News analytics: the tabs and what every number means

Analytics → News is the detailed analytics of a news site — the screen behind Full analytics in Newsroom. It appears for workspaces with a site of type News / Blog. For the live desk screen, see Newsroom: Live, alerts, front pages and the market.

At the top right of every tab you choose the site — All Sites or one of yours — and the period, from the last hour to the last year; 30 days by default. The choice is not kept in the page address: a reload brings you back to Overview and 30 days.

How reading is counted

• Opened — an article page was opened.
• Read — the reader stayed 25 seconds or scrolled through half of the article.
• Readers are browsers, not people: one person on a phone and a laptop counts twice.
• Bots are removed before anything is counted.

Overview

Counts articles only — not the front page, sections or galleries.

• Unique Readers counts people; the other cards count openings and reads. New Readers is all readers minus returning ones.
• Avg. Reading Time runs from opening to closing the article, including time in a hidden tab — so it is longer than the visible-screen attention in Newsroom. Both are right; they measure different things.
• Read rate in the card — the share of openings that became reads — and the last step of the funnel — the share of people who read at least one article — answer different questions; do not compare them.
• Reader Drop-off shows how far readers got in at least one article in the period.
• Further down: Top Articles (Last 24h) — always the last 24 hours, whatever the period — Category Performance, Engagement Rate Trend (the read rate by day), Peak Reading Hours, traffic sources, countries, on-site searches, and the period against the one before.

direct among the sources does not mean "loyal readers": moves between your own pages, links from messaging apps and most newsletters land there too.

Traffic

Counts every page opened — articles, front page, sections, galleries, search — so its numbers are larger than in Overview.

• The three comparison cards — Today vs Yesterday, This Week vs Last, This Month vs Last — ignore the period. Just after the hour, today looks up to an hour weaker; look again near the end of the hour before saying "we are down".
• Site Bounce is the share of visitors who opened exactly one page in the whole period; lower is better.
• The heat map of days and hours shows totals for the period, not averages.
• Below: devices, browsers and systems.

Reading Routes

How readers go from article to article, grouped by section — two articles by the same reader count as a step when there are at most 60 minutes between them. The period is at most 14 days. Reading depth counts a reader's articles over the whole period. The matrix shows the previous section in rows and the next in columns; the list below names the strongest bridges between different sections. A bridge shows that readers move, not how — a link, the back button or Google.

Article Intelligence

Tone, emotions, people, topics, format, readability and sources of your articles — everything except the word count is a language model's judgement, not a measurement. An article is analysed once two different readers opened it within 24 hours; galleries are skipped. The period here is when the analysis was made, usually minutes after publishing; the site choice has no effect. Names are not merged: "Kolev" and "Stefan Kolev" are two rows.

Audience DNA

Who your readers are. Most cards come from reader profiles — every reader who opened at least two articles in 30 days, recalculated every hour over their last 30 days — so they do not follow the period and site. Only Reader Loyalty, New vs Returning Readers, Readers by Country and Top Cities follow them.

• The reader segments — Casual, Regular, Highly Engaged, Power Reader — come from points over 30 days: a read article 2, an engagement 5, a comment 10, a share 8.
• Reader Loyalty counts the different days a reader opened articles in the period: Champion 7 days or more, Loyal 3 to 6, Returning under 3 days but at least 3 articles, New everyone else.
• Loyalty Mix (RFV) scores each reader 1–5 on how recently, on how many days and how many articles — the most useful level is At risk: readers who had a habit and stopped.
• The churn risk rises once a reader is late compared with their own habit; it means something for daily and weekly readers only.

A yellow line above the cards means the profiles have not been recalculated for over three hours; if it stays until the next day, tell us at [email protected].

Content Strategy

Which choices go with more reading — tone, section, length, hour. Everything here is a correlation, not a cause. Read rate is reads ÷ openings. Combinations need at least 10 articles; small ones are pulled toward the site's average so a handful of articles cannot top the list. Content Gaps compares readers' interest in a section with the number of its articles in the last 30 days. Best Publishing Times shows when people read, not how articles published at that hour did.

Real-Time

The site's pulse, refreshed every 15 seconds; the period does not apply here.

• Currently Reading counts people who opened an article in the last 10 minutes — not the exact number reading this second — for all your sites together.
• Trending Now ranks by acceleration: openings in the last hour against the article's average hour in the 23 hours before. Acceleration is not volume — check the openings before you promote an article from here.

If something doesn't work

News is not in the menu

No site of the workspace is of type News / Blog — set it in Settings → Tracking & SDK — or your access is limited to other CDP sections.

The numbers differ from Newsroom

Newsroom counts the last minutes with visible-screen attention; News counts the period with time from open to close. See Newsroom: Live, alerts, front pages and the market.

Article Intelligence lacks a new article

It is analysed after two different readers opened it, usually within minutes. Galleries are never analysed.