Glossary

Scan · Match · Winner · Wintraits Index · Wintraits Match Index · Wintraits Scan Index · Wintraits Success Ratio · Times more frequent · 95% interval · Sealed exam · Validated finding · Winning gap · Trap · Correlation, not causation · Match axes and attributes

Scan

A snapshot of the podium: what today's best sellers in a category have compared with those that sell less. It mixes veteran and new products, so it describes what today's sellers look like, not what makes a launch win. It is exploratory and does not diagnose your listing.

Match

Your product against the launches that won: we follow each launch from its first day and test the model with a sealed exam. It is what the diagnosis, the comparison and the concept evaluation use.

Winner

A launch that takes off. We measure its sales rank (the median between days 60 and 180 after release) and compare it only with those launched in the same quarter: the top 20% are the winners. That way we do not reward products that have been selling for years.

Wintraits Index

A 0-100 score of how closely an Amazon listing resembles the best ones in its category. It has two bases: the Wintraits Scan Index, against today's best sellers, and the Wintraits Match Index, against the launches that succeeded. We always give it with the reference band. It measures resemblance, not the probability of sales.

Wintraits Match Index

A number from 0 to 100 that says how similar your listing is to those of the winners in your category, looking at the attributes that set them apart. It is not the probability of selling: it is similarity.

Wintraits Scan Index

A number from 0 to 100 that says how closely your listing resembles today's best sellers in a Scan category: it looks at the traits that set them apart, weighting the top sellers more, and 100 is the most complete best-seller listing. We give it with the middle half of best sellers and of the rest, computed out of sample. It is a snapshot of today, not a probability of sales. How it is calculated

Wintraits Success Ratio

Out of every 100 launches in a category, how many become winners (and how many fail), by quarter. Only for groups of 30 launches or more.

Times more frequent

How many times more often an attribute appears among the winners than among the rest. “×2” means twice as frequent; “×0.5”, half as frequent. In statistics this is called “lift”.

95% interval

The range within which the true value lies with 95% confidence. If the interval does not include “no difference” (×1 in “times more frequent”), the difference is unlikely to be chance. We only publish differences like that.

Sealed exam

Before examining the model, we write down and date how we will judge it. Then we test it only once with products launched after training it, and publish the result whatever it is.

Validated finding

A pattern we have seen again in later launches, not just in the data where we found it.

Winning gap

A combination of attributes with more winners than usual that hardly anyone uses in recent launches. It can be an opportunity, if it is true of your product.

Trap

An attribute with fewer winners than usual, also among new products. Widely used does not mean it works.

Correlation, not causation

Winners having something in common does not prove that it makes them win. Recommend only changes that are true of your product.

Match axes and attributes

The six axes of the Match radar and the attributes each one groups. AI reads them from the listing (title, bullet points and description) or they are calculated from the price and size. If the listing does not mention something, it stays empty: we do not make it up.

Formula and actives

What the product contains and how much: the active ingredients the listing highlights, the concentration of the main one, how many different actives it has and the sun protection factor (SPF), if any.

Example: “Serum with 15% vitamin C and ferulic acid” scores on this axis; a “moisturising face serum” with no named actives does not.

Key ingredients

The actives the listing highlights: hyaluronic acid, vitamin C, niacinamide, retinol, peptides, centella asiatica…

Concentration of the main active

The percentage of the main active, if the listing states it (10 in “niacinamide 10%”).

Number of actives

How many different actives the listing highlights.

Sun protection (SPF)

The sun protection factor, in the categories where it applies.

Format and use

What the product is like and how it is used: packaging, texture, time of use, application, skin or hair type it targets, volume and units per pack.

Example: “30 ml dropper, gel texture, day and night, for oily and combination skin”.

Packaging

Dropper, pump dispenser, airless, ampoules or tube.

Texture

Gel, watery, oil, light cream or milk.

Time of use

Day, night, or day and night.

How to use

How it is applied, in the categories where it applies (for example, rinse-off or leave-in).

Skin type

Which skin types it targets: all, oily, dry, combination, sensitive, mature or acne-prone.

Hair type

Which hair types it targets, in hair-care categories.

Volume

Millilitres in the pack.

Units per pack

How many units it contains; 1 if single.

Claims

What the listing claims about what the product is or how it was tested: labels and guarantees, not results on the skin.

Example: “Dermatologically tested”, “vegan”, “cruelty free”, “fragrance free”, “paraben free”, “hypoallergenic”, “made in Spain”. By contrast, “reduces wrinkles” is a promise (Promise axis).

Claims

Stated labels and guarantees: vegan, cruelty free, fragrance free, paraben free, natural or organic, dermatologically tested, hypoallergenic, clinically proven, made in Spain or in Korea.

Listing (text and images)

How the listing is told on Amazon: length of the title and description, number of bullet points and images, clarity and tone of the text, and how much product information it gives.

Example: A title that says what it is, the active and the size, five complete bullet points and seven images, versus a short title with two photos.

Listing quality

Clarity and completeness of the text, from 1 (very poor) to 5 (excellent), as read by AI.

Listing tone

Technical (data and actives), emotional (feelings and results), mixed or poor.

Title length

Characters in the title.

Description length

Characters in the product description.

Number of bullet points

How many bullet points (“About this item”) the listing has.

Number of images

How many images the listing has.

Content index

How much the listing offers compared with products launched earlier: it combines number of actives, concentration of the main one and listing quality.

Value for money

What it costs versus what it offers: price, price per ml, price versus products of a similar size, and how much content and how many promises it gives for that price.

Example: Two 30 ml serums at €19.90: the one with 10% niacinamide and two more actives scores better on value for money than the one that names none.

Price per ml

Price divided by the total ml (volume times units).

Relative price

Price per ml compared with the median of similar-sized products launched earlier: 1 = the same; 1.5 = 50% more expensive.

Value for money

Content index divided by the relative price: more content for the same price gives a higher value.

Promises per price

Number of promised benefits divided by the relative price.

Price

Selling price in euros.

Promise

What result it promises the customer: the benefits (hydration, anti-ageing, radiance, anti-dark spots, pores…), how many it promises and whether it gives figures.

Example: “Hydrates and adds radiance” are two benefits; “reduces dark spots by 30% in 4 weeks” is also a promise with figures.

Promised benefits

The results it promises: hydration, anti-ageing, anti-wrinkle, radiance, anti-dark spots, anti-redness, anti-acne, pores, firmness, soothing or repairing.

Promise with figures

Yes, if the listing gives figures for results (for example, “−30% wrinkles in 4 weeks”).

Number of benefits

How many different benefits it promises.

More detail in the methodology.