Review article analytics
Article analytics covers one article at a time: how it performed, and what to improve next. It is available on eligible plans and development stores.
Open analytics
Analytics is not in the app menu. Open it by selecting View analytics at the end of an article row in the article list, or by selecting View analytics at the top of the editing screen for a saved article.
Searching, filtering, and sorting articles belong to the article list, not to analytics. See Find and open articles.
A left-pointing arrow at the top left returns you to the article list. It shows the arrow only, with no label next to it. View article (visible articles only), Refresh profile, and Edit article are at the top right.
The period covered
Below the title, the screen states which finalized period the analysis covers: the 30 days ending yesterday. Today is not included.
A newly published article shows Preparing analysis data until enough measurements exist.
Article purpose
Analytics weights different metrics depending on what the article is for. Four purposes are available:
| Purpose | Typical article |
|---|---|
| SEO and acquisition | Brings new readers in from search |
| Education and editorial | Explains how to use something, or shares knowledge |
| Product guide or comparison | Helps a reader choose a product |
| Campaign | A limited-time promotion or sale |
The purpose is labelled Selected by AI or Set manually. AI picks one from the article content on the first analysis. If it does not match, select Edit, choose the right purpose, and select Change and reanalyze. Changing the purpose changes the metric weighting and recalculates the article's results.
Goal achievement
A single score out of 100 that weights the metrics relevant to the selected purpose. A badge sits next to it:
| Badge | Meaning |
|---|---|
| Strong | 70 or above |
| On track | 50 or above |
| Improvement opportunity | Below 50 |
| Collecting data | Not enough data to judge yet |
Below the score, What to improve now shows the single highest-priority recommendation.
The four diagnostic scores
Each score rates one aspect of the article out of 100:
| Score | What it looks at |
|---|---|
| Reach | How many people read the article |
| Reading | Whether readers stayed and read properly instead of leaving quickly |
| Action | Whether product links, related articles, and buttons were selected |
| Conversion | Whether the article led to cart adds and purchases |
The colours follow the legend: green is good, yellow is worth watching, red needs improvement.
Each score is calculated by comparing this article against your other articles with the same purpose. When too few comparable articles exist, or when the article has too few measured views, the score shows Collecting data instead of a number. Check again after more data accumulates.
Article analytics has no SEO score. Use the four diagnostic scores and goal achievement instead. For search impressions, clicks, and position, use Google Search Console.
Recommendations
Up to three prioritized recommendations are listed, based on the article purpose and the measured results.
Each one carries a confidence level (High, Medium, or Low), the recommendation itself, the evidence behind it, and the related score. Low confidence usually means there is not much data yet.
Make one change rather than several, then compare the next period.
Detailed metrics
This section lists the measured figures. The confidence of the period's data (High, Medium, or Low) appears next to the heading; fewer views means lower confidence.
Key figures
| Metric | Meaning |
|---|---|
| PV | Times the article was displayed |
| Measured views | Views for which reading data was received. The rates below use this as their denominator |
| Measurement coverage | Of the views that finished, the share that returned reading data |
| Avg. active reading time | Average time readers spent actively viewing the page |
| Qualified read rate | Share of views that reached 90% of the article and stayed long enough to read it |
| Short exit rate | Share of views that left within 10 seconds with almost no scrolling or clicks |
| Return rate | Share of readers who came back to the article |
| Direct purchases | Purchases attributed directly to the article |
PV trend
Daily page views across the analysis period, with the period total.
Scroll reach
How many views reached the 25%, 50%, 75%, and 90% points of the article. A sharp drop between two points suggests that part of the article is hard to read.
Traffic sources
Where readers came from: Search, Direct, Social, Referral, or Internal. Sources with no views are not listed.
Actions and purchases
| Metric | Meaning |
|---|---|
| Product link CTR | Product links selected, as a count and a rate |
| Related article CTR | Related-article links selected, as a count and a rate |
| CTA CTR | Buttons and calls to action selected, as a count and a rate |
| Direct cart adds | Cart adds attributed to the article, as a count and a rate |
| Direct revenue | Revenue attributed directly to the article |
| Assisted revenue | Revenue the article contributed to indirectly |
| Assisted purchases | Purchases the article contributed to indirectly |
| Direct purchases | Purchases attributed directly to the article, as a count and a rate |
Content features
A count of what the article itself contains: word count, heading count (H2 and H3), images, internal links, product links, and CTAs. Recommendations draw on these figures.
Deciding what to change
Change one thing at a time.
Reach is low
There may not be enough ways into the article. Check that the words readers would search for appear in the text, and link to it from your other articles.
Reading is low
Shorten the opening and state what the article covers up front. Break up long paragraphs, and add headings and images. Use scroll reach to find where readers drop off.
Action is low
Reconsider where product links and related articles sit. Placing them next to the relevant passage works better than grouping them at the end.
Conversion is low
Feature products that genuinely match the article, and explain who they suit and when to use them. Check that price, shipping, and other pre-purchase details are not missing.
There is not enough data
Do not judge an article right after publishing. Wait until Collecting data disappears before comparing.
How data updates
Refreshing the article's title and author is a separate process from aggregating page views. Numbers that do not change immediately after you edit an article are not necessarily wrong.
Refresh profile re-reads the article content and recalculates the 30 days ending yesterday. A notice appears while it runs, and the results are replaced when it finishes.
Low page views alone do not make an article bad. Consider how long ago it was published, the season, product availability, and whether it was promoted through ads or social media.
If values are missing, see Analytics data is not displayed.