sample

5

Currency treatment

Not a money metric.

Average score

verdict.average_score

The mean TickClip score (0-100) across eligible verdicts. Verdicts with no score are left out of both the top and the bottom of the calculation — they are not counted as zero.

Numerator

Sum of scores, over verdicts that have one

Denominator

Count of verdicts that have a score

Eligibility

Same as verdict.count, further restricted to records with a valid numeric score.

Counted over

product

Minimum sample

3

Currency treatment

Not a money metric.

Average confidence

verdict.average_confidence

The mean evidence-coverage confidence (0-100) across eligible verdicts. Confidence measures how much evidence we had, not how sure we are. Verdicts with no computed confidence are excluded, never treated as zero.

Numerator

Sum of confidence values, over verdicts that have one

Denominator

Count of verdicts with a computed confidence

Eligibility

Same as verdict.count, further restricted to records with a computed confidence.

Counted over

product

Minimum sample

3

Currency treatment

Not a money metric.

Average rating

product.average_rating

The mean marketplace star rating across eligible products, counting each product once. A product with no rating is excluded from the calculation entirely — it never enters as a zero, which would drag the average down.

Numerator

Sum of ratings, over products that have one

Denominator

Count of products with a rating

Eligibility

Same as verdict.count, further restricted to products with a rating greater than zero.

Counted over

product

Minimum sample

3

Currency treatment

Not a money metric.

Average discount

listing.average_discount_percent

The mean of listing.discount_percent across eligible listings in the selected filter. A listing whose discount cannot be calculated (missing, zero or inverted reference price) is EXCLUDED from the average rather than counted as a zero discount — counting it as zero would drag the figure down with listings that simply have no reference price.

Numerator

Sum of each eligible listing's own discount percentage

Denominator

Listings whose discount could be calculated

Eligibility

Listings for which listing.discount_percent returns a number. Both prices are in the listing's own currency, so the percentage is currency-free and the mean is safe across currencies.

Counted over

listing

Minimum sample

3

Currency treatment

A ratio of two amounts in the same currency, so the unit cancels.

How figures are split

A release splits its figures along these dimensions and no others.

  • overall
  • category
  • brand
  • retailer
  • verdict
  • price_band
  • currency

Price bands

Bands are applied within a currency and never across them. A listing priced in one currency is never placed in a band computed from another.

  • 0-25
  • 25-50
  • 50-100
  • 100-250
  • 250-500
  • 500-1000
  • 1000+

What is deliberately not published

condition

A verdict row records alternate-condition PRICES but not the condition the verdict was reached for, so every row would fall in one bucket labelled 'new' that nothing verified.

country

There is no country on a verdict row. Marketplace is a near proxy and is published as `retailer` instead of being relabelled as geography it does not measure.

deal_type

Deals and promo codes live on their own tables and are counted as units, not joined per verdict. Joining them here would read them at a different moment from the prices beside them.

review_trust_level

Review integrity is published as a clean/flagged rate. Cutting it into named 'levels' would invent thresholds the metric dictionary does not define.

seller

TickClip publishes no seller trust score and does not rank sellers. A per-seller verdict split on a public index is a trust score wearing a table, so sellers are counted and never broken down. The aggregate figure over all attributed sellers is published instead.

store

Same rule as sellers: a storefront is a business, and a public per-store outcome table would rank it.

Known limitations

  • Coverage is Amazon-first. Products from other retailers are not analysed, so this is not a sample of the retail market.

  • Price history comes from Keepa. Where Keepa has no history for a product, TickClip has none either, and the price dimension is simply not measured rather than scored as zero.

  • Review-integrity signals are inferred from aggregate rating and review-count series. TickClip does not read individual review text and does not claim to detect fake reviews.

  • Verdicts are point-in-time. Each row carries the snapshotId it was published under and the timestamps it was true at; a verdict is not a durable statement about a product.

  • Coverage is not a random sample. Products enter the catalog because a shopper asked about them, so popular products are over-represented.

  • Routine snapshots are pruned to the most recent few per verdict; verdict-CHANGING snapshots are kept permanently. A release old enough to reference a pruned routine snapshot will not reproduce byte-identically, which the checksum will reveal.

Licence

CC BY 4.0