How The 10-Minute Shelf™ is measured

Public product pages on named platforms, at named pin codes, about once a day for each product.

Only what Rewalto collected itself is published. Data from a third party, such as a price-history aggregator, is never published as ours. We may estimate the present from what we observed; we do not forecast.

Rewalto measures public retail surfaces. No retailer or brand named here has a relationship with Rewalto unless that is stated. What is being collected today, platform by platform, is on the Data page; this page says only what the publication is allowed to use of it.

The last read of the day counts for that product on that day. A category figure is a share or a median across the products observed. Brand scorecards follow the separate Rewalto Score method below.

Every figure travels with four companions, on the same page. Definitions and calculations are numbered footnotes printed under Notes at the bottom of each report and page; the limitation and the coverage stay beside the figure:

• Denominator. What it is over: products, product-days, pin codes, platforms, brands.
• Period. The dates it covers, inclusive.
• Freshness. The date the underlying rows were last collected.
• Limitation. What was not collected or cannot be said.

The platforms a figure was read on sit directly beneath its title. Only The Index is quick-commerce only; every other section states its own platforms. A figure from a platform that is not ten-minute is never presented as a bare 10-Minute Shelf number, and a single average is never blended across the two.

Where a product is matched across platforms, the match is used only where the pack size agrees, and a match that has not been reviewed is labelled as such. Where we name brands in a ranking, the metric is defined in one sentence, the whole set is stated (“5 of 11”), only the top is named and ties are shown as ties.

A floor is the least a slice must clear before it may be published. A slice that misses is left out and named, never estimated and never shown as a shorter list.

These thresholds are publication rules, not observed results. Each finding must be checked against the applicable threshold before it appears on the publication. This page does not report the current eligibility of any slice.

The Rewalto Score is our own monthly score for one brand on one platform in one category over one closed calendar month. It describes how reliably a brand’s products could be bought across the locations we observed. It is not a measure of brand quality, sales or market share, and it is not a forecast. Method version rewalto-score-v1.

• Availability component, 60%. Equal-weight availability: for each listed, served product and pin code in a fixed panel, the share of known stock readings that were in stock, averaged with every product and pin code counting equally. Unknown and delisted readings are excluded. No sales weights and no search position take part.
• Distribution component, 40%. The mean of the daily share of served pin codes where a product could be bought, over the eligible product-days. Unknown and unserved pin codes are excluded, never counted as buyable. It is not a claim about national distribution.
• Why a weighted geometric mean. Both components must be strong for the score to be high, and a component of zero gives a score of zero. The weights are a publishing choice, fixed for this method version, and are changed only by a new version with a recorded note. The score is rounded to one decimal for display only.
• Scope. One brand, one platform, one category, one month. A brand on two platforms has two scores. Scores are never averaged across platforms, categories or months. Each Rewalto Score names its measured scope. Differences across months are not asserted where the panel changed.
• Inputs. Public retail surfaces only, never a client’s own data. The score is computed from reviewed aggregates; a score typed in by hand is rejected. No confidence interval or causal effect is implied.

A card that misses a floor, or lacks either component, is not scored. It says why and is never shown as zero; a measured zero is a real score of 0.0. Every card shows its products, pin codes, days, observed and expected cell-days, the eligible-cell and SKU-day denominators, its measurement date and its publication date. Panel coverage shows how much of the panel was read and is not part of the score. It is a different thing from the Coverage part of the brand Rewalto Score below, which is the share of a brand’s catalogue found on the channels we read.

Every brand is scored the same way. The formula and floors do not depend on whether a brand is a Rewalto customer, pays, joins, requests a correction or declines an offer. Monthly scorecards name brands, including low scores; this is the one exception to the rule below against naming a brand in a negative finding. Every scorecard states its scope, so a score is never read as a verdict on the brand.

Three report types carry a score, and each has its own method. Every part runs from 0 to 100 and the parts are combined with a weighted geometric mean, as above: a part that is 0 makes the score 0, and a part we could not read withholds the score, and is never filled in. Scores are never averaged across types, and a score never says anything about quality, sales or market share.

Brand Rewalto Score

• Coverage, 25%. For each channel we read, the share of the brand’s catalogue found listed there, averaged over the channels. A channel we do not read is named on the report and left out, never counted as zero.
• Distribution, 35%. Rewalto Score v1 from the section above: availability and reach across pin codes for the brand’s quick commerce slice.
• Visibility, 20%. The brand’s share of the top search results on a set of named search terms.¹
• Price and discount, 20%. Half is price consistency across channels for products matched with the pack size confirmed. Half is price stability over the month. Discount depth is shown beside the score and is not scored, because a deeper discount is not a better score.
• The brand’s own site audit is shown beside the score and is never inside it.

Platform Rewalto Score

• Coverage, 25%. The brands and products listed on the platform out of all we saw on any platform in the same categories.
• Time, 30%. Half is the median delivery promise and half is how steady that promise stayed. We read promises, not delivered orders.
• Served reach, 25%. The share of the pin codes we probe that the platform serves. It says where a shopper can order, not how many shoppers there are.
• Brands and products, 20%. How deep the range is and how much of it is new.

Shelf health

Shelf health is the category score. It appears on category reports only and is never ranked against another category. It describes the shelf and does not say a category is good or bad.

• Availability, 30%. Equal-weight in stock share, as in Rewalto Score v1.
• Competition, 25%. How openly brands share the shelf.
• Price stability, 25%. The share of products whose price held still between read days.
• Search crowding, 20%. The share of search results that were not paid placements.

A brand without a Rewalto account. A brand report on a draft page may show stock readings labelled “Preview reading”, with the coverage beside them. A preview reading is never a Rewalto Score, and nothing of it reaches a public page that search engines index.

The data reports are a separate, rolling product from the monthly score. They read seven days, so they never mix with a monthly edition, and each page states the day it was last computed from. Nothing shorter than seven days is published.

A page appears only when its slice clears a floor: for a listing, at least four days seen and, on apps read by pin code, three pin codes; for a category on an app, 30 SKUs and five days; for an app, five days and three categories or 500 SKUs. A slice under a floor is listed on the report’s type page with the reason and is never estimated.

Figures are shown per app and are never averaged across apps. A figure that looks wrong is held and checked before it is shown: a median discount above 60% for an app or category, above 90% for one listing, or a count that moved more than threefold since the last computation. A reading with no MRP is left out of every MRP ratio, and a price above a real MRP is only a candidate until a person has checked it against its source.

A listing appears on these pages only for a neutral fact, such as a price, an MRP or a rating the app itself shows. It carries no score and no ranking, and no seller is ever named.

Star ratings are the rating an app shows for a product, and only that: we hold no review text. A rating of 0 means not rated yet and is never counted, and a listing needs at least 10 ratings where the app shows a count. The Shelf score is 35% availability, 25% price stability, 20% content completeness and 20% rating, on a scale of 0 to 100. An app or category figure is the median of its products and is published only when at least half of its eligible products carry a rating, always with how many were scored. A product with any part missing is unscored, never zero, and no product is given a score of its own.

What we do not read is part of every figure. These gaps are structural, so they hold whichever slice is being published; the live picture is on the Data page.

• Flipkart carries no stock signal we can read, so no Flipkart stock figure is published anywhere. Its prices can appear; its availability cannot.
• A category read only outside quick-commerce appears in The Wider Shelf and never in The Index.
• A platform listed as “once it is collected” is not part of any figure until it is being read.
• A category we do not collect is absent. It is not estimated from a neighbour.
• Nothing sub-daily is published. The largest lane refreshes a product about once a day.

• A brand in a negative editorial finding. Findings publish at category level, and a brand is named as a winner, with the metric, the denominator and the full set it was measured against. The one exception is the monthly Rewalto Score, which names every brand that clears the evidence floors, low scores included, under the same formula, with its scope and evidence beside it.
• A bottom list. Ordering names the top only, and scores are never ranked.
• A blended number that mixes a ten-minute platform with one that is not.
• A single search-result position. We publish an average rank across platforms, never the position of one result on one search.²
• Anything from a client’s own data, anonymised or not.
• A person. No reviewer, seller, proprietor or employee is named, and no review is quoted verbatim.
• A forecast, or a claim that a retailer is a partner of ours.

If a figure is wrong, or a page says something it should not, tell us through the contact page and include a link to the page, the figure in question and any supporting evidence.

A corrected figure is marked as corrected on the page where it appeared, with the date, not replaced silently. A change to what a figure means is recorded here rather than edited in without a note.

Changes to this method

• 29 Oct 2026. A person now checks each candidate finding against its source.
• 27 Sep 2026. Added the rules for star ratings and the Shelf score on the data reports.
• 26 Sep 2026. Added the data reports: a rolling seven-day read, separate from the monthly score, with its own floors.
• 24 Sep 2026. Added the three scores on report pages. Defined Visibility as a share of results. A brand report may now show an average rank across platforms; the position of one result on one search stays unpublished.
• 21 Sep 2026. Version 1 published.

Back to The 10-Minute Shelf