Reading a row in ninety seconds

One row, four fields, one verdict before the kettle boils

The ninety-second read has a fixed order, and the order is the method. You look at the link, then the size figure, then the packing note, then the date. Four fields, four questions, and a verdict at the end that is either keep reading or put the row down.

The reason to fix the order is that rows are read for different purposes by different people and the sequence is what stops you from being talked into a purchase by the parts of a row that were never evidence. A link is not evidence that the item is right. A size figure is not evidence that the seller still stocks that size. Those are separate checks and they take separate seconds.

Ninety seconds is not a marketing promise, it is a budget. Forty seconds for the link, twenty for the size figure, fifteen for the packing note, fifteen for the date and the sample. Anything that runs over budget gets written down and checked later rather than guessed at now.

The failure mode the fixed order prevents is the slow skim. A reader who starts at the top of a row and works down spends the budget on whatever happens to be longest, which is usually a packing note written by an enthusiastic maintainer, and arrives at the size figure with nothing left to spend on it. Fixing the sequence means the expensive field is always read properly, the cheap field is read quickly, and no field gets read twice.

1. Field one, the link: condition is that it resolves to the same item, pass standard is that the model name matches the row, re-check is a fresh tab with no stored session.
2. Field two, the size figure: condition is that it names the two points measured, pass standard is millimetres with a repeat count, re-check is your own tape on an item you already own.
3. Field three, the packing note: condition is that it records a request that was sent, pass standard is a slip or a scale photo, re-check is one message to the warehouse before the parcel is built.
4. Field four, the date and sample: condition is that both exist, pass standard is a date inside the current quarter and a sample you can see, re-check is the batch note at the foot of the table.

The ninety-second figures: 46 rows, three pairs, two numbers

A row is worth reading only if somebody measured something, so it helps to know what a measured row looks like on this desk. The footwear run published 46 rows, each measured on three pairs of the same model, using insole length in millimetres and the widest point of the foot in millimetres.

Two numbers per row are the whole content of the size claim, and the second number is the one people skip. The difference between adjacent labelled sizes on the measured rows runs between 8 and 11 millimetres, which is roughly the width of a thumbnail, and that is the amount a wrong guess moves you.

Four models were dropped from the published run because the three pairs disagreed by more than 6 millimetres. That rule is the most useful thing on this page: if a model cannot be measured consistently three times, no row about it can be read quickly or slowly.

Note what the two numbers do not tell you. Insole length is taken with the foot out of the shoe, so it describes the shoe rather than the fit, and the widest point describes the shape rather than the volume around the toes. Both remain useful anyway, because both are reproducible, and a reproducible number beats a descriptive word the moment you have to explain a return to somebody who was not there.

Three questions a fast read always raises

The first question is whether the figure was measured or copied. Copied figures are not lies; they are somebody else measurement with the date rubbed off. The way to tell is to look for a repeat count, because a copy never carries one.

The second question is whether the row still points at the same item. Listings change variants without changing the link, and a row that was about a low-top can end up pointing at a high-top from the same store. Compare the model name in the row against the model name on the page, not the store name.

The third question is whether the size note applies to the size you are about to take. A note about a 42 says nothing about a 44 in the same model, because the measured difference between them is exactly what the row is silent on. If you buy a different labelled size from the one measured, treat the row as background rather than as evidence.

All three questions share one shape: they ask whether the evidence in the row applies to the thing in front of you. That is the only question a fast read is really answering, and it explains why the verdict at the end of ninety seconds is either keep reading or go and find out, and never buy or do not buy.

A useful side effect of asking the three in order is that each one has a cheap answer and an expensive answer. Checking a link costs a click. Checking a size costs a tape and a few minutes. Checking whether the row still applies to your model costs a decision rather than an observation, which is why it belongs last and why it is the question most often skipped.

1. Measured or copied: measured values carry a repeat count and a date, copied values carry neither.
2. Same item or same store: compare model names, and treat a store match as no match at all.
3. Your size or the measured size: a row measured at one labelled size does not transfer to the next one along.

Slowing down for the rows that deserve a second pass

Some rows should never be read in ninety seconds, and they are easy to spot once the fast pass is a habit. A row whose verdict says the model runs small is a row that has already cost somebody a return. A row measured two labelled sizes away from yours is a row you are extrapolating from. A row with no date is a row you are trusting on the strength of its handwriting.

The second pass is not a longer version of the first. It is a different action: you measure something yourself, or you send one question to the warehouse, or you find a second row for the same model and compare the two. Reading the same row again more slowly produces nothing except confidence.

The payoff of the fast pass is that it leaves you time to do that properly. Eleven rows read at ninety seconds each is fifteen minutes, and fifteen minutes is about what one careful measurement and one clear message cost.

One more trigger belongs on the list for a first order. If the row you are holding is the only row you have for that model, treat the ninety-second read as the start of the work rather than the end of it, and find a second source for the same two measurements before you choose a size. A single measured row is a data point rather than a pattern, which is precisely why models get retired here after three pairs disagree.

The second pass also has a stop condition, and it is worth setting before you start. Decide in advance which single action would settle the question, do that one, and close the row whether the answer pleased you or not. Rows that stay open for weeks are read ten times and settled never, and each of those readings costs the same ninety seconds.

1. Second pass trigger: the row verdict says the model runs small or runs large.
2. Second pass trigger: the measured labelled size is not the size you intend to take.
3. Second pass trigger: the row has no date, or the date predates the current batch note.
4. Second pass action: one measurement of your own, or one question to the warehouse, or one second row for the same model.

The data point behind this note

Own measurement run, KB-2026-03: 46 footwear rows published, three pairs measured per model at insole length and widest point; adjacent labelled sizes differ by 8 to 11 mm, and four models were retired because the three pairs disagreed by more than 6 mm.

◇KB-2026-03· 2026-09-30 Batch ledger · Challenge a figure

Notes in the same cluster

Related entries and gauges

• What does each column actually tell you?
• How do you find your item in a table with thousands of rows?
• Applicability protocol

  Work out which entries apply to your order right now.