At the end of every quarter the same question comes up in a lot of small companies: is the website actually doing anything? The answer is usually given from the gut. The owner searched for the business and was happy; a colleague in the office searched for it and was not. Yet the answer already exists. For every page that appears in a search engine, four plain numbers are produced. They tell you how often the page stood in the result list, how often someone clicked it, how those two relate to each other, and where the page was sitting while it happened. This article explains those four metrics in plain language, shows the three patterns that make the biggest practical difference, and turns them into a short list of tasks per quarter instead of a permanent watch that nobody keeps up with in day-to-day business.
Four metrics are enough to start with
To find out which terms bring your website impressions and which of them lead to clicks, you do not need a tool with dozens of dials. The raw data comes from the search engine itself. The performance report groups it by dimensions, among them queries, pages, countries, devices and search appearance (Google Search Console Help). The most productive of these is the query dimension, because it shows the words people actually typed before they landed on the site. That is a world apart from the words the business uses to describe its own work. How to close that gap systematically is covered in the article on the search terms customers really type.
Impressions
„How many times a user saw a link to your site on Google“ (Google Search Console Help). In other words: the page stood in the result list. Whether anyone consciously looked at it is not part of the number.
Clicks
„How many times a user clicked a link from Google to your site“ (Google Search Console Help). This is the moment a result turns into a visit, and the only number that costs something when it stays low.
Click-through rate
„The click count divided by the impression count“ (Google Search Console Help). It measures how convincing the entry in the list was, not how good the page behind it is.
Average position
„A relative ranking of your link position on Google, where 1 is the topmost position“ (Google Search Console Help). What is counted is the topmost result from your own site.
Taken on its own, each of these numbers is harmless. Only together do they say something. A worked example from one quarter: 12,480 impressions, 268 clicks, a click-through rate of 2.1 percent and an average position of 8.4 (project experience). That combination describes a business that is being found, whose entries in the result list rarely tip the balance, and that mostly sits in the lower half of the first result page. It is a completely different situation from 800 impressions with 90 clicks — there the presentation is convincing, but far too few people ever see the entry.
The order of the questions matters. As long as there are hardly any impressions at all, this is not an analysis problem but a visibility problem: pages may be missing, unindexed, or thin on substance. The most common causes and how to check them are collected in the article on why a website does not show up in search. Only once impressions exist does detailed work on click-through rate and position pay off.
What average position actually tells you
Average position is the metric most often misread. It is not a fixed rank but a mean across many individual searches that differ by location, device, time of day and search history. A position of 8.4 therefore does not mean the page sits in eighth place on every search. It may show up in third place for some queries and in eighteenth for others; the average lands somewhere in between and hides both ends.
The average position of the topmost result from your site.
The counting method deserves a look as well. If two results from the same site appear for one query, the chart counts that as a single impression (Google Search Console Help), and position is taken from the topmost of them. So if you compete with two pages on the same topic, the metric will not show it directly — but you will notice that one page collects clicks while the other comes away empty. That is a good moment to merge two competing pages into one strong one.
Compare first, then act
A second aspect is the device. Around 82 percent (Statista) of people aged 16 and over in Germany use a smartphone, and 93 percent (Federal Statistical Office of Germany) of people aged ten and over are online at all. On a small screen, a result in fourth place sits visually much further down than it does at a desk. It is therefore worth splitting the analysis by device once, especially for location-based queries of the kind described in the article on local visibility for trades and hospitality.
Pattern 1: plenty of impressions, few clicks
The first pattern is also the most rewarding one, because it can be fixed without producing new content. It occurs when a page collects many impressions while its click-through rate sits clearly below what is usual at that position. Translated: the search engine considers the page relevant enough to show, but the people looking at the result list do not. Between the two stands a single text of roughly three lines.
Visible is not the same as convincing
Search engines recommend descriptive, concise text instead of vague labels (Google Search Central). In practice a handful of corrections is usually enough: the service up front, the location where it genuinely applies, the company name at the end. Which patterns work for which page type, and why search engines sometimes rewrite titles, is covered in detail in the article on titles and descriptions in search results.
- The title names the service in the customer's language rather than in the company's internal wording.
- The location appears exactly once in the title, and only where the page really covers it.
- The description names an occasion and a condition, such as an on-site measurement, an appointment within days, or pricing based on effort.
- No sentence from the description appears word for word on another page.
- Title and description match what the page actually shows — otherwise the click-through rate rises and the visit ends after eight seconds.
- After the change, impressions, clicks and click-through rate are written down so the next comparison has a baseline (Google Search Console Help).
Pattern 2: the bottom of page one and all of page two
The second pattern shows where effort achieves the most. Pages in the top spots are usually close to their ceiling; further improvement there costs a lot and returns little. Pages far down the list are so far away that a text correction will not close the gap. The range in between — the lower half of the first result page and the second page — is where small improvements become visible, because impressions already accumulate there and the competition is within reach.
| Position range | What the numbers typically show | Sensible task |
|---|---|---|
| Positions 1 to 3 | High click-through rate, stable impressions. The page is already carrying its weight. | Watch it and keep the content current; do not rebuild it without a reason. |
| Positions 4 to 10 | Many impressions with a strikingly low click-through rate. | Sharpen title and description, and point the opening of the page at the concrete question. |
| Positions 11 to 20 | Impressions exist, clicks are barely measurable. | Add real substance: process, conditions, images, the questions that come up in customer calls. |
| Beyond that | Few impressions, no reliable click figures. | Check whether the page is needed at all, or merge it into a stronger one. |
| No impressions | The term does not appear in the report at all. | Sort out the basics first: does a suitable page exist, is it indexed, is it linked internally? |
Competition for these spots is dense: around 89 percent (Federal Statistical Office of Germany) of enterprises with internet access run their own website, and 99.3 percent (Federal Statistical Office of Germany) of all enterprises in Germany are small and medium-sized firms competing for the same regional enquiries. That makes it all the more important that a page really answers the question someone arrives with. Wording that manages this is described in the article on website copy customers actually read; which pages you need in the first place is covered in the article on the pages a business website needs.
Pattern 3: brand queries and demand queries are two worlds
The third pattern is not an error in the data but an error in the reading. Anyone who looks at brand queries and demand queries in one pot ends up with an average that says nothing. Brand queries contain the company name. There the website is as a rule the expected destination, the position is high and the click-through rate strong. Demand queries ask for a service, often combined with a location. That is where new business is decided — and where position and click-through rate are naturally weaker.
Brand queries
They contain the company name or a recognisable variation of it. They show how much recognition comes back from referrals, vehicle lettering, advertising or word of mouth. When this group grows, the work outside the website is paying off. It is a poor yardstick for copy work, because the click-through rate here is high anyway.
Demand queries
They describe a need: a service, a problem, a question, frequently together with a location. This group is the real measuring ground for titles, descriptions and content. Here position and click-through rate are worth watching, here new enquiries originate, and here you see whether the website speaks the customer's language.
The difference is substantial. In the example above, the click-through rate for brand queries was 18.6 percent (project experience) and for demand queries 1.4 percent (project experience). Viewed together, the two groups produce the unremarkable 2.1 percent from the overview — a number that describes neither of them. In practice it is enough to filter the report for queries containing the company name and to note both views separately. Which terms belong in the demand group at all can be worked out cleanly with the methods from the article on finding search terms for small businesses.
The list of tasks per quarter
The three patterns turn into a routine that needs four appointments a year and roughly an hour each. In a small business, more is neither necessary nor sustainable. The limit is not about convenience: changes need time before they are recrawled, processed and visible in the numbers. Anyone adjusting weekly ends up measuring their own restlessness.
- Pick the period: the completed quarter against the previous one, and for seasonal businesses also against the same quarter last year.
- Filter out brand queries and treat the demand queries as a separate list (Google Search Console Help).
- Mark the ten queries with the most impressions and note clicks, click-through rate and position beside each.
- Flag every row whose position sits between 4 and 20 — that is the working list for this quarter.
- Decide per row: sharpen title and description, add substance to the content, or merge two competing pages.
- Change at most five to ten pages per round so the effect can still be attributed in the next comparison.
- Write down the date and the change in a short note, and pull the list out again at the next appointment.
Do not change everything at once
This routine has a pleasant side effect: over the year it produces a list of topics that customers genuinely care about. Queries without a matching page become the next pieces of content, without anyone having to invent an editorial theme. How that list turns into finished pages in a few steps is shown in the overview of how a site goes from idea to launch.
Search data and site analytics complement each other
Search data ends at the click. It tells you how someone searched and that they arrived, but nothing about what happened next. Your own site analysis starts exactly there: which pages were opened, for how long, in what order, and at which point the contact form was submitted. Together the two sources form an unbroken line from the search query to the enquiry in your inbox.
| Question | Answer from search data | Answer from site analytics |
|---|---|---|
| Which terms produce impressions? | The full list of queries per page (Google Search Console Help). | Not included — analysis only begins after the click. |
| Why is an entry not clicked? | Click-through rate in relation to position shows the pattern. | Not included — a non-click leaves no visit behind. |
| What happens after the click? | not included | Pages opened, time on page, exits, the path to the form. |
| Which page produces enquiries? | Only indirectly, via clicks per page. | Directly: the path from entry point to submitted form. |
| Which period is reliable? | Comparison of completed quarters; preliminary daily values still change (Google Search Console Help). | Also a quarterly comparison, plus individual campaign periods. |
What matters is how that second analysis comes about technically. In the wild, several scripts are often embedded that hand visitor data to third parties — with consequences for load time, for consent requirements and for the length of the privacy policy. There is another way: a server-side, aggregated analysis without personal profiles needs no additional scripts in the frontend. It counts page views, referrers and completions, but it does not follow individuals across sites. What belongs to the scope of delivery is described in the XICflow product overview.
Fewer scripts, clearer numbers
A simple rule helps with attribution: search data answers the why before the click, site analytics the why afterwards. Put both side by side each quarter and breaks stand out quickly — a page with rising clicks but persistently few enquiries, for instance. In that case the cause is not in search but in the page itself: an unclear opening, missing conditions, an awkward form. Which pages should take on these jobs is described in the article on the pages a business website needs.
What the numbers cannot tell you
Sound analysis includes knowing where the data stops. The performance report is a statistic, not a log. It reflects what could be measured and deliberately leaves gaps where the analysis would otherwise point back to individual people. Knowing those limits saves you from false conclusions.
- No people. The report shows aggregated queries, not user profiles. Who called and why is not in there.
- No complete picture. Very rare queries are left out for privacy reasons. In small businesses in particular, those single cases add up.
- No cause. A position that has improved does not say why it improved. It may come from your own work, from a competitor taking a page offline, or from a changed result layout.
- No assurance. An improvement last quarter does not carry over to the next one. Reindexing can be requested but not forced (Google Search Console Help).
- No substitute for conversations. No report answers why an offer was turned down. Three phone calls with customers deliver more on that than three hours of spreadsheet work.
None of this is a reason to skip the analysis — but it is a reason to place it soberly. It is a compass, not a map. It shows the direction in which effort pays off, and it disproves assumptions that have settled in over the years. Which requirements a website should meet so that such changes can be made quickly at all is set out in the comparison of build approaches.
From finding to change in minutes
The most common reason analyses stay without consequence is not a lack of insight but a lack of access. The finding sits in the spreadsheet, but the change needs an appointment with an agency, a ticket, an approval and a release date. By the time the new title is live, the quarter is over. This is exactly where XICflow starts: analysis and editing live in the same tool.
Analysis is built in
Every website built with XICflow comes with an aggregated analysis — without personal profiles and without additional scripts in the frontend. Page views, referrers and submitted forms sit in the same account as the pages themselves.
Title and text per page
Title and description are their own fields per page and per language version, with a length indicator before publishing. A finding from the quarterly comparison can be translated straight into a change, without a detour through source code.
Publishing in minutes
Once approved, the page is delivered statically again. A note in the quarterly review becomes an amended page the same morning — instead of a ticket waiting for the next development cycle.
In practice this shortens the loop from weeks to minutes, and that changes how seriously an analysis is taken. Anyone who sees an observation turn into a visible change on the same day will look again next quarter. What the finished results look like is shown by the example sites from various industries; what it costs is set out in the overview of plans. And anyone who wants to understand the process first will find it step by step under how the build works.
Sources and studies