How Many Amsterdam Coffeeshop Menus Were Observed in October 2026?

Quick reality check: in the 30 days to 21 October 2026, 28 of the 251 shops listed here had a fresh menu in our archive. That’s 11% of this directory—not a claim about every menu in Amsterdam. Here’s where the rest of the snapshot went.

The October window in numbers

At the snapshot generated on 21 October 2026, the site lists 251 Amsterdam coffeeshops. Of those, 212 have at least one captured menu, meaning a menu that was found, read and stored. The 186 shops with a dated menu split into 28 fresh and 158 last known. Another 26 have captured menus without a usable date, while 39 have no captured menu. Across the same snapshot the archive holds 10,743 observed price lines, 10,690 of them trusted; the cheapest leaderboard draws on 4,061 entries. Fresh prices feed medians and rankings. Last-known prices stay visible with their dates, and differences against a current median are indicative rather than confirmed discounts.

Where the last-seen menus fall by month

A small number of archives reach back to 2004–2018, kept and labelled as historical. The fresh count of 28 is smaller than the October and September counts combined because the window is a rolling 30 days from the snapshot date, not a calendar month: only late-September menus land inside it. The monthly counts and the fresh count therefore measure different things, and both are printed here so neither can be read as the other.

Why 28 fresh out of 251 is the number worth publishing

Menu capture depends on menus being printed and legible, so partial coverage is the normal state of an archive like this one. The honest response is to print the coverage rather than hide it, and that is what every page here does. The medians on the price pages use fresh menus only. That is what keeps a €13 median for weed 1g honest: it is computed from 139 listings across the 27 shops that printed menus inside the window, and the page prints those counts next to the number. A statistic that claimed all 251 shops would be a fabrication. This one claims 27, in writing. The fresh share also moves: a window that slides one day forward drops the oldest menus and picks up the newest, so the count of 28 is a statement about this window, dated and re-readable next month. That is why the report prints the window’s dates rather than saying this month without qualification.

Why older menus still earn their place

A menu from July is not useless; it is a dated fact. 158 shops have last-known menus, and their pages show what the shop printed and when, with the difference to the current median marked as indicative. Read that way, an older menu answers the question it can answer: what did this shop last put on paper? The archives reaching back to 2004–2018 answer a different question still, about how menus used to read, and they stay labelled so they cannot pass for current. There is a practical use too. Before walking anywhere, a last-known page sets a rough expectation of what the shop printed last, and the date attached to every line says how much to discount it. The alternative, a blank page, would be less useful and less honest at the same time. Hiding all of that would shrink the archive to a tenth of itself while making the surviving tenth look more current than it is.

How the site keeps dates honest

Dates come from the menus themselves rather than crawl times or file names, and a daily rebuild does not make an old record newer. Fresh and last known describe different ages of evidence. As time passes, a dated menu eventually leaves the rolling fresh window even if its printed price is unchanged. The date-reading guide covers the rules, the last-menu-seen explanation covers the labels, and the method page describes capture. For a record without a usable date, the uncertainty remains visible rather than being replaced by the date the page was rebuilt.

The missing group between captured and dated

The October snapshot has 26 shops with captured menus but no usable dated menu: 212 captured minus 186 dated. The complete split of the 251 listed shops is therefore 28 fresh, 158 last known, 26 captured without a usable date and 39 with no captured menu. These figures derive from the site’s observed-menu snapshot generated on 21 October 2026, using the fresh window from 21 September to 21 October. Capturing a menu and establishing its date are separate steps.

That captured-but-not-dated group matters when reading the opening percentages. The share outside the fresh window is not entirely composed of older dated menus. It also includes records without the evidence needed to classify them by age, as well as shops without captured menus. Calling all of these shops last known would erase a useful distinction.

For the same reason, the count of trusted price lines is not a freshness count. Trusted describes whether a transcription passed the site’s checks. Fresh describes whether a dated menu falls inside the observation window. A clearly transcribed historical line may be useful archive material without belonging in the current price sample. Confidence in the reading and recency of the source are different dimensions.

Follow the denominator when reading a price claim

The catalogue total is the right denominator for a statement about directory coverage. It is not the right denominator for a category median. That median belongs to the listings in its own format, and its shop count can be smaller than the overall fresh-shop count.

For example, the weed single-gram sample described above comes from the shops contributing that format. The overall fresh count also covers menus that may contribute other categories or formats. You should not replace the row’s shop count with the catalogue count, or assume that every fresh shop supplied every row of a price table.

This distinction helps when discussing the report with another reader. Ask whether the number refers to shops, menus or price lines. A shop may contribute several offers, while an offer may have observations over time. The archive total describes the accumulated record; the fresh window describes the eligible recent evidence. Large archive totals cannot be used to make a small recent sample sound more representative.

A changing coverage count does not explain its own cause

If the fresh total changes in a later snapshot, first consider how the window works. Older menus leave the eligible period as time passes, and newly captured dated menus may enter it. A count can therefore change without telling you anything about a shop’s sales, popularity or opening status.

The same caution applies to a category median. A different set of contributing menus can change the middle observed price even when you do not have evidence that every shop changed its prices. To describe a trend confidently, you would need to inspect the underlying samples and matched observations. This coverage report does not perform that analysis.

When saving the report, keep its snapshot date and definition of fresh with the counts. Saying that the site has a particular fresh share without its period turns a dated result into an apparent permanent property. The date is part of the result, not a footnote that can be dropped for convenience.

Read coverage before choosing how much detail to trust

Coverage helps you calibrate a practical question. A fresh listing can provide a recent printed reference for a visit. A last-known listing can supply context and a question to ask at the counter. A shop with no captured menu offers no price evidence here, so the next useful step is finding its current menu rather than inferring a price from another shop.

None of these states is a recommendation for or against a business. They tell you how much the archive can contribute to your preparation. Keeping that modest purpose in view makes the report useful even when the fresh share is limited: you know where the evidence ends before treating a directory page as a current quote.

What a coverage number cannot tell you

The 11 percent fresh share describes recent evidence captured in this archive. It does not establish how many other shops printed new menus that were not captured, whether a shop is open or whether any line is in stock. Coverage is a property of the archive, not a review of the shops. The site is 18+ info only, no sales; verify on site for today’s offer. Inspect the shop directory for individual source dates.

Read the four groups as different evidence states

Four groups, four different states of evidence: fresh dated menus, older dated menus, captured menus with no usable date, and shops we haven’t captured at all. They’re not interchangeable. An undated capture proves an image exists - not how old it is. A shop with no image is missing from our sample; that’s not proof it has no menu. Every count here describes what the archive held on the snapshot date, and it shifts the moment another menu is collected or a date is corrected. If you cite the 11 percent, carry the snapshot and the 30-day window with it, or it starts sounding like a live city-wide measurement - which it isn’t. The listing scope guide unpacks those records.