7 German brands. 3 million internal links. 82.7% go nowhere new.

Why internal links decide what ranks

Search engines find pages by following links, and they decide how much each page is worth largely by counting and weighting the links that point to it. External links carry authority into a domain; internal links decide where that authority goes once it arrives. Every internal link is a small instruction — this page is worth reaching, and some of my standing should flow to it. Multiply that across a whole site and you get an internal PageRank distribution: a map, drawn by the site itself, of which of its own pages matter.

The decisive variable is where a link sits. A link inside body copy is an editorial judgement — a human decided this page is relevant to that one. A link baked into a navigation bar, header or footer is a template decision: it appears on every page, unchanged, regardless of context. Both pass equity, but only the editorial link carries a topical signal. Search engines have learned to discount boilerplate heavily for exactly this reason — when the same sixty links repeat on fifty thousand pages, they say almost nothing about any individual page.

This is not a new idea, but it is an underused one. PageRank — the algorithm Google was founded on — is at heart a model of how authority flows through links, and although the public ranking system has grown vastly more complex, the internal principle still holds: a page that is linked often, from pages that are themselves well-linked, is treated as more important than one that is barely linked at all. What has changed is the stakes. As AI answers and generated overviews increasingly summarise a category by naming a few entities, the clarity of a site's internal structure — which pages it treats as canonical, how its products relate to its content — is part of how machines understand a brand at all. A muddled internal graph is no longer just an SEO inefficiency; it is a comprehension problem.

We wanted to know how the typical well-run brand site actually behaves on this axis — not in theory, but in its live link graph. So we crawled seven recognizable German brand websites, spanning a small confectionery site of under a hundred pages to a drugstore retailer of more than fifty thousand, reconstructed every internal link, and measured the same things on each. The brands were chosen for recognizability and for spread across sectors (confectionery, beverages, baby and personal care, organic retail, drugstore) and platforms (Shopify, TYPO3, headless and bespoke CMSs), so the pattern we report is not an artefact of one vendor's defaults.

What follows is a reading of 3,031,977 internal links across 60,073 pages. The pattern is remarkably consistent — and, for anyone who cares about organic visibility, uncomfortable. We have published a separate, deeper teardown for each individual site; this report is the cross-brand view, the part that only becomes visible when you line seven graphs up next to each other.

Five in six internal links go nowhere new

Of the 3,031,977 internal links we mapped, an average of 82.7% are boilerplate — navigation, header and footer — and only 17.3% are editorial, placed inside the body of a page. Put plainly: for every six internal links these brands publish, five repeat the same menu on every page, and roughly one expresses an actual relationship between two pieces of content.

This is not one bad apple dragging down an average. Every single site in the study is boilerplate-dominated. The share runs from 68.4% at the most editorial site to 92.2% at the most templated, and four of the seven sit above 80%. The healthiest result we found, bahlsen.com at 68.4%, would still mean two of every three links are doing no topical work.

Why is it so uniform? Because it is the default behaviour of a content-management system. A global mega-menu plus a fat footer can easily contribute forty to eighty links to every page; a typical product or article page adds only a handful of in-content links, if any. As the page count grows, the boilerplate compounds while the editorial layer does not — so the bigger and more templated the site, the more lopsided the ratio becomes.

There is a second, quieter cost beyond equity: discovery and context. Editorial links are how a crawler learns that a recipe relates to three specific products, or that a guide belongs to a particular category. A boilerplate link cannot carry that meaning, because it is identical on every page and therefore context-free. A site leaning almost entirely on its menu is asking search engines — and, increasingly, the language models behind AI answers — to infer its catalogue's structure from a navigation list alone, and then to guess the relationships the content never states. Some will guess well; many will not.

The SEO consequence is a quiet loss of control. A site with a real editorial layer can deliberately push equity toward the products or campaigns it most wants to rank, by linking to them from relevant articles. A site that is 85–90% boilerplate cannot: the template distributes authority almost uniformly to whatever the menu and footer happen to point at, and the brand's actual priorities never enter the calculation.

What would a healthy ratio look like? There is no single right number — a thin brochure site will legitimately sit higher than a content-heavy publisher — but the direction of travel matters more than the absolute. A site that consciously interlinks its content tends to land somewhere in the 50–70% boilerplate range; once you are above 85%, the editorial layer has effectively stopped existing as a signal. On that scale, only one of our seven brands is even close to healthy, and the rest are well into the zone where the template is the only voice the graph has.

Authority pools in one percent of pages

If boilerplate decides that equity is spread thinly, PageRank concentration decides where the thin layer pools. Across the seven sites, the top 1% of pages hold an average of 20.6% of all internal PageRank. But the average hides the more interesting story, which is about size.

On the small sites, authority is relatively even — fritz-kola's top 1% holds just 3.2%, katjes' 10%. On the two largest sites it becomes extreme: alnatura's top 1% (41 of 4,143 pages) holds 48.4%, and dm's top 1% holds 54.8% — a single percent of pages absorbing more than half of the entire domain's internal authority. The mechanism is simple: the more pages a site has, the more times its global navigation repeats, and the more overwhelmingly that navigation's targets dominate the graph. Scale does not dilute concentration; it intensifies it.

The mechanism is a rich-get-richer loop. A page that sits in the global navigation receives a link from every other page; that makes it high-PageRank; and because PageRank flows proportionally, its own outbound links then carry more weight, reinforcing whatever it points at. Nothing in this loop references relevance or revenue — it is pure topology. On a fifty-page site the loop is gentle; on a fifty-thousand-page site it is a gravitational well, which is exactly why alnatura and dm show concentrations near and above 50% while the small sites do not.

That would be fine if the concentrated authority landed on the pages that matter commercially. It rarely does. The pages that win this lottery are whatever the template links most — and templates link to hubs, service pages and legal notices, not to the long tail of products and articles that actually capture demand. The result is a site that looks authoritative in aggregate but starves its own money pages of internal support.

The practical test is simple and worth running on any site: list your top 1% of pages by internal PageRank and ask how many you would actually pay to rank. If the answer is "the homepage, a store locator, and the privacy policy," your strongest internal signal is being spent on pages no customer is searching for — and the products that pay the bills are competing for whatever is left.

92,684 internal links into redirects

A redirect is not free. When an internal link points at a URL that 301s or 302s to another, the crawler has to make an extra hop, and a measurable fraction of the link's value is lost on the way. One such link is a rounding error. We found 92,684 of them.

Across the seven sites, an average of 5.1% of internal links point at a redirect rather than the final URL. The distribution is heavily skewed: dm alone accounts for 90,107 of the 92,684 — about 97% of the total — because at fifty-thousand-page scale, a single templated link to a redirecting URL is multiplied across the whole site. veganz shows the same disease in concentrated form: 1,354 links, 23.2% of its entire internal graph, point at redirects, because its primary navigation links to URLs that immediately 301.

How does it happen? Almost always through change over time. A site migrates from non-trailing-slash to trailing-slash URLs, switches http to https, restructures a category, or rebrands a product line — and sets up redirects so old external links keep working. That is correct. What gets forgotten is the internal links: the menu, the footer and the in-content references still point at the old URL, which now 301s. The redirect quietly does its job for years, and nobody notices the site is paying a small tax on every internal click.

None of this is visible from the front end. The menu works, the pages load, nothing looks broken to a visitor. It only surfaces when you crawl the site the way a search engine does and check where every link actually resolves. That is the difference between a site that looks fine and a link graph that is quietly taxing itself on every internal hop. The fix is unusually high-leverage: because the offending links are templated, repointing them at their final destination once corrects them across every page at the same time.

When the imprint outranks the product

Here is the finding that tends to stop marketers mid-sentence. On two of the seven sites, a legal or administrative page outranks every product the brand sells — by the site's own internal PageRank.

On bahlsen.com, a corporate Bahlsen-Family hub and legal pages sit at the top of the internal-authority ranking, above the biscuits. On ruegenwalder.de, the single most authoritative URL on the domain is the careers page (/de/ueber-uns/karriere) — ahead of every product and every recipe. And while dm.de's flag is technically clean, its most-linked page is a payment-information help page with hundreds of thousands of inbound internal links — not a product, not a category.

This is the signature of a footer-dominated architecture. Imprint, privacy, contact, shipping, careers and FAQ pages are linked from the footer of every page, so they accumulate more internal links — and therefore more internal PageRank — than almost anything the business actually wants to rank. Nobody searches for an imprint. Yet on these sites the imprint, or the careers page, is treated by the brand's own link graph as its crown jewel. The equity is pooling exactly where it delivers no commercial value.

It is worth being precise about what "outranks" means here, because it is not hypothetical. We are not saying these legal pages rank higher in Google for commercial queries — they do not. We are saying that the brand's own internal-authority signal, the thing it controls and uses to tell search engines what matters, places a careers page above its entire product range. That is the brand mis-instructing the search engine, every day, on every page. The product pages then have to overcome their own site's vote before they even begin competing externally.

What separates the least-bad from the worst

Step back from the individual numbers and a few cross-brand patterns emerge. First, boilerplate dominance is the rule, not a failure state — it is what every one of these sites does by default, and the variation between them is a matter of degree, not kind. Second, the problems scale with the site: the largest properties (alnatura, dm) show the most extreme PageRank concentration and, in absolute terms, the most wasted equity, because every templated mistake is multiplied by the page count.

Third, the sites that come off least badly share one trait: a genuine editorial layer. bahlsen.com, the most editorial site at 68.4% boilerplate, is the same site whose body content links out enough to lift its editorial share to 31.6% — roughly double the field. The lesson is not have fewer footer links; legal links are unavoidable. It is have more editorial ones, so the template stops being the only voice in the graph.

Finally, breakage and decay travel together with neglect: the sites with the most redirect links and the most internal 404s are also the ones whose navigation was clearly never audited against a crawl. Internal linking is the one ranking factor a site owner controls completely — no outreach, no third parties, entirely inside the CMS — and it is precisely because it is invisible from the front end that it is so consistently left to rot.

There is also a useful distinction the data forces: distribution problems versus breakage problems. alnatura and dm are internally clean — almost no broken links — yet they have the most extreme concentration; their authority is intact but misrouted. veganz and ruegenwalder are the opposite: smaller authority, but actively leaking it into redirects and 404s. The remedies differ. A distribution problem is solved by adding editorial links that pull equity toward priority pages; a breakage problem is solved first by repointing links at live, final URLs. A good audit tells you which disease you have before you start prescribing — which is why reading the graph has to come before touching it.

What it means for SEO and marketing teams

For an SEO or marketing team, the benchmark reframes a familiar problem. The instinct, when rankings stall, is to chase more external links or more content. But these seven sites suggest the cheaper, more controllable lever is usually internal: most brands are not short of pages or authority — they are misrouting the authority they already have. If 83% of your links say nothing and your most-linked page is your imprint, no amount of new content will rank predictably, because the site has no mechanism to send equity where it is needed.

The economics favour internal work, too. A backlink campaign is slow, expensive and outside your control; a content push takes months to compound; but repointing redirect links, adding contextual links and re-routing equity are changes a team can ship in days, on its own, and measure. For most brands the ceiling on organic performance is not a shortage of authority — it is the misallocation of the authority they already have.

There is also an AI-visibility dimension that did not exist a few years ago. When a model is asked to recommend a brand or compare a category, it leans on how clearly the web — including a brand's own site — describes and connects its entities. A site whose internal graph makes its product relationships explicit is easier for a model to represent correctly; one that hides everything behind an undifferentiated menu is easier to summarise wrongly or omit. Internal linking is quietly becoming part of how machines, not just rankers, understand a brand.

The good news is that internal linking is fixable without a redesign, a migration, or anyone else's cooperation. It is data work: crawl the site, see where the equity actually flows, and add the contextual links that the template never will. The teams that treat their internal-link graph as a deliberate routing system — rather than a by-product of their navigation — are the ones whose priority pages rank without a fight.

The site that links to itself

Seven brands, three million links, one consistent story: the modern brand website is a broadcasting machine for its own navigation. Boilerplate dominates, authority concentrates on hubs and legal pages, and a surprising share of links quietly leak value into redirects — all of it invisible until you read the graph. None of these brands is badly run; they are simply doing what their CMS does by default, at scale, where the defaults hurt most.

That is also the opportunity. Internal linking is the rare growth lever that is entirely in-house, low-risk, and measurable. The brands that map their link graph and route equity on purpose will quietly out-rank the ones that leave it to the footer.

A closing note on method, because it is the actual point. None of these findings required privileged access, a tool integration, or the brands' cooperation — every number here was computed from a public crawl and a reconstructed link graph. Any of these seven companies could have run the same analysis on themselves at any time and seen exactly what we saw. That they apparently had not is the most repeatable finding of all: internal structure is treated as plumbing, installed once and never inspected, when it is in fact one of the most direct levers a brand has over its own visibility. The benchmark above is a mirror; the only question that matters for any brand reading it is whether its own map looks meaningfully different — and the only way to know is to look.