Your product documentation was written in English because your engineers work in English. Your German pages arrived nine months later as translations. Search Console reports both, and the figures suggest the German market is weak — when what it actually shows is that your German pages are nine months behind your own thinking.
This is close to universal among Frankfurt fintech and financial-technology teams. The specialist material — API documentation, regulatory explainers, integration guides — originates in English because that is the working language of the people who build the thing. German follows when somebody has time, and by then the product has moved.
The numbers do not say the German market is difficult. They say your German material is older, thinner and derivative. Those are different problems with different remedies, and telling them apart takes three specific slices of the data.
Why the German figures lag your English ones
A page that has existed for two years, been updated four times and is linked from every release note behaves differently from a translation published last quarter. That difference has nothing to do with language and everything to do with age, maintenance and internal links — three things the translated version does not inherit from its source.
Comparing the two language trees on absolute figures therefore compares a mature page against a new one and calls the result a market difference. It is not a market difference. It is a maturity difference, and it will persist for as long as the German tree is a downstream copy of the English one.
The split that actually matters here
For a Frankfurt technology company, the informative cut is not country but query language — and the two do not line up the way people assume. A substantial share of German-market demand is expressed in English, by people who work in English all day and search accordingly.
Country and language, crossed
Four cells, four different audiences, and only one of them is served by a translation.
- Germany, English query. Engineering and product staff at German institutions, plus international teams based here. They will read your English documentation and are already served — this cell usually looks healthy and is often mistaken for the whole picture.
- Germany, German query. Procurement, compliance, risk and legal functions. They search in German, they read German, and a translated page written for engineers does not answer what they are asking.
- Outside Germany, English query. Frequently the largest cell by volume and the least commercially relevant if your licence or your product is Germany-bound. Worth separating before it flatters the total.
- Outside Germany, German query. Small, and disproportionately Austrian and Swiss. Relevant only if you can actually serve those markets — which is a licensing question, not a content one.
The second cell is where the money is and where translated documentation performs worst. A compliance officer evaluating a payments provider is not looking for an endpoint reference; they are looking for who holds which licence, what happens to funds in an insolvency, and which supervisory regime applies. None of that is in the English documentation either, so translating it produces nothing.
Which view answers which question
Within the Semalt workspace the Google reporting arrives split into eight separate screens instead of a single report with filters. For a bilingual property that arrangement is useful, because the interesting comparisons are all cross-sections rather than totals.
| Who asks | The question | The view |
|---|---|---|
| Founders or board | Is Germany worth a dedicated content effort? | Property dashboard with clicks, impressions and health flags |
| Product marketing | Which capabilities are people searching for? | Keywords with position history and click trend |
| Content | Which page is doing the work? | Pages, with the same metrics by landing page |
| Anyone arguing about translation | Is German demand real or imagined? | Countries, crossed with the query language |
| Whoever owns the roadmap | Are we gaining or losing ground? | Keyword dynamics: entries and exits from the top 3, 10 and 30 |
The last row is the only one describing a direction rather than a state, and it is the only one that justifies a weekly habit. Fifteen minutes a week is enough for it, and the entries-and-exits view in My SEO Stream reports the same movement into the project timeline, so the record survives whoever happened to be looking that week. Everything above changes slowly enough that a monthly or quarterly look is sufficient, and teams that attempt to review all eight weekly stop reviewing anything by the third month.
What your own property cannot show
Search Console stops at your domain boundary. Which other domains appear for the same terms, how strong they are, and whether the field is tightening is absent from it entirely. Six further views on the position-tracking side fill that gap, and in this market they surface the same finding with tedious regularity.
- A consolidated score per property. Aggregated across every tracked term, alongside average position and term count — the single figure a board tends to ask for.
- Strongest terms by volume and current position. Ordered by what people actually search rather than by what the product is called internally, which in fintech is rarely the same thing.
- The pages carrying the result. In technology companies this is almost always a documentation page or a regulatory explainer, essentially never the product overview the marketing team maintains.
- Domains sharing your term field. Listed with Domain Authority and the count of shared terms. In this market they are overwhelmingly trade publications, comparison sites and industry associations rather than competing providers.
- A ranked list across the whole estate. For companies running a product site plus a documentation subdomain plus a corporate site, one table instead of three separate reports.
What follows from the fourth point should be decided deliberately rather than by instinct. Going directly against a trade publication on a broad regulatory term commits several years to an uncertain outcome. Being cited by that publication produces qualified traffic within weeks. With a content team of one or two people, the citation route is the one that can actually be justified in a planning meeting. It also rules nothing out: narrower terms with a smaller authority gap remain worth a page of their own.
The regulatory explainer
Written once in German, updated when a rule changes, cited by anybody explaining the same regime.
- Accumulates references over years
- Read by the people who approve purchases
The translated feature page
Rewritten with every product change, never quite current, identical in structure to the English original.
- Never accumulates anything
- Consumes review time each release
The contrast is worth sitting with, because it points at where a small team should spend its German-language budget. A regulatory explainer costs one substantial writing effort and then a light annual revision. A translated feature page costs something every single release and reaches a group that was reading the English version anyway. Over three years the first accumulates history and references while the second stands still — and the running cost is the wrong way round.
Half your enquiries are invisible
In business-to-business technology sales, a large share of first contact happens through a channel no analytics package attributes: an email to a named person, a message on a professional network, an introduction at a conference. The person read three of your pages first. Nothing records that.
The form submission
Recorded, attributed, and in this segment usually the minority of genuine enquiries.
- Easy to count
- Systematically understates the site
The direct approach
Email or a message to a person by name, after reading. Invisible unless somebody asks how the contact arose.
- One question in the first call
- Written into the enquiry record
The remedy is a single field in whatever system records enquiries, filled in during the first conversation. It is tedious, easily forgotten, and it is the only bridge between what the reporting shows and what the company decides. Without it, a team argues for a year about whether German content is worth writing while the evidence sits unrecorded in a sales inbox.
Companies that maintain that field for twelve months end up with something no analytics package can produce: a list of enquiries where somebody named a page. Companies that do not end up arguing from impressions, which is an argument nobody wins. In a segment where a single enterprise contract can exceed a year of content spend, the difference between those two positions is not academic — it decides whether the German pages get written at all.
A review cadence for a small content team
| Cadence | What is reviewed | What it decides |
|---|---|---|
| Weekly, 15 minutes | Keyword dynamics, entries and exits | Fix technical losses, link recent gains internally |
| Monthly | Pages, plus country crossed with language | Whether German demand is growing or assumed |
| Quarterly | Competing domains and consolidated ranking | Which German pages to commission |
| Annually | Enquiries with recorded origin, year on year | Whether the effort paid for itself |
The weekly slot has two jobs and no others: spotting a dropped position early enough that the reason is still traceable, and registering a term that has climbed on its own — the moment at which an internal link from a strong page yields the most. Clicks, impressions and average position stay out of that slot deliberately, because at this volume they fluctuate for reasons nobody controls.
One addition worth making to the monthly slot for a bilingual property: check whether the two language trees are still pointing at each other correctly. Language annotations break silently during platform migrations, and the symptom — the German page vanishing from results while the English one remains — looks exactly like a content problem and is not one. It is a five-minute check and it prevents a quarter of misdirected effort. The workspace shows index coverage next to positions, which makes the disappearance visible rather than assumed.
Writing German pages rather than translating them
The recurring recommendation in this market is uncomfortable because it costs more: the pages that matter should be written in German, not translated into it. But the reasoning is narrower than it sounds, and it applies to a specific and small set of pages.
What to translate and what to write
Deciding this once removes most of the recurring argument about localisation budget.
- Translate the technical reference. Endpoints, parameters, error codes. The audience reads English anyway, and a translation here is a courtesy rather than a growth measure.
- Write the regulatory material in German. Licensing, supervisory regime, safeguarding of funds, data location. This is read by people who work in German and whose questions have no English-language equivalent on your site.
- Write the integration narrative in German. Not how the API works, but how a German institution actually adopts it: who signs off, what the procurement steps are, how long it takes.
- Leave the product marketing bilingual and thin. It performs least well in both languages and consumes the most review time. That is an uncomfortable observation and it holds across this segment.
The second and third points share a property that makes them cheap: neither requires the product team. The regulatory and adoption material is known to compliance and to whoever runs enterprise sales, and it exists already in slide decks and answers to due-diligence questionnaires. Turning that into public pages is an editing task, not a writing task.
Making the German number legible to an investor
A report going to a board or an investor has a different job from one used by the person doing the work. It will be read by somebody comparing this territory against others and without any feel for German search behaviour.
Keep it to figures with an attached decision: enquiries with recorded origin year on year, the German share of demand crossed with query language, the three pages carrying the result, entries and exits for the period, and a sentence of explanation. Full term lists, competing domains and country breakdowns belong in the export, where the working level can use them. Reports carry your own logo and colours, which matters more than it should — a document that looks like a company document gets forwarded, and a tool export gets filed.
The underlying campaign work — term selection, link building, page recommendations — sits in the same place as the reporting inside My SEO, which removes the step of exporting one system to compare it against another. How the initial German term list is assembled is described in our approach for keyword research, the editorial half in content strategy, and the technical prerequisites under technical SEO.
Open your German property in the dashboard
Questions from product and content teams
Our audience reads English. Do we need German pages at all?
Part of it reads English — the part that builds things. The part that approves purchases, assesses risk and signs contracts largely does not, and those are the people who stop a deal. Before deciding, cross the country breakdown with query language and look specifically at the Germany-plus-German-query cell. If it is empty, that may reflect genuine absence of demand or the absence of any page for that demand to land on, and only writing one page will tell you which.
Why does our English documentation outperform everything else?
Because it is older, maintained continuously and linked from every release note, which are the three things that make a page strong. This is worth understanding rather than resenting: it means the mechanism works and has simply never been applied to the German tree. A German page that is updated quarterly and linked from current material behaves the same way within a year.
Should we put German content on a subdomain or in a folder?
A folder on the main domain, in almost every case at this size. A subdomain starts with none of the accumulated history of the main site, and for a company with one strong domain that is a considerable amount to give up voluntarily. The exception is where a separate legal entity requires a separate presence — and that is a licensing decision rather than a technical one.
Our figures swing wildly week to week. Are we doing something wrong?
Probably not — you are reading small numbers at high frequency. At thirty clicks a month one visitor moves the percentage substantially. That is a property of small samples rather than a fault in the data. The practical answer is to lengthen the window: quarters instead of months, and year-on-year instead of month-on-month, with weekly attention reserved for entries and exits alone.
A trade publication holds every term we care about. What now?
Check the shared-term count and the Domain Authority gap first. Where the gap is large, that term is not winnable within any planning horizon you have, and the honest response is to spend the budget on terms with no such incumbent. Frequently the publication is not a competitor at all but a channel — a contributed article or a citation produces qualified readers far sooner than a page competing against it ever would.