Faragiri

Learn How links count Off page

A Shropshire basket maker had 214 links, 31 referring sites, and 4 that had ever sent a visitor, and the last number is the one worth reporting.

5Sections
4Min read
4Depth

Count referring domains, not links, and then count the ones with readers. Wenlock Willow makes baskets in Shropshire and its link report shows 214 links, 31 sites, and 4 that ever sent a visitor. The first number is the one every tool leads with and the last one is the only one that has ever predicted anything at the workshop.

The 214 collapse fast. Two directories account for 141 of them, one on every page of their own sites, and eleven of the 31 sites are the same regional listing network under different names. What is left is 12 real sites and four that people actually read.

What makes one count more than another

A link from a page about the same subject beats one from a page about everything. Wenlock's best link is from a page about hedgerow crafts on a countryside charity's site; its worst are from pages listing every business in a county. The first is a recommendation and the second is a phone book.

Link quality

A link's weight depends on whether the page shares a topic, whether a human wrote surrounding words, and where on the page it sits.

Signal strength by placement and context

Crowded generic listEditorial topic match
  1. Footer with ninety linksLow editorial intent and crowded with unrelated targets.
  2. County business directoryPhone book listing with no surrounding prose.
  3. Supplier list on a craft siteRelevant domain, though only other names sit beside it.
  4. Paragraph on hedgerow craftsWritten recommendation inside a page about the exact subject.

Exact weighting is unpublished, but the direction between a crowded list and an editorial paragraph is clear.

A link in a paragraph beats a link in a list of suppliers. Somebody wrote the paragraph, which means somebody decided, and the words around a link tell an engine what the link is about. A list entry has no words around it except other names.

Position on the page and the number of other links beside it both matter. A link in the third paragraph of an article with four links is not the same object as the same link in a footer with ninety. Nothing published says how much difference that makes, and the direction is not in doubt.

The click, and what it is evidence of

A link nobody clicks can still count, and a link nobody could click cannot. That distinction is the useful one: a link on a page a reader might plausibly reach is doing something even if nobody clicks it this month, and a link on a page no human has visited since 2019 is not doing anything at all.

Counting

Each band removes a way of overcounting, and the last one is the only figure at this workshop that has ever predicted anything.

  1. Links reportedTwo directories account for 141 of them, one on every page of their own sites.214
  2. Referring sitesEleven of these are one regional listing network under different names.31
  3. Sites a person might plausibly reachPages with readers, whether or not anybody clicked the link.12
  4. Sites that ever sent a visitorBetween 20 and 60 people a month between them.4

The last band undercounts on purpose: a good link on a page people read may send nobody in a given month. It is a floor rather than a measure of value.

The four links that sent visitors are the honest evidence, and they send between 20 and 60 people a month between them. Referral traffic is not what a link is for and it is the only sign available that anybody reads the page a link sits on.

Where this stalls: the score that can be bought

Domain authority scores are estimates sold by the companies that invented them, and they are not a search engine's numbers. They are computed from a crawl a third party owns and they move when links are bought, which is why a year can go into moving one while nothing about what the site answers has changed.

Vanity metric

Third-party authority metrics react quickly to paid directory listings because they come from a vendor's own crawl rather than a search engine.

Third-party authority estimate

Before spendAfter 400 pounds in 2024

  1. Wenlock's vendor score1119Recalculated by the tool company; moved zero positions on any query.

The score is real only in the sense that a private company really updated its own formula output after 400 pounds were spent.

The stall is that the score is the only number in the whole subject that responds quickly to money. Wenlock's went from 11 to 19 in 2024, which cost 400 pounds and moved no position on any query, and the improvement was real in the sense that a real company really did recalculate it.

Two smaller facts

Nofollow is a hint now rather than an instruction, which changed in 2019 and is still described as a blocking rule in a great deal of advice. A nofollow link may be counted or not, at the engine's discretion, and treating one as worthless is a mistake in the same direction as treating it as equal.

Rules of thumb

Treating nofollow as a hard block or repeating identical anchor phrases across directories misses how engines treat both patterns.

Common assumptionOld advice and bulk tactics
What actually happensHow search engines read it
Nofollow linksStrict block worth nothing at allA hint since 2019, weighed at discretion
200 identical anchors200 independent votes for a phraseOne unnatural claim repeated across sites

Real people writing sentences do not repeat the exact phrase 'handmade willow baskets Shropshire' two hundred times.

The same anchor text on 200 links is one claim, repeated, and it is a claim nobody writing naturally would make. Wenlock's directory links all read handmade willow baskets Shropshire, which is not a phrase any person has ever typed into a sentence, and identical anchors across many sites are a pattern rather than a set of independent recommendations.

What to open

The Links report in Search Console lists the sites linking to you from the index rather than from a third party crawl, free, and it replaced asking a tool that only knows what its own crawler happened to find. Its top linking sites list is the 31, and it is the closest thing to a first party answer that exists.

First-party checks

Skip third-party crawls and pair Google's own index list with referral traffic to see which links exist on pages people visit.

  • Search Console Links
  • GA4 Referral source
  • Search Console LinksShows the 31 top linking sites straight from Google's index, free.
  • GA4 Referral sourceReplaced Universal Analytics in 2023 and surfaces the 4 sites sending visitors.

If a site inside the 214 reported links never shows up in GA4 referrals, it is worth asking whether any human reads that page.

Google Analytics 4 supplies the four, free, replacing Universal Analytics, which stopped processing data on 1 July 2023. Referral source is the report, and a link that has never appeared in it is a link on a page nobody reads.

How the four were earned rather than bought is earning links without buying them, whether the 141 directory links are worth doing anything about is what actually harms, and what to leave alone, and the mentions that carry no link at all are being named, with or without a link. All of it is the signals that come from other sites, and 214 is the number that will be in the report.

Above this