Build reports you can compare, find where the gap changed, and confirm the cause before you act.
Why the numbers differ
A Google Search Console click is counted by Google when someone clicks your result and leaves Google Search. A GA4 session is counted by your site when the Google tag loads, is allowed to run and records a visit. They are different events in different systems, so they do not match one to one.
- A click can produce no session: the visitor leaves before the tag loads, declines analytics consent, blocks scripts, or lands on a file such as a PDF that cannot run the tag.
- A click can produce more than one session: a visit that pauses for more than 30 minutes starts a new session, and GA4 can credit a later direct return to Google organic.
- GA4 can record Google visits that are not web clicks: Search Console reports Image, Video and News results as separate search types, and Discover in its own report. Visits from them can still appear in GA4 as google / organic.
So GA4 can be lower or higher than Search Console. The size of the gap tells you little on its own. What matters is whether the two reports are comparable, and how the gap moves.
Prepare comparable reports
Do this once, before you read any pattern. Many “gaps” are two reports answering different questions.
In Search Console
- Open Performance → Search results with the search type set to Web. If Image or Video results get clicks, compare them separately.
- Use a property that covers the same hostnames as your GA4 web stream. A Domain property covers every subdomain and protocol; a URL-prefix property covers one.
- Filter by page, country or device rather than by query when you compare totals: query filters leave out anonymised queries, so the filtered total is lower.
In GA4
- Open Reports → Acquisition → Traffic acquisition, or an Exploration. Both are session-scoped.
- Filter Session source =
googleand Session medium =organic(in one dimension: Session source / medium =google / organic). - Do not use the whole Organic Search channel: it also counts Bing, Yahoo, DuckDuckGo and other engines.
- Do not use First user source or medium (the User acquisition report): it credits the source of each user’s first visit, not of each session.
- For page work, add Landing page + query string and Hostname.
Google explains the scopes in User acquisition vs. Traffic acquisition.
Align the dates
Search Console labels daily data in Pacific Time. GA4 uses the time zone set on the property (Admin → Property details). For a property set to Nairobi time (UTC+3), one Search Console day runs from 11:00 to 10:59 the next morning in Nairobi while the US is on standard time (November to March), and from 10:00 to 09:59 during US daylight time. Single days will never line up.
- Compare whole weeks or 28-day periods, starting on the same weekday.
- End the range at least three days ago: Search Console data usually arrives after two to three days.
- Use the same dates and period lengths in both tools.
Align page identity
Search Console assigns a page’s clicks to the canonical URL Google selected (how clicks are counted). GA4 records the page where the tag fired, after any redirect. Match them before you compare pages:
| Difference | Search Console shows | GA4 shows | What to do |
|---|---|---|---|
| Parameters and variants | /product/phone-a/, the canonical | /product/phone-a/?attribute_pa_colour=blue | Compare on the path without parameters. URL Inspection shows the canonical Google chose. |
| Redirects | The old URL, until Google processes the redirect | The redirect target | Match each redirected URL to its target. |
| Hostnames | The full URL, including www or a subdomain | Only the path, unless you add Hostname | Add Hostname and check every host sends data to this property. |
| Trailing slashes and case | /phones/ | /phones or /Phones/ | Normalise paths before matching. |
Consent Mode and modelled data
If your site uses Google Consent Mode in its advanced setup, GA4 can add modelled users and sessions for visitors who declined analytics cookies. Google does this only when the property meets its thresholds, such as at least 1,000 events a day with analytics consent denied for seven days, and the reporting identity is set to Blended. Modelled numbers appear in standard reports, but not in every exploration or in the BigQuery export (behavioural modelling for consent mode). Check whether modelling is on before you compare periods: the gap can move when it starts, or when you switch reporting identity. Many smaller sites never qualify, and declined consent simply means fewer sessions.
Identify the pattern
Put weekly Search Console clicks and GA4 google / organic sessions side by side and work out sessions per click for each week. Then read the pattern with care. None of these patterns proves a cause on its own.
| Pattern | What it can mean | What it does not prove | Next |
|---|---|---|---|
| A steady gap | Collection has not changed. | That collection is complete: a template that never had the tag gives a steady gap too. | Run the page checks once, then use the ratio as your baseline. |
| A gap that widens | A measurement change, a shift in traffic mix (more mobile, image or Discover visits, countries with stricter consent), a shift in page mix, or attribution. | Which of these it is. | Find the pages and segments where it changed. |
| A sudden step | A dated change: tag, consent banner, redirect, data filter or site release. | That a change on the same day caused it. | Match the date with your releases and GA4’s property change history. |
| Clicks but no sessions on a page | A missing tag, URL mapping, consent, filters, property coverage or mismatched reporting periods. | That the tag is missing. | Work through the page checks below. |
| Clicks and sessions both steady | Nothing changed in total. | That search performance is fine: gains in one segment can hide losses in another. | Split clicks by query, page, country and device. |
| Clicks fall with sessions | A possible search loss. | Which pages or queries lost. | Compare pages and queries between the periods. |
A fall in Search Console clicks is the clearest sign of a search loss, but not the only one. Steady totals can hide losses in particular pages, queries, countries or devices, so split before you conclude. The worked example shows how.
Check the affected pages
Work through these in order. Stop when a check explains the change for the pages involved.
1. Find where the gap changed
- Where to look: Search Console Performance → Pages, comparing the two periods, and GA4 Traffic acquisition by Landing page + query string, filtered to google / organic. To group a Pages export by page type and site section, paste it into the Search Console Landing Page Analyzer. To rank landing pages by sessions, engagement and key events, paste a GA4 export into the GA4 Organic Landing Page Analyzer. They are separate tools: each reads one export, and you compare the two results yourself. The GA4 analyser keeps only Organic Search rows when the export has a channel column, so for Google alone, export from an Exploration filtered to google / organic.
- What to look for: Pages or sections where clicks held but sessions fell, or the reverse.
- What it justifies: Narrowing every later check to those pages. If the change is spread evenly across the site, suspect a sitewide change and go to check 4.
2. Match the URLs
- Where to look: URL Inspection in Search Console for the canonical Google selected; GA4 Landing page + query string and Hostname.
- What to look for: Parameter versions, redirected URLs, other hostnames and trailing-slash differences.
- What it justifies: If the “missing” sessions sit under another URL, it is a mapping difference, not a loss. Compare matched paths from then on. Change the site only if the redirect or canonical itself is wrong.
3. Confirm the tag fires
- Where to look: Open the page with Google Tag Assistant, or watch GA4 DebugView or Realtime while you visit it from a Google result. Test one page of each template: product, category, article.
- What to look for: A page_view with the right measurement ID and page location, sent once.
- What it justifies: A missing or duplicated tag on a template is a measurement fix. Fix the template and verify it; do not change SEO work because of it.
4. Check consent, filters and settings
- Where to look: Your consent banner (decline, then check Realtime); GA4 Admin → Data filters; Admin → Data streams → Configure tag settings; Admin → Property change history.
- What to look for: A consent change, a newly active data filter, a referral exclusion or a time-zone change dated near the change.
- What it justifies: A dated settings change explains a sitewide step. Note it on your reports. Data filters are permanent and do not repair past data (data filters).
5. Check property coverage
- Where to look: The Search Console property type, the GA4 data streams, and the GA4 Hostname dimension.
- What to look for: Hosts or subdomains that Search Console covers but that send data to no GA4 stream, or to a different property, and URLs that cannot run a tag, such as PDFs.
- What it justifies: Leaving those URLs out of the comparison, or adding the tag or stream if you need them measured.
6. Split the search side
- Where to look: Search Console Performance, filtered by query (branded and non-branded), page, country and device.
- What to look for: A segment that fell while another grew.
- What it justifies: Treating it as a possible search loss for those pages and queries, even when total clicks are flat. Then look at their rankings, indexing and snippets.
A worked example
Illustrative example, not real data. The numbers are made up to show the method.
An electronics store compares two 28-day periods that both end more than three days ago. Its GA4 property is on Nairobi time, and both reports are prepared as above.
| Period A | Period B | |
|---|---|---|
| Search Console clicks (Web) | 2,000 | 2,000 |
| GA4 google / organic sessions | 1,760 | 1,750 |
| Sessions per click | 0.88 | 0.88 |
Totals and the gap are both steady. A rule of “it is a search loss only when clicks fall” would stop here. Check 6 splits the clicks by query:
| Search Console clicks | Period A | Period B |
|---|---|---|
| Branded queries | 1,000 | 1,500 |
| Non-branded product queries | 1,000 | 500 |
| Total | 2,000 | 2,000 |
- Pattern: a brand campaign added 500 branded clicks, and non-branded product queries lost 500. The total hid both.
- Pages: the losses sit on product pages; the homepage gained the branded clicks.
- Measurement: on the product pages, sessions per click is unchanged, the tag fires and the URLs match. Checks 2 to 5 pass, so this is not a tracking problem.
- Next: treat it as a search loss on the product template and check its rankings and indexing, for example with how to check SEO after a WordPress update.
Verify the suspected cause
A cause is confirmed when evidence other than the date points to it, and fixing it moves the affected pages back. Test one cause at a time.
| Suspected cause | Confirm it by | After the fix, expect |
|---|---|---|
| Missing tag | Tag Assistant shows no page_view on the template. | The pages appear as google / organic landing pages within a day or two, and their sessions per click moves back towards your baseline. |
| URL mapping | The canonical Google selected, or the redirect target, matches the GA4 URL. | Matched paths line up. There is nothing to fix on the site. |
| Consent change | The step lines up with the banner change, and declining in a test removes the visit from Realtime. | A new, stable baseline. Annotate your reports. |
| Filter or setting | Property change history shows it near the step. | A change from that date on only; past data stays as it was. |
| Attribution | Traffic acquisition and User acquisition moved differently. | No fix. Report with session scope every time. |
| Search loss | The same pages or queries show fewer clicks, with fewer impressions or a lower position. | Clicks recover on those pages. Check with the same filters and period lengths. |
If GA4 sessions rise while Search Console clicks do not, automated visitors that load your tag may be the cause: see GA4 traffic went up but enquiries did not. Traffic Quality estimates how much of your traffic shows signs of automation, beside GA4, without changing your GA4 numbers.
Continue in Ndovu
- Search Console insights draws GA4’s google / organic sessions over the clicks chart, with organic search’s share of sessions and Google’s share of organic.
- Analytics matches Search Console clicks to GA4 google / organic landing pages by path in Search value. Its findings list pages that get Google clicks but no GA4 sessions, and pages GA4 records as one URL.
- Traffic Quality shows suspected automation beside GA4.
- Without an account, the Search Console Landing Page Analyzer and the GA4 Organic Landing Page Analyzer each read one export in your browser.
Google’s documentation
- Performance report: troubleshooting data discrepancies
- What are impressions, position and clicks?
- User acquisition report vs. Traffic acquisition report
- Behavioural modelling for consent mode
- Data filters