Daily Noodle Admin

Overview

Readers per day

Unique readers each day, split by the section they used. A reader who used two sections counts in both lines.

Signals

Automatic flags from the rules in Basis of preparation. Small numbers are left out.

News

Which articles get read, and read properly.

Readers by topic

Unique readers of each topic's articles in the period.

Read-through vs attention

Each dot is an article: time spent against the share who reach 75% of the page. Top right is best.

Puzzles

Which types get played and finished, and how hard they are.

From opening to solving, by type

Of the puzzles opened: how many were started, and how they ended.

Puzzle types compared

Rates are shares of puzzles started.

Games

Rounds played, results and how far people get.

Audience

Who comes, from where, and whether they come back.

Which sections people use

Each reader counted once, by the combination of sections they used in the period.

Do new readers come back?

Readers grouped by the week of their first visit. "Came back" means seen again on a later day, at any time.

Where readers arrive from

The site that linked to the first page of a visit. Apps and many social sites send no referrer and show as direct.

Devices

By screen width when the page opened: phone under 480px, tablet under 1024px, desktop above.

Front pages

What gets tapped on the News, Puzzles and Games lists.

Taps by position in the list

How strongly position drives taps. Position 11 includes everything further down.

Most tapped cards

Average position shows whether a card was found high up or scrolled to.

Basis of preparation

How every figure on this page is produced, and what it can't tell you.

General basis

Source. Events recorded by dn-track.js on every Digi page and stored in the digi Supabase database (London). Figures are computed live when this page loads or you press Refresh.
Reader. One browser on one device, identified by a random code stored in that browser. The same person on a phone and a laptop counts as two readers; clearing browser data or a private window starts a new reader. No names, emails or IP addresses are held.
Visit. A run of activity by one reader with no gap longer than 30 minutes.
Page load. One opening of one page. Reloading, or returning via the back button after the page was discarded, counts again.
Dates. UK dates (Europe/London). A range includes both its start and end date.
Not counted. Readers who switched recording off, browsers sending Global Privacy Control or Do Not Track, readers whose browser blocked the script, visits to this admin page, and any browser on the exclusion list (0 at present, e.g. your own test phone; add the device you're on with "Exclude this device" at the top).
Losses. Events are sent in small batches. Some are lost when a phone closes a page abruptly or goes offline, most often the "left the page" event that carries time on page. Timings are averaged over page loads where that event arrived; loads without it are left out of those averages rather than counted as zero.
Bots. Automated traffic is not filtered. Most crawlers don't run the script, so they are largely absent, but treat sudden spikes with care.
Small numbers. Percentages from a handful of readers swing wildly. Signals ignore anything with fewer than 5 readers or starts.
Retention. Raw events are kept for 13 months, then reduced to daily totals; ranges older than that will read as empty here.

Definitions

Every measure used above.

Sign in