How to Read the Google Discover Performance Report

The Discover performance report in Search Console shows impressions, clicks and average CTR for content that Google surfaced in Discover. It behaves differently from the regular Search report: it may not appear at all, its totals can disagree with its table, and its lines swing more than most publishers expect. Each of those quirks has an explanation in Google's help pages. Knowing them is the difference between reacting to noise and spotting a real problem.

This guide is for publishers and SEO owners who open this report weekly. If your question is about whether a sponsored article suits Discover at all, that is covered in a separate piece on sponsored content expectations in Discover.

Why the report may be missing

An absent report feels like a configuration error. In most cases it is a threshold.

Why the minimum impressions bar hides the report

Google's Discover performance report help page (Opens in a new tab) says the report appears only when your property has reached a minimum number of impressions in Discover. The page does not say what that number is. A missing report therefore does not mean your content is weak. It means Discover has not yet shown the property often enough. Rows in the table also need a minimum number of impressions, but rows under that threshold still count toward the site totals.

That second rule explains a common puzzle. The chart can show a total that the listed pages do not add up to, because some pages are in the total without being listed individually.

Is there a shortcut past the bar?

Not by any documented switch. The help page offers no method, and the sensible reading is that you meet the Discover content and image guidance consistently and let the system decide. That guidance includes headlines that reflect the content, large images of at least 1200 px wide and a good page experience. A routine that applies those rules to every new page beats any one time attempt to force the report to open.

Wrong property, empty screen

Another possible cause is looking in the wrong property. If you verified both a domain property and a URL prefix property, data appears only in the property that contains the canonical URL. One property can look empty while the other holds everything. Confirm which properties you own and where your canonical URLs point before concluding that nothing is there.

Symptom and cause lookup

While the report is absent, check what you can control. The Discover documentation (Opens in a new tab) says content is automatically eligible if it is indexed and meets Discover's content policies, that no special tags are needed, and that eligibility does not guarantee appearing. Policy violations can trigger a Discover manual action, which shows in Search Console under Security and Manual actions. If your interface is not in English, menu names may look different from the ones used here.

How impressions, clicks and CTR are counted

Before comparing numbers, read the definitions. They are not the same as in the Search report.

How an impression is counted

According to the help page, an impression is counted when a Discover item is scrolled into view, either as a standard item or inside a carousel. Only one impression is counted per result per session. Two practical effects follow. An item that never scrolls into view is not counted, and repeated exposure of the same item inside one session adds nothing extra.

What counts as a click

A click is the user clicking the Discover item. Sharing and other actions do not count as clicks. Average CTR is clicks divided by impressions. For Web Stories there is a special rule: an impression or click also counts for each additional story a user opens from the initial one.

Do not compare Discover CTR with Search CTR. A Search result answers a typed query, whereas a Discover item is proactively recommended to someone who typed nothing. Compare Discover figures only with your own earlier Discover figures.

Definitions in one table

CTR up or down: three readings

• Impressions up, CTR down: the item may be reaching a wider and less specific audience. Check total clicks before calling it a problem.
• Impressions flat, CTR down: headline and image may be pulling the same audience less strongly. This is a good case for a presentation test.
• Impressions down, CTR up: Discover may be showing the item to a narrower audience. If clicks hold steady, nothing is wrong.

These are hypotheses, not rules. Google's documentation does not map CTR movements to specific causes.

Filter options and the Pacific Time quirk

The report can be filtered by date, page, country and Discover appearance type, and the date filter cannot be removed. Appearance types are AMP articles, News Showcase, Web Story and Videos. Regular expressions in the page filter use RE2 syntax and ignore case by default; prefix (?-i) for a case sensitive match. Invalid syntax returns no matches.

Granularity can be hourly (24 hour view only), daily, weekly or monthly. The 24 hour view uses local time, while the other views use Pacific Time. If you compare a daily bar against your own server logs, expect the day boundaries to differ.

Using the country and format tabs

The country grouping reports where the content was viewed or clicked, and several countries can be combined in one filter. The help page does not list supported countries, so the documentation gives no reason if a country you expected is absent. The appearance tab separates formats. A site that publishes only articles will see little there, but if you also run stories or video, this is where a shift in format mix becomes visible.

Stories and carousels count differently

Because Web Story activity counts for every additional story opened after the first, a site that publishes stories can see impressions and clicks that grow faster than its article numbers. Do not blend the two without noticing. Filter by appearance type before comparing formats, otherwise a strong story week can mask a weak article week, or the reverse. The same caution applies to carousel placements, where an item is counted when it scrolls into view inside the carousel.

Canonical attribution and duplicate URLs

The mistake that most often misleads readers of this report is expecting data on the address a visitor landed on.

How page metrics are assigned

Page metrics are attributed to the canonical URL, not the landing page, and data appears only in the property that includes the canonical URL. Duplicate URLs show zero clicks, impressions and CTR. If you search the table for a tracking-tagged or alternate address, you will see zeros even though the article is performing well under its canonical address.

The reasoning behind canonical grouping

Search systems group duplicate pages under one canonical address so that signals are not split. The Discover report follows that grouping and reports activity on the representative address. The help page states the rule but does not discuss how Google picks the canonical URL, so for that question the documentation here is silent. Your job is to make the intended canonical obvious and consistent, and to read the report accordingly.

After a restructure or move

During a migration or a restructure, the report follows the canonical URLs, so old addresses fade to zero and new ones grow. Judge the transition by the site total instead of single pages. Check redirects and canonical tags with the URL Inspection tool, and do not read zeros on retired addresses as a loss. The details of migrations themselves fall outside this guide.

Case: a match report on a local sports club site

A local sports club runs a news section. Its match report opens at the main address, at a shortened sharing link with a campaign parameter, and under an old archive path left from an earlier site design. The canonical tag points to the main address. In the report only that address carries numbers. A volunteer who searches for the archive path and sees zeros may decide Discover never showed the report to anyone, while a colleague who knows the attribution rule finds every impression and click under the canonical address.

Address checks before you panic

• Does the address you are searching for match the canonical tag on the page?
• Do http, https, www and trailing slash variants resolve to one address?
• Do parameterised URLs point to a clean canonical?
• Are you looking in the verified property that contains the canonical URL?

The same logic applies to the standard Search report, where page and query views have their own aggregation rules. For a walkthrough of that report, see the guide to the Search Console performance report.

When the graph and the grid disagree

The help page lists several reasons. The table shows at most 1,000 rows, and that cut happens before the data reaches the report. Filters can cause discrepancies, and totals from contains and does not contain filters may not add up to the unfiltered total. Exported values shown as "~" or "-" download as zeros, so a zero in a spreadsheet may not be a true zero. Before comparing totals, make sure the filters match and the table is not truncated.

Telling volatility from a real drop

A fall in Discover traffic is alarming at first sight, yet Google describes this surface as unpredictable from the start.

Google on unpredictable traffic

The Discover documentation says traffic from Discover is less predictable than traffic from keyword driven Search and recommends treating it as supplemental. It lists changing user interests, shifts in content types and Search updates among the reasons for traffic changes, and it says some changes are unrelated to content quality or publishing frequency. The help page repeats that predictability is lower because Google serves the content proactively.

Preliminary data and time zone traps

Three data effects can imitate a drop. Today's data, and sometimes yesterday's, is preliminary and shown with a dotted line. Data is normally available within two to three days. And the 24 hour view uses local time while other views use Pacific Time. Reading the last few days as a decline is therefore often premature.

Seven checks in order

1. Switch to weekly granularity and set aside the preliminary segment. Google's Search performance report help page (Opens in a new tab) notes that weekly and monthly views smooth daily swings.
2. Check whether the drop is site wide or concentrated in specific pages or countries.
3. Split by appearance type: AMP articles, News Showcase, Web Story, Videos.
4. Look for canonical, redirect or template changes made around that date.
5. Check manual actions and index status.
6. Check whether a Search update overlapped; the guide on diagnosing traffic loss after a Google update offers a sequence.
7. Finally, review your own publishing log for changes in topic, image or headline style.

When to call a decline real

Treat a decline as real when it is not preliminary data, persists across several weekly readings, and is broad instead of confined to one page group, country or format. Treat it as a swing when it lasts one week, ties to a single topic and leaves other pages stable; log it and look again next week. Google's documents do not give a numeric threshold, and every site has a different baseline, so the rule uses none.

Case: a hardware retailer's blog loses two weeks of clicks

A hardware retailer's blog sees Discover clicks fall for two consecutive weeks. The first reaction is to rebuild the article layout. The analyst works through the checks instead. The decline is not preliminary data. It sits mostly in posts about power tools, the format mix is unchanged, and nobody touched a canonical tag or a redirect. The best hypothesis becomes a shift in interest or content type, so the team tests new headlines and lead images on the power tool posts and compares results two weeks later. The case proves nothing about cause. It only shows how an ordered check narrows the questions.

Mistakes to avoid after a dip

The costliest mistake is changing headlines, images, templates and posting schedule at once on the strength of one week of data. You will not know which change helped or hurt. Change one thing, record the date, wait at least two weeks, and use a second source, for instance organic traffic analysis in GA4, to cross check what the Search Console numbers suggest.

Editorial questions when numbers fall

• Do headlines still deliver what the page promises?
• Do lead images meet the 1200 px width and carry the large preview permission?
• Have recent posts drifted toward types Discover may not recommend, such as forms, petitions or unframed satire?
• Has page experience degraded through slow loads, layout shifts or persistent pop-ups?

A weekly review table

A short, repeatable routine keeps panic out of the process. The layout below takes 20 to 30 minutes.

Lock the report view

Open the report with the same settings every time: weekly granularity, the default three month range and all countries. Comparison mode allows only one comparison at a time. If you change filters, write it down, because filters can create discrepancies in totals.

The weekly checklist table

Misreadings and corrections

Small dated tests

A report is most useful when you compare like with like. Pick one variable per test: a new headline style on one content category, or a new lead image style on another. Keep the other variables stable for at least two weeks, use the page filter or a regular expression on a URL pattern to isolate the group, and compare the same weekday range before and after. Record the result in the log even when it is inconclusive. A series of small, dated tests gives you more usable knowledge than a single dramatic redesign, and it keeps you aligned with Google's advice to treat Discover as a supplemental source of traffic.

A publishing log beside the report

The report does not know what you published. A simple log of publication date, headline, lead image size and crop, topic category and any template or technical change makes patterns visible. Often a drop lines up with a log entry; just as often nothing lines up, which supports Google's description of shifting interest. The same log shows which pages perform well, so you can reuse their traits in experiments without treating them as ranking rules.

Briefing editors and managers

Put the date range, filters and granularity on every screenshot. State the definition of an impression once, so nobody mistakes it for a pageview. When presenting a decline, show the triage steps you completed, so the conversation is about which possibilities were eliminated rather than who erred. A summary sentence can look like this: clicks changed in this direction over four weeks, the data is final, the change sits in this page group, next we test this.

What the report cannot tell you

The report covers content from the last 16 months. The table shows at most 1,000 rows, and data normally arrives within two to three days. These limits can change, so check the current help page. The report also stays silent on why a page was not shown and on the cause of a decline. You close that gap with your own log, a second measurement source and patience.

Related guides: Video and Discover Visibility for Publishers, Large Images for Google Discover, Video Sitemaps and VideoObject Markup Guide.

Conclusion

Reading the Discover report well rests on three habits: remember that a minimum threshold controls whether it appears, count impressions and clicks the way Google defines them, and look for data under the canonical URL. For drops, filter out preliminary data, move to a weekly view and follow a fixed order of checks. Because Google itself calls Discover traffic less predictable than Search traffic, judge the trend over weeks and change one variable at a time.

If your plan includes placing editorial content on relevant sites, you can look at our guest post service.

Frequently Asked Questions

Why is the Discover report missing in Search Console?

Google's help page says the report appears when your property reaches a minimum number of Discover impressions. The number is not published. A missing report does not mean the content is poor; the threshold may simply not have been reached.

How are Discover impressions counted?

An impression is counted when a Discover item is scrolled into view, either as a standard item or in a carousel. Only one impression is counted per result per session.

Why does my page show zero in the Discover report?

Metrics are attributed to the canonical URL, and data appears only in the property that contains it. Duplicate URLs show zero clicks, impressions and CTR.

My Discover traffic dropped. Is it a real decline?

First set aside preliminary data, which shows as a dotted line, and switch to a weekly view. Google says Discover traffic is less predictable and that some changes are unrelated to content quality. Investigate declines that persist across weeks and affect the site broadly.

Why do the chart and the table show different totals?

The help page names the 1,000 row table limit, filter effects and rows below the threshold that count toward totals without being listed.

How soon does Discover data appear?

Data is normally available within two to three days, and today's data and sometimes yesterday's is preliminary. These timings can change, so check the current help page.