An English-language page on a Swiss domain lives with a permanent handicap that no report explains: the market it serves mostly searches in German. The page can be well written, correctly built and genuinely useful, and still show a fraction of the impressions its German sibling collects — not because it is worse, but because fewer people phrase the question in its language.
This matters more in Zurich than almost anywhere else. The city runs a large share of its business in English while the surrounding search behaviour stays German. A company with an international client base, an English-speaking staff and a German-speaking neighbourhood ends up with two audiences that never appear in the same row of any report.
What follows is a way of reading search data that keeps those two audiences apart from the first screen onward. The views described sit in the Semalt workspace, where Search Console figures and rank tracking share a project, but the reasoning applies to any tool that can split a report by language and country.
The report averages two markets that have nothing in common
A single domain serving both audiences produces one set of headline figures. Those figures are the weighted average of a large German-language demand with low intent and a small English-language demand with high intent. Averaged together they describe neither. Read separately they describe two workable markets.
- German-language queries higher volume, broader phrasing, a large share of impressions from outside Switzerland.
- English-language queries lower volume, more specific phrasing, and frequently the ones attached to an actual budget.
- Brand queries arrive in both languages and belong in neither bucket — they measure recognition, not acquisition.
- The averaged view hides the only comparison worth making: how each language performs against its own baseline.
Doing this consistently requires a decision that many teams postpone: which language a given page is meant to serve. A page that exists in both versions but is linked only from the German navigation is an English page in name only, and its figures will say so.
Why the four headline numbers behave differently per language
Impressions, clicks, click-through rate and average position are the same four figures in both languages, but they fail in different ways depending on which side of the split you are reading. Knowing which failure applies saves most of the wasted analysis.
Volume with a wide geographic spread
Plenty of rows, plenty of impressions, and a large fraction of them from people who cannot buy from you.
- Typical failure Click-through rate looks poor because it is computed across a market you do not serve.
- Correct reading Apply the country filter first, then judge the rate against the domestic slice only.
- Useful signal Queries that appear domestically but not abroad — those are the locally rooted ones.
- Trap Chasing the larger foreign demand with pages nobody in your pipeline will read.
Few rows, high intent, unstable ratios
Small denominators make every derived figure jump, and rare queries disappear from the breakdown entirely.
- Typical failure A rate computed over nine impressions is treated as a trend.
- Correct reading Use absolute counts over a quarter; ignore ratios below roughly a hundred impressions.
- Useful signal The keyword dynamics view, which names entries into and exits from the top ten.
- Trap Concluding a page failed when it simply serves a question few people ask in writing.
The practical consequence is that the two language sets need different decision thresholds. Applying one rule to both is how English pages get retired for underperformance when they were never measured against a fair baseline in the first place.
| Question | German side | English side |
|---|---|---|
| What counts as enough data? | A month is usually sufficient | A quarter, sometimes two |
| Which figure leads? | Click-through rate, domestic only | Absolute clicks and query count |
| What does position mean? | Contested; small moves matter | Often uncontested; presence matters more |
| What is the main distortion? | Foreign impressions | Rare queries omitted from the breakdown |
| When to rewrite a page? | Steady impressions, weak rate | Impressions present, no landing page fit |
| When to leave it alone? | Seasonal dip | Fewer than fifty impressions a quarter |
There is one more asymmetry worth naming, because it explains a recurring argument between marketing and sales. The German side generates most of the traffic and almost none of the pipeline; the English side generates almost none of the traffic and a disproportionate share of the pipeline. Both departments look at the same dashboard and reach opposite conclusions about which pages deserve investment, and both are reading the figures correctly. The disagreement is not about data quality. It is about which of two markets the report is describing at any given moment, and it disappears the day the split becomes the default view rather than an occasional filter. Setting that up takes a saved segment and ten minutes, and it is the highest-return configuration change available in a managed campaign or a self-run one alike.
Queries describe demand, pages describe supply
The query view reports what people typed; the page view reports which of your addresses appeared. Neither answers the question that matters, which is whether the right page appeared for the right question. Only the crossing of the two does that, and in a bilingual estate it is where the interesting failures hide.
The German page answering an English query
Common where the German version is older and better linked. The visitor arrives, cannot read it, and leaves.
- Check whether an English equivalent exists
- Check whether it is linked from anywhere
- Check the language annotations on both
Two of your own pages competing
The English page and a translated variant of it both appear, and neither settles into a stable position.
- Decide which one is the original
- Point the other at it as the preferred address
- Consolidate the internal links
Demand with no page at all
The most valuable finding. A phrasing that produces impressions from a page that only mentions the topic in passing.
- Confirm it persists across two quarters
- Confirm it matches something you sell
- Then build the page, not before
Failure C is worth a habit of its own. Once a quarter, sort the English query rows by impressions and read the ones with no obvious destination. In an estate that has grown organically, three to five of those usually turn out to be services the company genuinely offers and has simply never described in English.
The country filter is not optional here
English pages on a Swiss domain attract attention from everywhere. Some of it is valuable — an international client base is exactly why the pages exist — and some of it is noise from markets where you have no delivery capability. Unlike the German side, where the foreign share is predictable, the English side needs a judgement call about which countries count.
Once that list exists, a second question becomes answerable: is the English demand concentrated in Switzerland or spread across the continent? The two answers lead to very different content plans. Concentrated demand argues for pages about the local context in English. Spread demand argues for pages about the subject matter, with the location as a footnote.
This distinction also settles a recurring argument about the domain itself. A Swiss domain serving concentrated local demand in English needs no further justification. The same domain serving spread international demand competes against providers with more established addresses, and that is a strategic decision rather than a technical one — worth taking deliberately rather than discovering after a year.
Devices, and what they say about the reader
In a bilingual professional estate the device split carries an unusually clear signal. English-language business pages are read overwhelmingly at a desk, during working hours, often from a corporate network. German-language pages covering the same subject show a much larger phone share, because they catch a wider and less committed audience.
- Desktop dominance on an English page suggests it is reaching the intended professional audience.
- A sudden phone share increase usually means the page started ranking for a broader, more casual query.
- Weak phone click-through points at a title that truncates before the distinguishing word.
- Tablet rows are too small to support any conclusion at Swiss volumes.
The practical use is in deciding what to fix first. A page read at a desk can carry a comparison table and a longer argument. A page read on a phone needs its distinguishing claim in the first line and a way to make contact without scrolling. Applying the wrong remedy to either wastes the revision.
Comparing two periods without fooling yourself
The 28-day and 90-day presets answer the two questions anyone actually has, and the two-day reporting lag catches out everyone at least once. Pull a report on Monday morning covering through Sunday and the weekend will look like a collapse that never happened.
Comparison across periods also assumes the underlying estate stayed the same. In bilingual sites it frequently has not: a language version was added, a set of pages was consolidated, or the navigation changed and with it the internal link structure. Any of those makes two periods incomparable. Note such events on the chart, or the team will spend a quarter explaining an effect that a migration caused. The technical side of those migrations is covered under technical work.
The competitive field looks different in each language
Run the competitor view twice, once per language, and the two lists rarely overlap by more than a third. The German list contains regional providers and large directories. The English list contains international firms, often much larger, that appear because they publish in English about the same subject and are not competing with you for clients at all.
| Who appears | Why | What to do about it |
|---|---|---|
| Regional providers, German | Same clients, same city | Compare term by term, close the gaps |
| Large directories | Authority without depth | Compete on specific questions, not broad ones |
| International firms, English | Publishing overlap, no client overlap | Treat as background; do not benchmark against them |
The third row is where most misreadings happen. Seeing a global consultancy in your English competitor list is not evidence that you are losing to it; it means you both wrote about the same topic. Benchmarking your domain authority against that is a reliable way to conclude that nothing can be done. The generated market view is more useful here, because it separates the firms occupying your actual market tier from those merely sharing vocabulary with you.
Turning the split into decisions
Reading two markets separately is only worth the effort if it changes what gets built. That requires written thresholds, one set per language, agreed before the next report rather than argued about after it.
Revise an existing page
At least 200 domestic impressions a quarter, click-through below four percent, position between five and fifteen.
- Title and description first
- Body text second
- Re-read after 28 days
Build a new page
A query with impressions across two quarters, inside the served-country list, with no page addressing it.
- Confirm it maps to something sold
- Write it as an original, not a translation
- Link it from at least two places
Do nothing
The case that is almost always missing, and without which every fluctuation becomes a task.
- Under fifty impressions a quarter
- Movement smaller than three positions
- Seasonal terms out of season
The middle card carries the instruction that matters most in a bilingual estate: write the second language version as an original. Translations of the same argument compete with each other for the same ideas and read as thin in both languages. Two pages built from the same facts but aimed at different readers do not have that problem. The editorial side of that work sits under content.
What the figures will not tell you
The deeper limit is that search data carries no notion of value. Forty clicks from people comparing providers for a mandate are worth more to a Zurich advisory firm than four thousand from students looking up a definition — and in a bilingual estate the forty are usually the English ones. That weighting exists nowhere in the reporting. It has to be written down once, by whoever owns the pipeline, and placed beside the figures every time they are read.
If you want to see your own estate split this way, connect the domain in the Semalt workspace and give it a quarter before drawing conclusions about the smaller language. What usually changes the plan first is not the average position but the list of English queries that produce impressions and have nowhere sensible to land. Further pieces on the subject sit in the blog, and the campaign side is covered under automated campaign work.
Should English and German pages sit on the same domain?
In almost all cases yes, in separate directories with mutual language annotations. Splitting them across domains halves the accumulated authority of both and doubles the maintenance. The exception is a genuinely separate international business with its own brand, which is a corporate decision rather than a search one.
Why does the English version show so few impressions?
Usually because fewer people search in English for that subject in this market, not because the page is weak. Check the absolute query count rather than the ratio, and compare it with the English demand for the same subject in a market where English is dominant. The gap is the market, not the page.
Is it worth tracking the same keywords in both languages?
Track the terms that have a destination page in that language, and only those. Tracking the English equivalent of a German term for which no English page exists produces a row that can never improve, and a report full of such rows stops being read.
How should brand queries be handled in the split?
Separately from both. They measure whether people already know the name, which is a different question from whether the site attracts new demand. Leaving them inside the language buckets inflates whichever language the brand is more commonly typed in and flatters the overall click-through rate.
How often should this analysis be repeated?
Monthly for decisions, quarterly for the language comparison itself. The split changes slowly, and reviewing it too frequently invites conclusions drawn from a handful of clicks — which on the English side is nearly always what a month of data amounts to.
What is the single most common mistake?
Judging the English pages by the German pages' numbers. They serve different demand, at different volumes, with different intent. Once each language is measured against its own baseline, most of the apparent underperformance turns out to be an artefact of the comparison rather than a property of the pages.
Ready to improve your SEO?
Request a free SEO audit and see where the rankings can move.
Request a free audit