How do I read the site statistics?

The top of the page says how many translations, languages and chapters the site has. Below are two tabs with figures.

1. Open the “Last 30 days” tab.
2. Point at a slice of a pie chart or press it: a tooltip shows the number and the share.
3. Under the list of translations, press “show all” to expand it fully.
4. Press the name of a translation: its own statistics open.

Tooltips are not only on the pie charts. They also appear on the charts by day and by hour, on the countries on the map, and on the icons and cells of the language table.

“Yesterday” is the past day; “Last 30 days” is the thirty days that ended yesterday. The dates of the period are written under the tabs. If there is no data for some of the days, there is also a note like “data for 22 of 30 days”.

Days are counted in Coordinated Universal Time (UTC). There are no figures for today: a day gets into the statistics at night, after it has ended.

This block is only on the “Last 30 days” tab. It has three tracks, one under another: “Robots: requests”, “People: page loads” and “People: unique”. Each track has its own scale, so the heights of the lines are not compared with each other. What is worth watching is how one track changes from day to day. Point at a day to see its figures.

Readers is the main line of the page. The exact number of people cannot be known, so it shows a range, “… to …”. The lower bound is the visitors we count as people. The upper bound is the same plus the doubtful ones (“people?”) and those who asked not to be counted. The real number is somewhere in between.

A visitor is a tag that lasts only one day, so over 30 days visitors are added up day by day. A person who came every day is counted thirty times in this sum.

Pages served is all the pages the server gave to anyone at all, robots included. That is why this number is much larger than the page views.

Reading time is how long a visitor reads per day on average, and how much time everyone read together. Next to it is the date from which it is counted.

These charts include everyone who said they were a browser. Ordinary analytics services would show roughly these figures. The first chart counts visitors, the second counts loaded pages. Each chart has its own 100%; they do not add up with each other.

• people — real readers;
• people? — an estimate: these visitors look no different from people, but their behaviour shows an anomaly, so we estimate how many of them are probably not people;
• scripts — programs that present themselves as a browser but, by a number of signs, are unlikely to be controlled by a person;
• Asked not to be counted — a rounded number of browsers that asked not to be tracked.

Translations, languages, books and chapters, time of day and countries count only people together with “people?”.

The “Translation language” chart shows which language the Bible text is read in, and the “Interface language” chart shows which language the site itself is open in.

Further down the page, the “Which language people read in” table connects the two. Its rows are the interface language, its columns the translation language, and a cell holds visitors. Blue cells mean people read in the interface language, red ones in another language. An empty cell means there are too few visitors. The totals add up only the cells shown.

Each line is a translation. The length of the bar shows visitors, and its colour matches the colour of the language in the chart above. On the right are the person icon with the number of visitors, the eye icon with the number of page views, and the share in per cent. Hovering over an icon tells you what it means, and hovering over the name shows the full title of the translation.

The most-read translations come first; the “show all” button expands the rest. Translations read by very few people are gathered in an “others” line.

A name in the list leads to the statistics of that translation; its address looks like /stat/<translation>/. It has the same two tabs, the “Readers” line, books with bars, chapters and countries, and the “Read this translation” link opens the text itself. If there were few readers in the period, there are no figures: we do not show small numbers, so that nobody can be identified.

Some rights holders with several translations have a summary page. Its figures are the sum over the holder’s translations, so a reader of two translations is counted in each of them. The site has no links to such pages; we give the address to the rights holder.

These are the most-read books and chapters of the period, with the number of visitors and page views. All translations are counted together. The “others” line at the bottom gathers everything that did not make the list and shows the total.

The summary line shows how long people listened in the period and how many listeners, plays and finished chapters there were. The median session means that half of the listening sessions are shorter than this time and half are longer. Next to it is the date from which it is counted.

Below come the “Recordings”, “Books listened to” and “Chapters listened to” tables.

The “Devices” charts divide readers into “Phone” and “Desktop”. “Desktop?” is an estimate: like “people?”, these are visitors with an anomaly in their behaviour. “Asked not to be counted” are those who asked not to be tracked; the numbers in this slice are rounded.

The “When people read” block has 24 bars, one per hour. The hour here is local, each reader’s own, so morning in this block is the reader’s own morning. An hour with too few readers stays empty and is labelled “below the threshold”.

The table shows visitors and page views by country. On the map, the colour of a country shows its share, and the values of the colours are given in the legend under the map. Point at a country to see its figures.

The “unknown” line gathers visitors whose country could not be determined. It and the “others” line are not on the map.

Robots name themselves honestly: search engines, programs of companies that train AI, link previews in messengers. They are not counted as readers, and for them we count requests, not visitors.

• The “Robots” chart divides requests by type, and the “AI crawlers” chart shows whose artificial intelligence robots they are.
• The “Which robots came” list names them one by one.
• Pretended to be a robot — called themselves a known robot but did not come from its network. The “Scripts posing as robots” list shows which names are used this way.
• Pretended to be a browser — called themselves an ordinary browser, but a real browser does not behave like that.
• Posed as visitors — downloaded pages like an ordinary browser, but the page view was never confirmed.
• Looked for vulnerabilities — those who tried addresses in search of the site’s weak spots and fell into our traps. This line appears only if there were any.

This is the last block of the tab, and it is about the work of the server, not about readers. It counts all the server’s responses to everyone who contacted it.

• Responses — how many responses the server gave in the period.
• Average and Median — the usual load: responses per second (RPS) and per minute (RPM).
• The RPS and RPM lines show the peak and the minimum with the time they happened.
• Responses by hour, UTC shows how the load changes over the day. The hour here is in UTC, not local as in “When people read”.
• Responses by status class divides them into successful ones, redirects, “not found”, blocked, rate-limited and errors.

The total of visitors includes scripts and those who asked not to be counted, while countries count only people together with “people?”. So the countries add up to less. On the map the sum is smaller still: it has no “unknown” and “others” lines.

The same is true of books and translations: they also count only people, and their sum will not match the “Visitors” chart. This is not an error.

We do not show small numbers, so that a rare value cannot identify an individual reader. Lines with too few visitors are gathered into one “others” line with the total. Point at it: if enough values have gathered, the tooltip shows what went into it.

The “below the threshold” note on a bar means that there were too few readers in that hour. The “nothing above the threshold” label means the same for a whole block: there is nothing to show.

Turn on the request not to be tracked in your browser: Do Not Track or Global Privacy Control. Where to find them is explained on the “Privacy” page.

With this setting we store neither a tag nor what you read. We count only how many pages readers with this setting opened and roughly how many of them there were. On the statistics page this is the “Asked not to be counted” slice; its numbers are rounded.