Most businesses open Search Console exactly twice: once when the website goes live, and once when the enquiries dry up. In between, a report keeps running that answers the only question that matters: what are people in your region searching for, and does your page appear in the results? This article takes apart four numbers, three filters and two reports – enough to turn a wall of figures into a list of tasks.
Key takeaways
- Search Console measures the search that happens before your website, not the behaviour on it. It is the stage before a visitor counter, not a replacement for one.
- 62.1% (SparkToro) of German Google searches end without a click. Impressions without clicks are the normal case, not a defect on your page.
- The four headline metrics cannot be read one at a time. Only the pair of position and click-through rate tells you whether a page needs better copy or more visibility.
- Without filters the report is an average of job ads, brand searches and stray clicks. With three filters it becomes a list of concrete tasks.
- Data reaches back 16 months (Google) and appears with roughly 48 hours (Google) of delay. Both decide when you are even allowed to compare.
What the performance report actually shows
Search Console does not measure your website. It measures the search in front of it. Every row in the performance report describes a moment when someone typed something into the search box and your page appeared in the result – with or without a click. A visitor counter on your own site only starts after that, at the first page load. Anyone who holds one figure against the other and finds a discrepancy has not found a bug; they have asked two different questions.
The demand behind that is well documented. 80% (Statistisches Bundesamt) of people aged 16 to 74 in Germany recently searched online for information about goods and services, and 92% (Bayerisches Landesamt für Statistik) of Bavarian companies with at least ten employed persons had their own website in 2025 – a state figure rather than a national one, but a solid indication of the order of magnitude. The question is no longer whether a website exists, but what it brings in.
Search in Germany is also almost synonymous with a single channel: 88.49% (StatCounter) of measured queries went to Google in August 2026. What sits in Search Console is therefore not a slice of the search market but very nearly all of it, and what is missing there is usually missing from your enquiries too. Once you have sorted out the basics of search optimisation, the report is where you check the result.
Before you judge a single figure, two timings matter. 48 hours (Google) pass before collected data appears in the interface at all, and Google names 2–3 days (Google) as the usual period until it is fully available. The last points of any curve are therefore systematically too low – counting is still under way. Judging Friday on a Monday morning measures the lag, not reality.
The most common false alarm
The four numbers at the top of the report
Four tiles sit above the chart: clicks, impressions, average click-through rate, average position. They look like four equivalent metrics and they are not. Two are counts, two are averages – and averages mislead as soon as they are formed across unequal things.
An impression occurs when your page appears in a result the searcher sees; a click is the visit that follows. The click-through rate is the quotient of the two, and position is averaged across all queries. A page ranking third for one important term and sixtieth for twenty unimportant ones therefore shows an average position in the double digits – and looks worse than it is.
For that reason the four numbers are read in pairs. Position and click-through rate together tell you whether copy needs sharpening or whether visibility is missing; impressions and clicks together tell you whether a page is growing or whether demand is fluctuating. The table below sorts out what each number answers and where it stays silent.
| Metric | Answers | Does not answer |
|---|---|---|
| Impressions | How often Google considered your page relevant at all | Whether anyone saw or read it |
| Clicks | How often visibility turned into a page visit | Whether an enquiry or an order followed |
| Click-through rate | Whether title and description match the search intent | Whether the landing page delivers what the result promised |
| Position | Where you stood on average across all queries | Where you stand for the one query that counts |
A second point affects every year-on-year analysis: performance data reaches back 16 months (Google), after which the oldest days drop out. The remedy is unspectacular: export once a quarter and file the spreadsheet. Five minutes that cannot be made up a year later.
Why many impressions bring no clicks
The most common observation in small accounts: impressions rise, clicks stay flat. It feels like a defect and usually is not one. In Germany, 62.1% (SparkToro) of all Google searches end without any click, measured against a clickstream panel for January to April 2026, and only 37.9% (SparkToro) lead to a click at all – somewhere, not necessarily onto an external website. Anyone expecting every impression to be a chance of a visit is working with a base population that does not exist.
A third figure makes the arithmetic tangible: out of a thousand German Google searches, 287 clicks per 1,000 searches (SparkToro) land in the open web – the ceiling of what can be collected at all, across every website. Germany does well in the country comparison: German searchers click at rates more than 20% higher (SparkToro) than British ones, so click-through benchmarks from English-language studies do not transfer one-to-one.
How much the environment is shifting becomes clear across the Atlantic. In the United States, 68.01% (SparkToro) of Google searches ended without a click in early 2026 against 60.45% (SparkToro) in 2024 – from two panels that are not identical, as the author points out. Searches with at least one click fell over the same period by -9.51 points (-22.9%) (SparkToro). If your clicks soften while impressions hold steady, the explanation may lie outside your website.
The one question that rescues the report
On top of that comes a shift in search behaviour now documented for Germany too. 50% (Bitkom) of internet users in the country at least sometimes use a chat instead of classic search, while 47% (Bitkom) still search exclusively the classic way. The performance report therefore still reflects the largest part of demand – just no longer all of it.
Something is shifting inside the results page as well. 24% (Bitkom) frequently read only the AI summary placed above the results without looking at them at all – an additional, nameable reason for impressions without clicks. At the same time 43% (Bitkom) still click through to the results despite the summary, to form their own picture. So the click column does not lose its meaning, it loses volume. What that means for the visibility of a local business is something we took apart in a separate article on AI summaries in search.
Only the filters turn the report into a task
Without filters the report is an average across everything: brand searches, job ads, stray hits from the legal notice, and somewhere in there the two dozen queries your business depends on. Three settings separate signal from noise, and all three sit in the same filter row.
The first filter is the device. 50.89% (StatCounter) of page views in Germany came from a phone in August 2026. Positions and click-through rates differ noticeably between the two views, because fewer results are visible without scrolling on a small screen. A page that performs decently on desktop and poorly on mobile has a design problem rather than a ranking problem – and belongs next to mobile-first design rather than next to keywords.
The second filter is the query. Exclude your company name and the numbers fall – that is intended. What remains is demand from people who did not know you yet, and only that remainder can be moved by work on the website. The third filter is the page: select a single subpage and look at which queries it appears for. Surprisingly often it is not the topic it was written for.
That third view pays off on pages every business has. 64% (Bayerisches Landesamt für Statistik) of Bavarian companies with their own website advertise open positions there; career pages generate impressions and sometimes plenty of clicks, but nothing to do with sales. Conversely, only 27% (Bayerisches Landesamt für Statistik) of companies with a website offer ordering, reservation or booking – for everyone else the deal closes by phone or form and appears in no search statistic. Which is why the analysis does not end at the click but at the enquiry on the page.
- Set the period to 28 days and mentally subtract the last three days.
- Switch the device filter to “Mobile” and compare positions against the desktop view.
- Exclude queries containing your own company name.
- Remove career, legal notice and privacy pages from the page list.
- Sort the remaining twenty queries by intent: informing, comparing, commissioning.
- Move only the rows with commissioning intent into your task list.
Six patterns every business finds in the report
Once the report is filtered this way, the same constellations keep appearing, and each calls for a different response. The six cases below cover most of what occurs in a small account.
Good position, weak click-through rate
Ranked third to fifth but barely clicked: that is a copy problem in the result, not a ranking problem. Title and description promise something other than what was searched for. Only 67.7% (Web Almanac) of pages have a meta description at all – the rest leave the wording to Google.
Many impressions, poor position
You appear for a topic but are not seriously positioned within it. A new title will not help here; a dedicated page with enough substance on exactly that question will.
Place names without their own page
Queries with neighbouring towns for which no matching subpage exists. That is the classic gap local visibility closes – one page per location, with real content instead of a list of names.
Seasonal swings
Heating in October, gardens in April, gifts in November. Without a year-on-year comparison every swing looks like a triumph or a crash. With one it becomes a pattern you can plan against.
One page for two topics
The same subpage appears for two different search intents and serves both halfway. Splitting is almost always the better answer here than adding.
Duplicate titles
Two pages with nearly identical titles compete against each other. Titles are basic equipment – 98.6% (Web Almanac) of pages have one – a distinguishable title is not.
Both comparison figures come from a crawl across 16,213,084 websites (Web Almanac) in July 2025 – a cross-section of the open web rather than a customer sample. That makes them usable as a benchmark: if two thirds of all pages carry a meta description, its absence on yours is a gap against the average.
The second half: indexing and page experience
Next to the performance report sits the indexing report, and in small accounts it is overlooked almost across the board. It answers the question that precedes everything else: is your page included at all? A page outside the index produces no impressions, and no amount of copywriting changes that.
The most common causes are technical. On 13% (Web Almanac) of websites the robots.txt returns a 404 error, while 85% (Web Almanac) return a valid 200 status code – which only says the file is reachable, not that the rules inside are correct. 66.01% (Web Almanac) of pages set a canonical tag; without one, the same content shows up under several addresses as duplicates. And around 20.3% (Web Almanac) use hreflang – the declaration that prevents duplicate impressions on a bilingual website.
The third report covers page experience. Only 48% (Web Almanac) of pages pass all three Core Web Vitals on mobile, and the bottleneck is usually loading time: 62% (Web Almanac) of mobile pages achieve a good score for the largest contentful area, while layout stability and responsiveness look considerably better. If you see red here, the order of the fixes is in our article on Core Web Vitals and the practical steps under improving loading speed.
One last point from this block concerns how you appear in the result. 50% (Web Almanac) of home pages carry structured data, 43% (Web Almanac) of all home pages use JSON-LD, by far the most common format – the basis for address, opening hours and ratings showing up directly in the search result. For a local business that is the difference between a bare blue line and a result that already half answers the question. The practical side is in our article on structured data; the entries themselves belong to the business profile.
The order is not arbitrary
What Search Console cannot do
Handling the tool honestly means knowing its limits. The most important one sits in the table below the chart: it shows at most 1,000 rows (Google). The sum of the visible rows is therefore regularly smaller than the figure in the chart above it – a display limit, not an arithmetic error.
If you need more, you go past the interface. The Search Console API returns up to 25,000 rows (Google) per request, twenty-five times what the user interface shows, and the quota is no bottleneck either: Google allows 1,200 QPM (Google) – queries per minute – combined across all of a user's websites, not per property.
| Question | Search Console | Another tool needed |
|---|---|---|
| What are people searching for? | Yes, within the display limit | not included |
| How many people called? | not included | Yes – phone log or form analysis |
| Which copy on the page convinces? | not included | Yes – visitor counter or survey |
| Is the page in the index? | Yes, with a reason per exclusion | not included |
| What does a click cost? | not included | Yes – advertising account |
A third limit is rarely named: Search Console stops at the click. Whether it becomes a phone call, a form or an appointment is decided by the page behind it – which is why the other half of the work is in our article on how to turn visitors into enquiries.
A monthly routine of thirty minutes
The practical value comes from repetition, not from depth. A fixed routine once a month, always in the same order, beats any sporadic deep dive – thirty minutes are enough once the steps are settled.
- Set the period to the last 28 days and switch on the previous period for comparison.
- Note the four tiles – four numbers in a spreadsheet, nothing more.
- Exclude brand searches and note the numbers again. The difference is your share of existing awareness.
- Look at the ten queries with the most impressions and decide per row: relevant or not.
- Mark every row with a good position and a weak click-through rate – that is the copy list for the month.
- Switch to the indexing report and watch only for new exclusion reasons.
- Write exactly one task from all of this into the calendar. Not five.
The last point is the one most often skipped. One page revised per month leaves you with twelve improved pages after a year and a time series to go with them; five things started per month leave you with fifteen half-finished pages and no idea what worked. The same discipline that defines ongoing website care applies to the numbers.
How you notice the numbers are working
After three months of routine, less changes in the report than in the conversation about it. “We are bad at Google” becomes “our services page ranks seventh for comparison queries and is clicked too rarely”. That is the gain: a language in which decisions can be made. Whether that leads to a new subpage, a better title or a rebuild of the website is the second question.
Two neighbouring topics come up regularly. Hospitality businesses discover impressions for ingredients and intolerances nobody had thought of – an argument for a menu online with allergen labelling rather than a PDF. And anyone moving their address should export the report beforehand: a domain migration without downtime can only be judged if the state before it is on record.
It stays as it began: Search Console is a measuring instrument, not a tool. It says where something is missing, not how to build it. How the numbers translate into plannable search engine optimisation is a question of sequence and patience.
A report that is read every month and never translated into a task is more expensive than no report at all: it costs time and creates the feeling that something has been done.
Sources and studies