5 Sites, 3,671 Internal Links, and 2% Editorial at Best

Every internal-linking guide ends with the same instruction: add more internal links. On 2026-09-07 I checked whether that was actually the problem. I ran five sites through the free analyzer on my own SEO tool and looked at one thing — not how many internal links each site had, but what kind.

None of them was short on links. Between them, 3,671 internal links on the crawled pages. The share classified as editorial — a link written inside body copy, pointing somewhere specific — topped out at 2%. On three of the five it rounded to zero.

What I ran

No account needed, so this is reproducible:

• POST https://rankforge.cc/api/analyzer with {"seed_url": "..."}, then poll GET /api/analyzer/<id> until status is completed.
• Anonymous tier, which caps each crawl at 10 pages. Every run came back with crawl_complete: false.
• Five seeds, all run on 2026-09-07: ahrefs.com/blog, css-tricks.com, kubernetes.io/docs/home, rankforge.cc, www.thestackanalyst.com.

The 10-page cap matters, so I'm putting it up front rather than in a footnote. This is a sample of the pages nearest each seed — the homepage or hub, plus what it links to first. It is the top of a site, not the whole site.

That cuts in an interesting direction, though. The top of a site is where internal linking is supposed to be strongest and most deliberate. It's the friendliest place to go looking for editorial links, not the harshest.

The classifier sorts every internal link into four buckets: editorial (inside body content), navigational (menus, headers, breadcrumbs), list (archive and card grids), and boilerplate (footers and other site-wide furniture).

The numbers

Site	Internal links	Editorial	Navigational	List	Boilerplate	Nav-sat.
ahrefs.com/blog	1,276	0%	35.6%	31.2%	33.2%	yes
css-tricks.com	59	0%	83.1%	16.9%	0%	no
kubernetes.io/docs	1,335	0%	95.4%	4.6%	0%	yes
rankforge.cc	503	0.6%	14.3%	9.9%	75.1%	yes
thestackanalyst.com	498	2%	74.7%	22.7%	0.6%	yes

Four of five sites got flagged for weak internal linking. All five had plenty of links. None had many editorial ones.

Not one of these sites had an orphan page in the crawled sample — all five report orphan_pages_count: 0. Four of the five still got "Internal Link Strategy" flagged as their primary bottleneck.

Navigational links can't tell your pages apart

This is the part that changed how I read internal-link reports.

A menu link appears on every page and points at the same handful of destinations. In a PageRank-style model, that's a dense, near-uniform block of edges. It moves authority around, but it moves it the same way for everybody, so it can't express a preference. It has no opinion about which of your pages deserves to rank.

Editorial links are the only ones that carry an opinion. Somebody writing a paragraph decided this page was worth pointing at from that sentence. That's the signal.

When the navigational block dominates, the analyzer sets nav_saturated: true — and then warns you about its own output:

Navigation/footer links reach nearly every page, so authority differences between pages are largely structural (a byproduct of the menu), not a reliable signal of which pages are strategically under-supported.

I like tools that do this. A per-page "authority score" on a nav-saturated site is mostly measuring your template. Ranking your pages by it and "fixing" the bottom five is chasing a menu.

What it looks like on my own site

My own site came out worst on composition: 75.1% boilerplate, 0.6% editorial. That's a footer doing the work that body copy should be doing.

The analyzer proposed 7 specific in-body placements, each with a source page, a target, an anchor, and the sentence to put it in. It also simulates the result:

Metric	Value
Links modeled	7
Total authority before	922.34
Total authority after	922.34
/docs/authority-flow before	22.34
/docs/authority-flow after	77.66 (+247.7%)
Pages gaining	1
Pages losing	9

The two numbers I'd underline are the identical totals. Internal linking creates no authority. It's redistribution: one page went up 55 points, nine went down an average of 6.2 each. That's a fine trade when you know which page you want to win, and a pointless shuffle when you don't.

/docs/authority-flow was sitting at 22.34 on a site whose median page scores 98.93. It was reachable, indexable, in the footer — and structurally invisible.

The outlier worth looking at

Kubernetes docs is the extreme case: 95.4% navigational, and the docs sidebar is doing essentially all of the linking. Its homepage holds 49% of the internal authority in the sample and the median page sits at 11.57.

That is not a bug in their docs. A giant navigation tree is the correct interface for reference documentation, and their readers arrive from search on a specific page anyway. It's a good reminder that "editorial links are low" is a finding, not automatically a defect — the question is whether the pages you actually care about are getting any differentiation at all.

Check yours in about two minutes

# 1. queue a crawl (anonymous, no signup)
ID=$(curl -s -X POST https://rankforge.cc/api/analyzer \
  -H 'Content-Type: application/json' \
  -d '{"seed_url":"https://example.com/"}' | jq -r .id)

# 2. poll until it finishes
until [ "$(curl -s https://rankforge.cc/api/analyzer/$ID | jq -r .status)" = "completed" ]; do
  sleep 10
done

# 3. the only number this post is about
curl -s https://rankforge.cc/api/analyzer/$ID \
  | jq '.summary_metrics.link_quality | del(.pages_with_only_boilerplate_links)'

Real output from the thestackanalyst.com run:

{
  "editorial_pct": 2,
  "navigational_pct": 74.7,
  "boilerplate_pct": 0.6,
  "list_pct": 22.7,
  "total_internal_links": 498
}

The browser version of the same crawl lives at rankforge.cc/audit if you'd rather not shell out for it.

What I'm doing about it

Three changes, in the order I'd rank them:

1. Stop treating link count as the metric. 503 internal links and 0.6% editorial is a worse position than 60 links with 20 of them written on purpose.
2. Read nav_saturated before reading any per-page score. If it's true, per-page authority deltas are template noise and the ranked "weak pages" list is not a to-do list.
3. Write the links into the prose. Not a related-posts widget, not another footer column — a sentence that earns the link. That's the only bucket a template can't fake for you.

Run it on your own site and drop your editorial_pct in the comments. Mine was 0.6%, and I build SEO tooling for a living — I'd like to know whether I'm the outlier or the norm.

---

Crawls run 2026-09-07 via the public RankForge analyzer API, anonymous tier, 10 pages per site. Raw JSON for all five runs is saved alongside this post; every figure above comes from summary_metrics in those files.