What internal linking looks like across South West businesses
Most writing about internal linking, including the version of this article that stood here until recently, explains what internal links are and asserts that they matter. That is easy to say and impossible to check.
So here is what we actually found. As part of the Local Digital Visibility Index we parse the homepage of every business we measure and record whether it links to three things: an about page, a team page, and a credentials page. Those are internal links, and they are the cheapest possible test of whether a site was built for people who want to know who they are dealing with.
Across 1371 businesses in 34 South West indices, 1222 homepages parsed cleanly. Every figure on this page is computed from the published datasets when the page is built, so it tracks the index rather than describing some earlier version of it.
The finding
Section titled “The finding”| Internal links on the homepage | Firms | Median Digital Visibility Score | Median Visibility pillar |
|---|---|---|---|
| 0 of 3 | 252 | 46 | 0 |
| 1 of 3 | 544 | 50 | 4 |
| 2 of 3 | 325 | 57 | 21 |
| 3 of 3 | 101 | 60 | 28 |
Every step up adds points, with no reversal. Between the bottom and top groups there is a 14-point gap in overall score and a 28-point gap in search visibility.
But medians are the weakest way to put this, because they hide how little the groups overlap.
Read along the bottom row and then the top one. The worst-performing firm with all three links scores 47. Of the 252 firms with none, 113 manage to beat it. The quartiles move together too, not just the middle:
| Links | Lowest | 25th | Median | 75th | Highest |
|---|---|---|---|---|---|
| 0 of 3 | 18 | 41 | 46 | 51 | 75 |
| 1 of 3 | 31 | 46 | 50 | 56 | 82 |
| 2 of 3 | 38 | 52 | 57 | 64 | 87 |
| 3 of 3 | 47 | 54 | 60 | 69 | 89 |
The number that should worry you most
Section titled “The number that should worry you most”Read all of that with the next section in hand. Pooled across sectors it overstates the case, for a reason worth being explicit about.
The individual links are not equally common:
- 871 of 1222 (71%) link to an about page
- 356 of 1222 (29%) link to a team page
- 270 of 1222 (22%) link to anything resembling credentials, accreditation or registration
Only 101 of 1222 firms — 8% do all three. 252 firms do none of them: their homepage offers no route to who runs the business.
What this does not prove
Section titled “What this does not prove”This is the part most articles skip, so it is worth being direct about it.
The credentials check was sector-blind, and that skews the pooled table. For most of this corpus the third check looked only for property-sector bodies — Propertymark, NAEA, ARLA, RICS, an ombudsman. A construction firm listing FMB and Gas Safe, or an accountancy practice listing ICAEW, scored zero on it. So “three links” was far easier for an estate agent to reach than for anyone else. The detector is now sector-specific, but these figures predate that fix.
Which is why the honest evidence is the within-sector table below, not the pooled one. Comparing firms only against others in the same index removes the sector effect entirely — and the relationship still holds.
This is correlation, not causation. Adding three links to a homepage will not move a site 14 points. The far more likely explanation is that these three links are a proxy: a business that maintains an about page, a team page and a credentials page is usually a business that maintains its website generally — which is also the kind of business that has a fast site, valid structured data and a complete Google Business Profile. The gradient is real; the arrow of causation is not established by this data.
Three binary checks are not an internal-linking audit. These three say nothing about click depth, orphaned pages, anchor-text distribution or link equity flow. A site could pass all three and still have a poor internal structure.
The sample is 1222 firms in one region. It is enough to be interesting and not enough to be a law.
The same test, within each sector
Section titled “The same test, within each sector”Each index compared only against itself, so no sector can flatter another.
| Index | Firms | 0 links | 1 | 2 | 3 |
|---|---|---|---|---|---|
| Bristol Double Glazing | 107 | 47 (n32) | 50 (n49) | 55 (n18) | 59.5 (n8) |
| Gloucestershire Double Glazing | 88 | 46 (n22) | 48.5 (n38) | 57 (n22) | 65 (n6) |
| Bath Architect | 78 | 46 (n19) | 49 (n38) | 56 (n16) | 59 (n5) |
| Bristol Architect | 66 | 43.5 (n20) | 51 (n29) | 55.5 (n14) | 51 (n3) |
| Cheltenham Physiotherapy | 63 | 44 (n11) | 50 (n29) | 59 (n20) | 67 (n3) |
| Bath Physiotherapy | 60 | 47.5 (n6) | 48 (n27) | 55 (n23) | 53 (n4) |
| Exeter Architect | 58 | 45 (n13) | 52.5 (n22) | 54 (n19) | 60.5 (n4) |
| Bristol Physiotherapy | 57 | 48.5 (n10) | 51.5 (n20) | 56.5 (n22) | 58 (n5) |
| Cheltenham Architect | 53 | 46 (n14) | 52.5 (n28) | 52 (n10) | 55 (n1) |
| Exeter Physiotherapy | 49 | 53.5 (n2) | 49 (n22) | 56.5 (n20) | 57 (n5) |
| Worcester Construction | 46 | 49 (n13) | 48 (n21) | 54 (n11) | 55 (n1) |
| Gloucester Physiotherapy | 45 | 51.5 (n8) | 51 (n23) | 61 (n14) | — |
| Gloucester Restaurant | 38 | 48 (n19) | 53 (n16) | 56.5 (n2) | 63 (n1) |
| Cheltenham Construction | 37 | 47 (n13) | 49 (n20) | 57 (n3) | 68 (n1) |
| Swindon Physiotherapy | 37 | 52 (n5) | 53 (n21) | 64 (n10) | 51 (n1) |
| Worcester Physiotherapy | 37 | 40.5 (n4) | 48 (n21) | 57 (n10) | 58.5 (n2) |
| Gloucester Accountant | 33 | 45.5 (n4) | 54.5 (n16) | 56 (n10) | 65 (n3) |
| Gloucester Architect | 28 | 45 (n7) | 49 (n11) | 61.5 (n6) | 56.5 (n4) |
| Swindon Architect | 26 | 46 (n9) | 55.5 (n14) | 58 (n2) | 48 (n1) |
| Bristol Vet | 25 | 53.5 (n2) | 59 (n10) | 57 (n9) | 58.5 (n4) |
| Worcester Architect | 21 | 39 (n5) | 56 (n14) | 51 (n2) | — |
| Cheltenham Vet | 19 | 45 (n3) | 67.5 (n8) | 57 (n4) | 67 (n4) |
| Gloucester Vet | 19 | 45 (n2) | 50 (n3) | 52 (n9) | 74 (n5) |
| Bristol Estate Agent | 18 | 42 (n1) | 59.5 (n6) | 59 (n9) | 62.5 (n2) |
| Exeter Vet | 18 | 47 (n1) | 56 (n5) | 50 (n8) | 54 (n4) |
| Bath Estate Agent | 15 | — | 61 (n3) | 65.5 (n6) | 73 (n6) |
| Swindon Vet | 15 | 39 (n1) | 71 (n3) | 56 (n8) | 63 (n3) |
| Worcester Vet | 14 | 50.5 (n2) | 50.5 (n6) | 65 (n3) | 78 (n3) |
| Gloucester Construction | 13 | 47 (n2) | 49.5 (n10) | — | 57 (n1) |
Median Digital Visibility Score by number of homepage links, within each index.
In 28 of 29 indices the highest-linking group outscores the lowest-linking group — though in 15 of them that top group is one or two businesses, so treat those rows as indicative rather than measured. Strictly, step by step, only 13 of 29 rise without a dip — and the dips sit on the thinnest cells, where a band of one or two firms moves the median on its own. Read the direction, not the individual steps.
Worcester Construction is the clearest example of why. It runs 49, then 48, then 54, then 55 — an apparent dip in the middle. Those groups hold 13, 21, 11, 1 firms. A median over 13 businesses moves if one of them moves; a median over 8 moves further. The dip is thirteen firms disagreeing slightly with twenty-five, not a finding. This is also why the article does not report a single pooled correlation coefficient: it would put a decimal point on something this uneven and invite you to trust it.
That is the finding, and it is the honest version of it: the effect is not an artefact of comparing estate agents with builders, because no index here is being compared with any other. It is also weaker than the pooled table at the top of this page implies, which is why that table should not be read alone.
What the data does support is narrower and still useful: among comparable local businesses in the same sector and city, the ones whose homepage tells you who they are tend to be the ones performing better in search. If your homepage fails all three checks, you are keeping company with the bottom group.
It varies more by sector than you would expect
Section titled “It varies more by sector than you would expect”| Index | Firms | Median links | Median score |
|---|---|---|---|
| Exeter Solicitor | 8 | 2 | 73 |
| Swindon Estate Agent | 8 | 3 | 70.5 |
| Bath Estate Agent | 15 | 2 | 66 |
| Cheltenham Law Firm | 6 | 2 | 64.5 |
| Bristol Dentist | 9 | 2 | 59 |
| Bristol Estate Agent | 18 | 2 | 59 |
| Swindon Vet | 15 | 2 | 58 |
| Bristol Vet | 25 | 2 | 57 |
| Cheltenham Vet | 19 | 1 | 57 |
| Worcester Vet | 14 | 1 | 55.5 |
| Swindon Physiotherapy | 37 | 1 | 54 |
| Worcester Architect | 21 | 1 | 54 |
| Bristol Physiotherapy | 57 | 1 | 53 |
| Exeter Architect | 58 | 1 | 53 |
| Exeter Physiotherapy | 49 | 2 | 53 |
| Gloucester Accountant | 33 | 1 | 53 |
| Gloucester Physiotherapy | 45 | 1 | 53 |
| Exeter Vet | 18 | 2 | 52.5 |
| Swindon Architect | 26 | 1 | 52.5 |
| Cheltenham Architect | 53 | 1 | 52 |
| Cheltenham Physiotherapy | 63 | 1 | 52 |
| Gloucester Restaurant | 38 | 0.5 | 52 |
| Gloucester Vet | 19 | 2 | 52 |
| Bath Physiotherapy | 60 | 1 | 51.5 |
| Gloucester Architect | 28 | 1 | 51.5 |
| Bristol Double Glazing | 107 | 1 | 51 |
| Worcester Physiotherapy | 37 | 1 | 51 |
| Bath Architect | 78 | 1 | 50 |
| Bristol Architect | 66 | 1 | 50 |
| Gloucester Construction | 13 | 1 | 50 |
| Gloucestershire Double Glazing | 88 | 1 | 50 |
| Bath Vet | 8 | 1 | 49 |
| Worcester Construction | 46 | 1 | 49 |
| Cheltenham Construction | 37 | 1 | 48 |
Counts are firms whose homepage parsed and which carry a score, so a sector’s row can be smaller than its published index.
Sector norms differ sharply, and a single signal never carries a score on its own — which is why the index weights six pillars rather than one. The useful comparison is against your own sector’s median, not a general benchmark.
How this was measured
Section titled “How this was measured”Every figure above is reproducible.
- The homepage of each business is fetched once per quarter and parsed for links whose href or anchor text matches about, team/people/staff, and credentials/accreditation/registration patterns.
- Each check returns
true,false, ornullwhere the page could not be parsed. A null is excluded, never scored as a zero — 149 homepages failed to parse and are absent from the percentages above, which is why the denominator is 1222 and not 1371. - The three checks form the Content & trust pillar, which carries 10% of the overall Digital Visibility Score.
The scoring itself is deliberately plain arithmetic — a ratio and a weighted sum, no model — so that any figure here can be recomputed from the published data with a spreadsheet. The methodology states the rules, the pipeline is open source, and every dataset is CC BY — one file per index, listed on the indices page.
If you want to check the gradient rather than take it on trust, the statistical methods for doing so are written up separately — regression analysis for the trend, ANOVA for whether the differences between those four groups are larger than the noise. We have one quarter of data, which is not yet enough to run either properly; that changes as the monthly series accumulates.
What to do about it
Section titled “What to do about it”Ranked by how much it matters for a local service business, not in general.
1. Make your homepage answer “who are you”
Section titled “1. Make your homepage answer “who are you””If a stranger lands on your homepage, can they reach a page naming the people who will do the work, within one click? For 71% of the firms we measured, no. For a solicitor, a dentist or a builder, that is the question the visitor is actually asking, and it is also what E-E-A-T is reaching for.
Credentials are the biggest gap: 78% of firms show none. If you are registered, accredited or regulated, the homepage should link to the page that says so.
2. Link with anchor text that names the thing
Section titled “2. Link with anchor text that names the thing”“Click here” and “read more” are wasted. The anchor is the strongest single statement you make about what the destination is about, and generic anchors throw it away. Name the entity: “our Bristol conveyancing team”, not “find out more”.
3. Prioritise the pages that earn
Section titled “3. Prioritise the pages that earn”Internal links distribute authority, and most sites spend theirs on navigation furniture — privacy policies and cookie pages get linked from every page while the service page that generates enquiries is three clicks down. Count how many internal links point at your most commercially important page. For most local sites the honest answer is “fewer than point at the privacy policy.”
4. Build the cluster, not the page
Section titled “4. Build the cluster, not the page”A page that stands alone competes alone. A service page supported by a location page, a process explainer and a case study — all linked to each other — states a topic rather than a single term. That is the principle behind topical authority, and it is what the Visibility pillar is really measuring when it counts how many of a sector’s keywords a firm ranks for at all.
5. Fix orphans before adding anything
Section titled “5. Fix orphans before adding anything”A page reachable only from the sitemap is a page you are paying to host and getting nothing from. Crawl your own site and list every page with zero inbound internal links. Deal with those before writing anything new — see the technical SEO guide and the indexing and crawling checklist.
Where your own firm sits
Section titled “Where your own firm sits”All 1371 measured businesses are listed on the indices, each with its own scorecard.
If your business appears in one of the 34 indices, your scorecard shows your Content & trust score and which of the three checks you pass, alongside your position against your sector median.
If a figure about your business is wrong, tell us and it gets corrected and logged in the changelog. Scores are never adjusted on request — the method is applied identically to every business in a cohort, including businesses that ask.
Data: PYC Local Digital Visibility Index, Q3-2026, 1371 businesses across 34 South West indices, recomputed at build time. Published under CC BY 4.0 — reuse with credit to Phil Yarrow (PYC), https://hub.pyc.agency/indices/. Analysis by PYC.