A purchasing engineer sourcing a machined component does not type the name of a city. He types a material, a tolerance, a certificate and a delivery window. Search Console records which country he was sitting in when he did it, and that column is the closest thing a Czech supplier has to an early warning system.
Search analytics is usually explained as though every business sold to the street outside its door. A large share of the companies registered in Prague do not. They sell capacity: engineering hours, machining time, laboratory throughput, back-office headcount. The customer is in Munich, Malmö or Austin, has never been here, and is running a comparison in which Prague is one of five candidate locations.
The query with no city in it
Consider a toolroom on the eastern edge of the city that produces injection moulds and does five-axis work for German tier-one suppliers. Its Czech pages are read by people who already know the firm: recruits, local subcontractors, a regional customer, the bank. Its English pages are read by somebody who does not know it exists and is halfway through a sourcing exercise.
That second reader arrives with a requirement document open beside him, and what goes into the box is a fragment of it: a process, a material grade, an accreditation, sometimes a batch size. The word Prague appears nowhere, because location is not the requirement — at most a filter applied afterwards.
- The process, named the way the trade names it. Not "manufacturing" but wire EDM, anodising, build-to-print machining — phrases that survive being pasted from a specification into a search box.
- The certificate or accreditation. An automotive quality standard, a laboratory accreditation, an information-security certification. Buyers search for these because a missing one ends the conversation.
- The constraint that hurts. Small series, prototype quantities, a tolerance band, a lead time in days. Whatever he has been told is hard to source is what he searches for.
- The commercial shape of the deal. Contract development, managed team, dedicated capacity, framework agreement — words for how the work will be bought, which rarely map onto their Czech equivalents.
None of that is a local search. It is a sourcing search, and the result page it produces lists suppliers in Poland, Portugal, Turkey and Vietnam alongside you. You are not competing with the firm in the next industrial park; you are competing with one whose hourly rate is a third of yours, on evidence the reader can gather in an afternoon.
Which market is sourcing from you this quarter
Most people treat the country breakdown as decoration — a map for the last slide. For a supplier it is the most commercially loaded view in the dataset, because it answers a question sales cannot answer for months: which economy is looking for what you make.
The Search Console side of the panel runs to eight connected views — a dashboard across every verified site, then site overview, keywords, pages, devices, countries, traffic and keyword dynamics — each of which can be narrowed. A supplier should start with countries, and read it as a time series, quarter against quarter, rather than as a snapshot.
Here is the pattern worth watching. German impressions on your capability pages hold steady, as they have for two years. Then over ten weeks Swedish and Dutch impressions on the same pages roughly double, on phrases you never targeted. Nothing about your site changed. Buyers in two other countries began sourcing in your category.
Your commercial team learns about that wave when enquiries arrive, six to twelve weeks later. The country column shows it while it is still happening. That is not a marketing metric; it is an input to a capacity conversation.
| What the country view shows | Plausible reading | What to check next | Who should hear about it |
|---|---|---|---|
| A new country appearing with real impression volume | A sourcing wave has started in a market you do not cover | Which pages and which phrases carry it | Sales, before the enquiries land |
| Impressions up, clicks flat, in one country only | You are being listed but not shortlisted there | Titles and descriptions on the pages involved | Whoever owns the page copy |
| A long-standing market declining steadily | A competitor has improved, or demand has moved | Competitor domains in the same keyword space | Commercial management |
| Domestic impressions dominating an English page | Local staff, recruits or partners are reading it | Whether the page is doing an export job at all | Whoever set the target for that page |
The same company described in two vocabularies
The Czech pages and the English pages of a Prague supplier describe one business and share almost no vocabulary. That is not sloppiness. The two audiences stand in different places: one already knows the company and is checking a detail, the other is comparing unknown vendors against a written requirement.
The failure mode is predictable. Somebody takes the Czech page that performs well at home and has it translated faithfully. The result is grammatical, accurate and invisible, because the term carrying the domestic business has a literal English rendering no buyer has ever typed. The page ranks for nothing, produces no impressions, and therefore leaves no trace of what went wrong.
The faithful translation
Rendered word for word, the Czech term is correct English and a phrase nobody outside this country uses, so it has no volume anywhere.
- Reads well to a Czech reviewer
- Produces a page with no impressions at all
Start from the requirement
Write from a real specification rather than from the Czech page. The phrases you need are in the tender you last lost.
- Certificates, processes, tolerances, quantities
- Verified afterwards against actual query data
Two neighbouring failures do the same damage: a broad, safe word lands the page on a phrase contested worldwide, and a service described by its internal meeting-room label competes only with itself. The keyword pool in the panel is built for all three. It draws candidates from three sources at once — Search Console, live result pages and seed terms you supply — and each candidate is approved, rejected or deferred individually rather than accepted wholesale. That review step is where a Czech-thinking term gets caught before it becomes a page. The generative market-research views add intent classification, separating a phrase typed while sourcing from one typed while writing a school assignment.
Position four, and still nobody clicks
Once your English pages rank, a frustrating pattern shows up: a page sits in the upper half of the first result page for a relevant sourcing phrase, collects hundreds of impressions a month and converts almost none. That is not a ranking problem, and no amount of link building fixes it.
The searcher is running a first-pass elimination on the result page itself. Five candidates, three tabs, and the choice made from roughly fifteen words each. If your fifteen words give a company name and a generic phrase while the competitor below gives the standard he needs plus a lead time, he opens the competitor. The on-site suggestion engine flags pages where that gap is widest.
- Put the credential where it can be read without a click. A standard or a capability limit in the title carries more weight than any adjective. The buyer is looking for a reason to keep you on the list.
- Answer the constraint the query contains. If the phrase names a small batch or a prototype quantity, the snippet should acknowledge it. A generic description reads as a firm that does not do that work.
- Do not spend the space on geography. Naming the city where the buyer expected a capability wastes your most valuable characters. Location can be established on the page.
- Read click-through rate query by query. A site-wide average across a bilingual property describes nothing; the gap appears only per phrase.
- Watch the pages view alongside. When several pages compete for one phrase, the one shown may not be the one you would have chosen, and its snippet is doing the arguing.
| What the query contained | Snippet that loses the click | Snippet that keeps you on the list |
|---|---|---|
| A process plus a certification | Company name and "quality manufacturing services" | The process named, the standard named |
| A tolerance band | "Precision engineering in Prague" | The tolerance held, and on which machines |
| A prototype quantity | "Your serial production partner" | Minimum quantity stated, lead time in days |
| A managed development team | "Innovative software solutions" | Team size, contract model, security certification |
Why band changes beat click totals
Sourcing phrases have low volume. A precise description of a process, a material and a certification might be searched a few dozen times a month across all of Europe. That is not a weakness but the nature of the demand: behind each search sits a budget, a specification and an approval process.
Click totals are therefore useless in the short term. Two clicks becoming five is noise, and you cannot wait a year to find out otherwise. What is readable at this volume is band movement: whether a phrase entered or left the top three, ten or thirty. The keyword dynamics view tracks those transitions, and signals months before the click count does.
Bands also carry meaning positions do not. Reaching the top thirty means a page has been recognised as relevant. Crossing into the top ten means it appears while a buyer is comparing. Entering the top three means it is seen by people who open two tabs and stop. Averaging those into one figure erases the distinction.
First measurable movement after a change typically appears in four to eight weeks, and on thin export volumes expect the slower end. Judge the wait on band transitions, not on a click line with too little data to draw.
The competitor list nobody expects
The rank-tracking side runs to six views: a ranking score across your sites, average position, keyword counts, top phrases by volume and position, best pages, and a competitor view listing the domains contesting your keyword space with a domain authority figure and a shared-keyword count. A portfolio ranking figure with a twenty-eight-day trend sits alongside it.
For a Czech exporter that is the most uncomfortable screen in the account, and the most useful. The domains listed are rarely the firms named in your commercial reviews. They are suppliers in other mid-cost manufacturing countries, directories aggregating hundreds of vendors, and occasionally a customer's own procurement portal.
The equivalent supplier elsewhere
A firm with comparable equipment in another country, competing on price or proximity to a different customer base.
- Compare on certifications and evidence, not rates
- Watch which phrases they hold that you do not
The aggregator
A platform or directory ranking for hundreds of capability phrases, listing suppliers including, possibly, you.
- Rarely beatable on broad phrases
- Beatable on specific, constrained ones
The authority comparison is where strategy gets decided. If the domains above you carry far more authority and far more shared keywords, contesting the broad phrase is a multi-year project. The constrained phrase — the one carrying a standard, a tolerance and a quantity — is where a specialist holds a position a general platform cannot, having nothing specific to say about it.
Turning a reading into something somebody does
Analytics pays for itself only when a number changes a decision. For an export supplier those decisions are few and large: which capability to describe next, and which market to work this quarter.
AutoSEO — the campaign runs itself
For a company with no in-house search specialist and no plan to hire one.
- Keywords found and prioritised automatically. Candidates are gathered and ordered without anyone maintaining a list, which matters when nobody in the building has time to.
- Backlinks built without manual placement. Placements draw on a partner network of more than 230,000 websites.
- On-site suggestions and full analytics. Search Console and rank-tracking data in the same place, plus a live chat assistant tied to the project.
FullSEO — with a hand on the wheel
For a supplier whose terminology is technical enough that automated choices need review.
- Manual keyword selection with automatic fallback. You approve the phrases that describe your capabilities; anything you do not reach is handled automatically.
- Backlink placement with a domain-rating target. Placements are chosen against a quality threshold rather than accepted as they come.
- Human review before on-site changes. A team of specialists, developers and writers works alongside the automation — the part that matters when a wrong word in a technical description costs a tender.
Whichever level you run, the reading must survive leaving the screen. Exports go to CSV or JSON at up to 10,000 rows and to PDF at up to 250 rows, rendered server-side, through a report builder carrying your own logo and colours. A quarterly country heatmap and a band-movement table are usually enough for a board. Sites can be shared to individual email addresses, so an export manager sees one property and nothing else. Semalt's reporting and export layer covers time series, metric cards, sortable tables and heatmaps.
There is also a continuous version of the same reading. The Stream feed inside My SEO keeps answers, automatic reports, newly placed backlinks with donor rating and traffic, to-dos and campaign news in one chronological line, and takes keyword and URL lists in batches. A router model decides per question which data blocks to load, and the feed is full-text searchable.
Common questions
Our English pages get impressions from countries we do not sell to. Is that a problem?
Usually not. Capability phrases are international by nature and attract researchers, students and competitors along with buyers. It matters only when that traffic distorts the average position of pages you care about, in which case read those pages individually rather than in the site total.
Should we translate our Czech service pages or write new English ones?
Write new ones, from a requirement document rather than from the Czech page. The two audiences share the subject but not the vocabulary, and a faithful translation reliably produces a page that is accurate and unfindable. Keep the technical facts; rebuild everything around them.
How much history do we need before the country view means anything?
Two full quarters is the practical minimum, because sourcing cycles are seasonal and one ninety-day window cannot separate a real shift from a holiday period. Use the twenty-eight-day preset for operational checks and the ninety-day one for a trend.
Our best phrase has forty searches a month. Is it worth targeting?
Frequently yes. Forty searches naming a specific process and certification are forty buyers with an open requirement, and one of them converting can outweigh a year of general traffic. Judge export phrases on the value of the enquiry, not on volume.
Can we see the Czech and English sides separately without two accounts?
Yes. Filter the pages view to the English directory, read the figures, then invert the filter. With several domains, site tags apply the same separation across the account.
Where to start on Monday
Open the countries view, set it to ninety days and restrict it to your English pages. Write down the top five countries and their impression counts, then set the same view to the previous ninety days and write them down again. The difference says which markets are sourcing in your category now, and it will not match the ranking your sales team would give from memory.
Then read the click-through rates of the three phrases carrying the most impressions. Anything sitting high with a poor rate is a snippet problem you can fix this week. Anything with no impressions is a translation problem, an indexing problem, or both — and indexing is cheaper to rule out first, which is where technical SEO work earns its place. Further reading sits in the English articles section.
To see this on your own property rather than in the abstract, run the two ninety-day windows side by side: open the dashboard and connect a verified property, then compare the country lists quarter against quarter. It changes who hears about a new market first — the report, or the enquiry two months later.