Hand soaps: price moves rank here, but only just
I measured 225 price changes across 22 hand soaps on amazon.de, each against the products that held their price. Both directions are real and both are small: a 10 percent cut bought about 2 percent of rank, a 10 percent increase cost about the same. That puts hand soaps at the weak end of this series — demonstrably below seven of the ten categories published before it.
In short
- A 10 percent cut improved rank by about 2 percent (range 0.3 to 5). Small but demonstrable, across 225 price changes from 22 products.
- A 10 percent increase cost about 2 percent of rank (range 1 to 4). Symmetric and equally small — this category registers price without caring much.
- Only the extremes clear the no-change line. Cuts past 25 percent bought about 15 percent of rank; increases past 25 percent cost about 12. The four middle brackets are indistinguishable from no change.
- Against the series, this is the quiet end. The cut response sits demonstrably below seven of the ten categories published before it; against shampoo, hair oils and conditioner the gap stays an observation.
- The market answers in weeks, not hours. Median first competitor response: five days. Same-day movement: one cut in ten. And the corridor scoring returns no recommended entry band at all.
Price cuts
A 10% cut improved rank by roughly 0.3 to 5 percent.
Price increases
A 10% increase worsened rank by roughly 1 to 4 percent.
ONLY THE EXTREMES REGISTER. THE MIDDLE OF THE LADDER IS SILENT.
Hand soaps register price at the edges and shrug at everything in between. Cuts past 25 percent produced ranks about 15 percent better; increases past 25 percent, about 12 percent worse. Both extreme brackets clear the no-change line. The four brackets in the middle — every move between 5 and 25 percent in either direction — do not.
One wrinkle in the chart deserves naming rather than smoothing: the medium-cut bracket shows a better point value than the deep-cut bracket, 19 against 15 percent. Its range spans the no-change line on 22 events, so it is noise, and I read it as noise — but it is the kind of noise a small category produces, and hiding it would misrepresent how solid the rest is.
At category level both directions are demonstrable because the full 225 events pool what the brackets cannot: small effects, clearly present, of roughly a fifth the strength that eye creams or fragrances showed. If price is a lever here, it is a short one.
The full picture
01 · 1 · DOES DISCOUNT DEPTH MATTER?
Only at the far ends — the middle four brackets are indistinguishable from nothing.
TWO BRACKETS CLEAR THE LINE. FOUR DON'T.
Bracket by bracket: deep cuts at 15 percent better, range 5 to 24 — clear. Deep increases at 12 percent worse, range 2 to 20 — clear. Everything between ±5 and ±25 percent includes the no-change line, including the noisy medium-cut bracket discussed above. In a category this small, 18 to 55 events per bracket set the resolution, and the resolution says: moderate price moves produced no demonstrable rank effect.
Note what that does and doesn't license. It doesn't prove moderate moves are free — the category-level averages say price effects exist. It says the per-bracket evidence only becomes unambiguous past 25 percent in either direction.
02 · 2 · WILL COMPETITORS FOLLOW?
Slowly — the median first answer takes five days, in both directions.
ONE CUT IN TEN GETS A SAME-DAY ANSWER. THE REST WAIT DAYS.
Hand soaps and beard oils form the slow corner of this series together: median first answers of five days after cuts and increases alike, same-day shares of 9 to 10 percent, and reaction lifts of 1.13 and 1.12 over baseline across 289 cut and 264 increase triggers.
The combination with the elasticity numbers is coherent: a shelf where price barely moves demand is also a shelf nobody monitors hourly.
03 · 3 · WHERE WOULD A NEW PRODUCT ENTER?
The scoring returns no corridor — and the band data backs that refusal up.
THE FIRST CATEGORY WHERE THE SCORING ITSELF DECLINES TO RECOMMEND
No corridor: no price band in this category met the entry criteria, so none is marked recommended.
This is the first category where the corridor scoring itself returns nothing — no best band. Looking at the data, the refusal is earned: across all eight bands there is not a single judged recent entrant, no band beats the one-third baseline on more than two products, and the busiest shelf holds seven.
21 products did enter in the last 24 months and 4 are winning, so entry happens — it just leaves no price-band pattern this analysis can responsibly generalise. A tool that knows when to say nothing is worth more than one that always answers; this week it said nothing.
04 · 4 · WHAT DOES THE MARKET LOOK LIKE UNDERNEATH?
A small, slow shelf where half of all price changes were too disturbed to use.
The fit set spans €2.40 to €18.96 in three small tiers — budget €2–7, mid €8–12, premium €13–19, with 7 to 8 products each. The gap analysis covers 55 products; the busiest band is €11.31–€14.64 with seven.
21 products entered within the last 24 months, 4 are winning, none has exited. That one-in-five win rate on a tiny denominator matches the mid-field of the series — this niche admits and keeps the occasional newcomer.
The most striking figure sits in the funnel: 51 percent of all detected price changes had to be excluded, the highest disturbance share in this series. The dominant single reason is control scarcity — 168 of 553 events failed for lack of five quiet comparator products, the structural tax of a 71-product set. The 225 events that survive clear this series' gates, with little room to spare.
Data appendix
The set started from an Amazon category export of 717 products in the hand soap path on amazon.de, curated on 19 August 2026 under the frozen rules of this series. In order: 98 foreign types removed — bar soaps and wash bars, foam soaps, no-touch systems, curd soaps; 100 products outside the 200–500 ml single-pump-bottle band, with 281 doubt cases — 142 unparsable titles and 139 refill formats, which are a different price-per-millilitre logic — held on a doubt list rather than guessed at; 57 multipacks; 4 children's lines; 51 outside the €2–20 price frame; 37 below 20 reviews; and 18 above the cap of three products per brand. That leaves 71 products from 47 brands. Category membership comes from Amazon's own tree: 54 of the 71 synced products sit in the majority path for hand soaps, 16 sit elsewhere in the tree, and their events were excluded from the figures where they occurred. One product returns no category path at all; the current guard treats it as conformant — a known limitation I plan to close. The product list and the curation record are available on request for methodological review.
| Step | Events | Removed because |
|---|---|---|
| Price changes detected | 553 | — |
| Gaps in the price history | −16 | the price series around the change is incomplete |
| Too little ranking data | −78 | not enough rank observations to measure an effect |
| Overlapping price change | −64 | another move by the same product inside the window |
| Deal or coupon markers | −0 | the change was a flagged promotion, not a price |
| Stockout signals | −22 | availability broke during the window |
| Burst of new reviews | −8 | a review spike could move rank on its own |
| Listing too young | −11 | less than 60 days of history |
| Too few control products | −168 | fewer than 5 quiet products to compare against — the structural tax of a small set |
| Passed every check | 272 | one event can fail several checks at once |
| Used for the figures | 225 | after removing 47 events with a pre-existing rank trend |
price change ≥ 5% and ≥ €0.30, held ≥ 7 days · comparison: 14 days before, days 3–14 after · control products: same category, price moved less than 2%, at least 5 of them · rank coverage ≥ 70% of days · listing older than 60 days · review burst = more than 10% and at least 15 new reviews · ranges are 90%, from a bootstrap clustered by product
Observed price changes ran from −72.8 percent to +234 percent. The extremes are mostly listings returning to a nominal price after being unavailable, not commercial decisions. The figures above are read from the dense middle of that range.
Limits
- LIMITS — I measure the ranking, not units sold, and only over a 14-day window after each change. This is one of the smallest fit sets in the series: 225 events from 22 products, which is why four of six depth brackets stay undemonstrated and why the series comparison against shampoo, hair oils and conditioner remains an observation. Competitor reaction and advertising interplay are not modelled, and a rank position says nothing about margin.
- 71 products
- 553 price changes found
- 272 passed every check
- 225 used for the figures
- 22 products in the fit set
- ref 2c9ea884
The Monthly
Each weekly report measures a single category. The Monthly is where they get compared, and it carries the quarterly report when one is published.