Somebody else is writing your employer description

Leipzig SEO editorial team21 August 2026updated 31 August 2026

Ask what it is like to work at your company and see where the answer comes from. Almost certainly not from you. It will be assembled out of anonymous reviews, a few forum threads and whatever a job portal wrote in its company profile — none of which you commissioned, and all of which is now the description a candidate carries into their decision.

Employers have largely accepted that review platforms exist and that they cannot control what appears there. What has changed is that those platforms are no longer one source among several that a candidate reads and weighs. They are increasingly the raw material from which a single paragraph gets built, and that paragraph is read instead of everything else.

The description you did not write

A candidate at the start of a search does not look up your careers page. They ask a general question — is this a good place to work, what do people say about the management, is the pay competitive — and those questions have no canonical source. Where none exists, the answer gets assembled from whatever is findable, and what is findable is almost always the review platforms.

The bottom two rows are the interesting ones. Nobody outside the company can answer them, so nothing gets assembled — the question simply goes unanswered, or gets answered with a generic statement about the country rather than about you. For an employer competing for international candidates, that is an open field with no incumbent in it.

What an assembled paragraph is built from

A ranked list puts whole documents in order using a blend of topical fit and standing accumulated over years. Assembling a paragraph works differently: individual assertions are lifted out of individual pages and written into continuous prose. What gets picked is a sentence rather than a document — often just the words surrounding a fact somebody can use.

Careers pages consist almost entirely of the right-hand card, and that is not carelessness. Read from top to bottom the wording is perfectly acceptable and it is what recruitment copy has always looked like. It is simply unusable by anything that takes one sentence and discards the rest — which is a growing proportion of how candidates form a first impression.

The test is quick and mildly humbling. Take any paragraph from your careers section, remove it from its surroundings, and read it alone. If it still means something to somebody who has read nothing else, it can be used. If it needs the sentence before it, or if removing your company name would leave a paragraph that could belong to any employer in the country, it cannot.

Applying that test across a careers section usually produces the same result: three or four sentences survive out of several pages, and all of the survivors contain a number. That observation is more useful than any amount of advice about tone, because it identifies precisely what to write more of.

What can actually be done about the review platforms

The distinction between the two blocks is not about tone. It is about whether a sentence contains something that can be checked against reality. An anonymous review saying management is unresponsive cannot be countered by a page saying management is responsive — but it can be sat alongside a page stating that every applicant receives a decision within ten working days, because one of those is a fact and the other is an opinion.

There is a limit worth naming, because it marks where effort pays. Questions with an obvious authoritative source — statutory notice periods, national holiday entitlements, immigration rules — will be answered from that source, and nothing you write displaces it. The productive ground is everything specific to your company and your city: how your process runs, what your ranges are, which of your roles work in which language. That is the material nobody else holds, and it is also the material employers are least accustomed to publishing.

Finding out what is currently being said

No report covers this. What exists is running the questions yourself and writing down what comes back — about a quarter of an hour a month, which almost nobody spends, and that is precisely why it produces an advantage.

• Ten questions a candidate would genuinely ask. Including the two or three about relocation and language that only you can answer, because those are where the field is empty.
• Identical wording every month. Otherwise this month cannot be compared with last month and the whole exercise yields nothing.
• Write down what was said, not just whether you appeared. The description matters more than the mention. A confident wrong statement about your pay does more damage than absence.
• Note which employers were named instead. Over a quarter this becomes an inventory of who in your region has published the more specific material.

Filing these observations alongside the ordinary reporting in one workspace is the single thing that keeps the routine alive past week eight, and week eight is where it usually dies.

Why one kind of material is drawn on so heavily

Reference works appear inside composed paragraphs at a frequency wholly disproportionate to how much of the web they occupy. The explanation is structural rather than mysterious. Such material is broken into headed sections, it carries a dense web of internal cross-references, and the style it is written in yields sentences capable of standing alone. As a result a slot there pays back more than the same slot bought anywhere else. Nor is the price steep: ten dollars per slot, and the only quantities on offer are nought, one, five and ten.

There is no option in between. Decide that seven feels right and the choice collapses back to five or ten, separated by six hundred dollars over twelve months. An employer this size should open with one: a hundred and twenty dollars answers the question that has to be answered before any other.

Where a bigger number changes nothing

A different principle governs the other option. Each network placement is priced at one dollar and the quantities available are nought, twenty, a hundred and five hundred. All that figure captures is how many placements come into being; nothing whatever about the standing of the sites that carry them is implied. Choose the top quantity and five hundred of them will exist. What will not exist is an effect five times the size of the hundred quantity.

Of the two, only the reference-work option speaks to the aim set out in this piece. What the network quantity governs is a site's weight inside an ordinary results list; it has no bearing whatsoever on whether a sentence can be extracted for use in a paragraph. Listing the two separately within the same account is what stops them being merged into a single figure when somebody reviews the budget.

That is no argument against network placements as such — they serve a separate purpose perfectly well. It is a warning about a budgeting habit that recurs everywhere: a company decides it wants to be described accurately to candidates, then commits the whole allocation to whichever option produces the larger figure in a monthly summary, which is precisely the option with the least connection to that aim. Reviewing both justifications side by side inside a single environment costs nothing and makes the year-end conversation possible.

The four pages that cover most of it

None of the four requires a new tone of voice or an agency. All four require somebody to write down what already happens, in sentences that contain figures. That is uncomfortable in a way that culture copy never is, and it is the entire difference between material that gets quoted and material that gets skipped.

None of the above matters if the text stays behind

When a page constructs itself inside the visitor's browser after the load completes, there is every chance that nothing beyond that browser ever handles its text as text. This applies with particular force to careers areas hosted inside a recruitment platform or thrown together on a modern site builder — for those, it is simply what happens unless somebody decided otherwise on purpose. Internally there is no clue at all: every monitor in the building shows a page that looks exactly as intended.

A technical review establishes in a morning whether the text actually ships. Which of these questions are genuinely being asked in your region comes from keyword research based on measured volume rather than assumption, and turning the answers into pages that inform rather than advertise is what content strategy covers. Handling all three from a connected account keeps them from being pursued separately and arriving at conclusions that cannot all be true.

A closing word on expectations. None of these four pages produces applications in the following month. What they change is the description a candidate carries into a decision they will make two or three quarters later, and in a labour market where good people move every three or four years, the effect surfaces correspondingly late. Judged after a quarter against the number of applications received, this work will always look like a failure. The measures that mean something early are different ones: whether the new pages were catalogued at all, whether they generate impressions, and whether the monthly readings show the description shifting away from the review platforms and towards material you wrote.

Check how your company is currently described

Questions from recruiting and leadership