Free report 2026

What actually blocks growth.

1,675,833listings analysed

2.56 M listing versions · 45,177 listing-months of revenue · 352 deep analyses

The four states every listing sits in — ranked by impact, not by frequency. The result turns the usual order upside down.

As of 1 September 2026Marketplace Amazon.de

48.8 %of listings carry a purchase-blocking defectn = 55,431

4.6 %of listings carry 51.4 % of revenuen = 7,820

€789upside per hidden champion per monthmedian

4.27 %unit session percentage, medianfrom 30 sessions

The finding

Three quarters of listings carry one sixth of revenue

Every listing of one real month, sorted into four states. Then two questions asked separately: how many are there, and how much revenue do they carry? The two answers are near mirror images.

Three quarters of listings carry one sixth of revenuen = 7,820 listings · July 2026 · Amazon.de

How the listings split

75.2 %

14.5 %

5.7 %

4.6 %

How the revenue splits

15.2 %

17.7 %

15.7 %

51.4 %

Visibility issueHidden championConversion leakScaler

n = 7,820 listings with traffic, July 2026, Amazon.de. Cohort revenue that month: €1.38 M. Below 200 sessions visibility leads — without a denominator no conversion verdict holds. The reference is the 4.65 % median of the same month.

The most common problems are not the most expensive ones. And the most expensive ones are not what everybody works on.

4.6 % of listings carry 51.4 % of revenue. At the other end sit 75.2 % that together contribute less than a sixth. Working a catalogue evenly therefore spends time almost exactly inversely to impact.

The four blockers

Each state has its own lever and its own speed

Mixing them up is the most expensive mistake in catalogue work.

Hidden championThe biggest lever per listing

1,137listings · 14.5 %

€789median upside

€1,304upper quartile

9.33 %conversion — twice the benchmark

61median sessions

Visibility issueThe bulk

5,879listings · 75.2 %

€483median upside

€791upper quartile

2.33 %median conversion

10median sessions

Conversion leakThe fastest money

446listings · 5.7 %

€342median upside

€811upper quartile

1.84 %median conversion

476median sessions

ScalerWhere the money sits

358listings · 4.6 %

51.4 %of monthly revenue

€187median upside

9.47 %median conversion

384median sessions

Upside is computed differently per state and is shown as a median, never a sum — single extreme cases distort any addition. Visibility issue and hidden champion: sessions raised to the median of visible listings (440), at benchmark or own conversion. Conversion leak: the way from actual to benchmark. Scaler: 25 % more traffic at unchanged conversion.

Ranked by upside per listing the order is hidden champion, visibility, conversion leak, scaler. Ranked by count it is exactly reversed. And ranked by speed it is a third order again. That is why “optimise your listings” is such a weak instruction: it names none of the three.

The state of the catalogue

How many listings are actually broken?

Not “not optimal” — broken. A defect is something that mechanically impedes the purchase, independent of copy, price and category. Three of them can be checked on every listing.

How many listings are actually broken?n = 55,431 live listings · Amazon.de

At most 3 of 9 gallery slots filled28.7 %

Main image missing or not zoomable24.0 %

Variant without a parent SKU10.0 %

48.8 %carry at least one of these defects

13.3 %carry at least two

51.2 %are clean on all three

n = 55,431 live listings, Amazon.de. The title is deliberately excluded here: it is stored for only 7.3 % of the sample. Counted in for exactly those 4,030 listings, the share with at least one defect rises to 94.3 %.

Every second listing is broken before anyone even talks about copy. And the most common defect is not a wording problem but an empty surface: more than every fourth listing fills at most three of nine gallery slots. On almost every fourth main image Amazon does not even switch on the zoom, because the short edge sits below 1,000 pixels.

Gallery slots filled — Amazon shows ninen = 55,431 live listings · Amazon.de

7slots filled, median

13.4 %have the main image and nothing else

7.4 %use all nine slots

76.0 %of main images are zoomable

A gallery slot is a surface, not a file: the count is distinct image slots (MAIN, PT01 through PT08), not stored size variants.

This is the link to everything that follows: a listing with three images has nowhere to show a use case, a before-and-after or a size comparison. The bottleneck is not the idea, it is the empty slot.

The four states, one by one

Visibility: the biggest problem nobody treats as one

Three in four listings get a median of ten sessions across a whole month. Talking about conversion at ten sessions is not analysis, it is noise.

5,879listings — 75.2 % of the cohort

10median sessions, whole month

15.2 %of revenue, taken together

€483upside per listing per month

The advice “work on your conversion” simply does not apply here. Without a denominator there is no rate, and without a rate there is no verdict.

What works here is findability: search terms, variant wiring, category, product type. And it works slowly — the blocker with the longest path between action and effect. First by bulk, last by speed.

Hidden champions: the proof is already in

1,137 listings convert at 9.33 % — twice the benchmark — and get 61 sessions for it. That is the single biggest lever in the whole dataset.

1,137listings — 14.5 % of the cohort

€789upside per listing — the highest of all four

9.33 %conversion against a 4.65 % benchmark

61median sessions

A hidden champion is the only listing where buying traffic is not a bet.

In every other state additional traffic is an experiment: whether it converts is open. On a hidden champion the conversion is already measured — across 61 sessions, twice as good as the market median. Buying visibility here means buying it against a known rate. That makes it the only state that scales profitably right away, and calculably so.

And this is exactly the state that stays invisible day to day: a listing with 61 sessions and six sales appears in no report, no bestseller list, no revenue ranking. It only shows up when conversion and sessions are looked at separately — and almost nobody does that.

Conversion leaks: not the biggest problem, but the fastest money

Only 5.7 % of listings are genuine conversion problems. That contradicts the entire consulting market — and changes nothing about the money sitting closest to hand here.

446listings — 5.7 % of the cohort

476median sessions — the traffic is already there

1.84 %conversion against a 4.65 % benchmark

€342upside per listing per month

What sets this state apart is when the effect lands. Visibility has to be built first; a hidden champion needs budget or ranking. A conversion leak already has the traffic — bought, delivered, paid for. It runs through the page every day and leaves again.

476 sessions at 1.84 % means 467 people looked at the listing that month and did not buy. The cause demonstrably sits on the detail page, not before it — the click is a test already passed. That makes this the only state where a content change becomes measurable within the same month.

Scalers: 4.6 % of listings, 51.4 % of revenue

This is where the money is buried, quite literally: half of monthly revenue hangs on 358 listings.

358listings — 4.6 % of the cohort

51.4 %of monthly revenue

€709,024actual revenue that month

€187upside per listing — the lowest of all four

Where the money sits is not where the growth sits. Confuse the two and you polish your bestsellers while nothing moves.

The last number is the uncomfortable one: scalers have the smallest upside per listing. They already run above benchmark, their conversion sits at 9.47 %, their traffic is there. There is little to repair — which is precisely what makes them scalers.

First, risk: a catalogue whose revenue half rests on 4.6 % of its items is not protected against a suspension, a stockout or a new competitor on exactly those ASINs. Second, attention: these 358 listings get the most care day to day because they top every report — and they are the place where one more hour of work moves the least.

The right work on a scaler is therefore not optimisation but defence and replication: protect what carries the revenue, then understand why it works and carry that reason over to the hidden champions.

The diagnosis

Six phases, 36 core questions — and a very clear pattern

Conversion leaks need a cause, not a guess. The SWF framework splits every purchase decision into six phases and checks six core questions per phase: answered or not. No model sets a score.

What almost every listing managesn = 352 complete analyses · 36 core questions

see_1Is it instantly clear what the product is?94.9 %

see_4Is the product understood without a long explanation?94.4 %

Same listings, different question

want_1Is it shown how the product improves the buyer's life?20.5 %

want_6Is there a before-and-after feeling?4.1 %

want_3Are benefits made emotionally and visually tangible?2.4 %

The German Amazon catalogue explains its products. It does not sell them.

The diagnosisn = 352 analyses

Phase	What it checks	Median	below 60
SeeStand out	Is the product seen and understood?	60	38.9 %
RelevanceApply	Does the buyer recognise their situation?	50	71.3 %
WantDesire	Does a result image and a desire form?	50	88.9 %
FixResolve	Are concerns, dimensions and contents settled?	65	29.0 %
TrustBelieve	Is there proof instead of assertion?	55	52.6 %
CloseDecide	Does one buying argument stick?	60	47.7 %

Want is the weakest phase in 218 of 352 listings — 61.9 % of cases; relevance follows far behind at 18.2 %. Overall score median 58 out of 100; only two analyses in the entire sample reach 70 or more.

Fix is the strongest phase. German sellers are good at removing doubt: dimensions, material, contents, warnings. That is mandatory work and it gets done. The desire still does not form — because nobody is told, without contradiction, why they should want the product.

The gaps

The eight most expensive gaps

The core questions answered least often — a list that can be worked through today.

The eight most expensive gapsn = 338–352 analyses per question

see_2Does the product stand out against competitors?1.1 %

want_3Are benefits made emotionally and visually tangible?2.4 %

want_6Is there a before-and-after feeling?4.1 %

trust_3Are reviews, seals and guarantees put to work?4.4 %

trust_5Is there social proof or a real usage example?5.9 %

relevance_2Is a concrete use case shown?9.5 %

close_3Is there a clear reason to buy now?11.5 %

want_2Does a clear result image form?13.3 %

Six of the eight gaps are closed with images, not with copy. Use case, before-and-after, result image, social proof: those are gallery slots and A+ modules — the very surfaces that sit empty on 28.7 % of listings.

see_2 at 1.1 % is the hardest single finding. The question is not “is the image good” but “does it stand out in the results list next to the others”. Almost nobody checks their main image in the context of its competition — only on its own, in an image editor.

The market

Why visibility is so expensive

Across the 60,000 most-searched terms attention is distributed extremely unevenly — and the concentration gets stronger towards the long tail, not weaker.

Why visibility is so expensiven = 60,000 search terms · Amazon.de · snapshot 31 July 2025

36.1 %of clicks go to the top 3, median

29.8 %of terms: the top 3 take more than half

8.9 %conversion of the most-clicked product

Rank 1–1,000 → 30.1 % click share, rank 10,000–50,000 → 36.6 %. The niche term is not the easier market, it is the narrower one. Only one snapshot available — read it as a moment, not a trend.

The most-clicked product of a term converts at a median of 8.9 % — almost twice the median of the overall catalogue. Visibility and conversion are therefore not separate problems: holding the top 3 buys better traffic, not just more. Missing them means fighting for the same rate with worse traffic.

Unit session percentage — the comparison tableUnit session percentage · n = 39,144 listing-months · 40 sellers · Dec 2024 – Jul 2026

Threshold	p10	p25	Median	p75	p90
from 30 sessionsn = 16,603	0.99 %	2.17 %	4.27 %	8.51 %	15.08 %
from 100 sessionsn = 7,991	0.81 %	1.82 %	3.74 %	7.44 %	13.50 %
from 300 sessionsn = 3,206	0.64 %	1.62 %	3.50 %	7.00 %	13.02 %

The more traffic a listing carries, the lower its median — broad traffic is blunter traffic. Traffic-weighted across all sessions: 6.02 % (5.80 M sessions, 348,898 units). 4.2 % of listing-months with at least 30 sessions end without a single sale.

The limits

What stays open in 2026

A higher SWF score ought to mean more revenue. The available data does not support that — so it is not stated as a claim.

What stays open in 2026

Approach	n	Result
Cross-sectionSWF score against conversion, across listings	59	Rank correlation −0.03 — no relationship demonstrable
Before / afterthe same listing before and after publishing	11	Median +49 % relative, 73 % improved — too small to claim

The cross-section cannot work: conversion depends on price, category, reviews, brand and ad pressure — all of which vary more between two listings than their content quality does. The sound route is before-and-after on the same product. What is missing there is cases, not method.

This is not an admission but the strongest position in the report. The 2027 measurement design is already fixed: at least 200 listings, two months either side of the change, control group drawn from unchanged listings of the same seller.

The 2027 report has therefore already assigned its most important headline.

Method

Data basis

Every figure in this report has its own population. They are listed in full so any statement can be traced back.

Data basis

Analysis	n	Basis
Catalogue structurevariants, EAN, product type, brand, fulfilment	1,675,833	full census across all listings
Listing versionscorpus size	2,562,584	full census
Title length75-character rule since 27 July 2026	136,961	all listings with a stored title
Quality scoresix content surfaces	83,000	sample of live versions
Defect and gallery countmain image, filled image slots, variant wiring	55,431	sample of live listings, Amazon.de
Traffic and revenuesessions, orders, revenue, returns	45,177	listing-months across 20,908 listings, 40 sellers
State distributionthe four growth blockers	7,820	listings with traffic in July 2026
Search termsclick share, conversion of the top 3	60,000	search frequency report, snapshot 31 July 2025
SWF deep analyses36 core questions per listing	352	completed analyses across 258 listings

Median, never a sum

Any addition over upside is carried by a handful of extreme cases — the sum across all conversion leaks would be €637,286 for one month, €71,310 of it from a single listing with 63,186 sessions and 0.11 % conversion. That is an ad-targeting fault, not an opportunity.

Every figure with its base

No percentage without an n and a date, not even inside a chart.

Missing data is not zero

Where a field is empty that is stated, not computed. A+ coverage is therefore absent from this report: the flag is empty on 99.4 % of versions, and “99.8 % without A+” would have measured missing data, not missing A+.

Second half

Unlock the complete report for free

The four states one by one — what to do in each

The diagnosis ends here. What follows is the work: the four states with lever and order, the 36 core questions, the eight most expensive gaps, the market figures and the full data basis.

The four states one by one, with lever and speed
Six phases, 36 core questions — why listings do not sell
The eight most expensive gaps, as a checklist
Click concentration and the conversion comparison table
What stayed open — and the complete data basis

Check your own catalogue against these numbers.

Which of the four states your listings sit in shows up after the first catalogue scan — with upside per listing, not as an average.

Request access How it works