Internal linking as product design, not a plugin setting
Written by AERVIO, a Montreal studio. Human-reviewed before publish. Contact: hello@aervio.io · +1 438-498-3913.
Ask a team to show you their site's navigation and they can. Ask them to show you which pages link to the page that earns the most revenue, and you often get silence. That second map — the one made of links inside the content — is the internal linking structure. On most sites nobody designed it. It is what the theme, the footer, and a "related posts" widget happened to produce.
Treating it as a product changes the outcome without adding a page. This is how we approach it.
What internal links actually do
Google documents the basics plainly in its link best practices: crawlers find pages by following links, and link text helps them understand the destination. That gives internal links three jobs.
- Discovery. A page nothing links to is hard for a crawler to find, and it signals that you do not consider it important.
- Context. The words in a link describe the destination. "Managed IT in Montreal" tells a crawler and a reader something. "Click here" tells neither.
- Navigation for people. Before any ranking argument, a link is a decision about where a reader goes next. A site whose links lead nowhere useful loses readers regardless of where it ranks.
Notice what is not on the list: "pass PageRank around". That framing leads to scatter-gun linking. Think about the reader first; the crawler is following the same road.
Start with the question: what do we want a visit to become?
Every article, guide, or location page exists to move someone one step toward a conversation. The step might be a service page, a contact form, a calculator, or a sibling article that answers the next objection.
For each page type, we write down its destination before we write its links:
| Page type | Its job | Where it should send people |
|---|---|---|
| Homepage | Orient, route | The two or three offers and one or two explanatory notes |
| Service page | Qualify | Proof, pricing context, contact |
| Insights article | Answer a question | The relevant service, one or two sibling articles, contact |
| Location page | Confirm fit | Services, a relevant article, contact |
| Contact page | Convert | Little else. Do not send people away |
This is design work. It decides which pages are hubs, which are spokes, and which are deliberately left as dead ends — such as a thank-you page.
The structure we use: hub, spoke, and a short path to revenue
Our default shape for a content cluster is plain:
- A hub page on the topic, usually a service or a pillar article
- Spokes: articles answering specific questions, each linking to the hub and to one or two siblings
- A path to contact within two clicks of every spoke
This Insights section is built that way. For example, the SEO side links SEO and GEO in Montreal to the system argument in site, CRM, and search as one system, which links onward to the CRM piece CRM built around how you sell. A reader who arrives from a search about one topic can walk the cluster without hitting a wall.
The homepage should not link to twenty equals. If everything is promoted, nothing is. Pick the two or three pages that sell and make them obvious.
Write anchors like labels
Anchor text should name the destination. Compare:
- "To learn more about our approach, click here."
- "We cover this in site, CRM, and search as one system."
The second reads naturally, helps a screen-reader user who navigates by links list, and tells a crawler what the target is. A few rules we follow:
- Use the page's topic or name, not a generic verb
- Keep it short enough to scan, long enough to be unambiguous
- Vary the wording naturally. The same exact phrase on fifty pages looks mechanical
- Do not stuff keywords into anchors. If a sentence would read oddly aloud, rewrite it
Fix the structural leaks first
Before adding links, find where the existing structure leaks. In a crawl, look for:
- Orphans. Pages in the sitemap with no internal links pointing to them
- Links to redirected URLs. After a redesign, old URLs linger inside content. Each one adds a hop. Update them to point at the final URL
- Links to 404s. Dead ends for people and crawlers alike
- Navigation-only pages. Pages reachable only from a menu, never from content that explains them
- Pagination and filter traps. Faceted URLs that create thousands of near-duplicates from a handful of products
A crawl with a tool like Screaming Frog, plus Search Console's coverage and links reports, answers most of these in an afternoon. If you recently relaunched, Search Console after a redesign covers the first month.
Plugins help, but they do not decide
Related-post and auto-link plugins produce links based on tags or keywords. They are fine for filling the bottom of an article. They do not know that one page earns your revenue and another is a news post from 2019. Auto-linking every occurrence of a phrase to one page also tends to produce awkward text and over-linking.
Our approach: let a plugin suggest candidates, then have a person decide. The editorial rule is that each new Insights piece should earn at least two links from pages that already matter, and should send readers to one service page, the contact page, and a sibling article. Anyone publishing should be able to say where the page sits in the cluster before it goes live.
Make it part of the publishing checklist
Internal linking decays unless someone owns it. We add four lines to the publishing checklist:
- Which hub does this page belong to?
- Which two existing pages should now link to it?
- Which service and contact destination does it send readers to?
- Is the anchor text the name of the thing?
That takes two minutes per page. After a year it is the difference between a cluster and a pile.
Link structure and answer engines
AI answer surfaces and traditional search both read the same HTML. A clear link structure helps them understand that your site has a topic and that specific pages are the authority on sub-topics. It does not require special markup or a separate file. Links with descriptive text, claim-complete paragraphs, and consistent contact details are the same quality bar Google describes for AI features.
How to measure whether it worked
Be careful here. Internal linking changes are slow and noisy, and we do not promise a ranking outcome. What we watch:
- Whether orphan and low-link pages enter the index and begin earning impressions
- Clicks and impressions on the target pages for their main queries, over eight to twelve weeks
- Click paths: how many readers of an article continue to a service or contact page
- CRM sources: whether form submissions list the cluster pages as the first touch
If those do not move after a reasonable period, the problem is usually the page content, not the links.
Where to start
Pick the one page that matters most to revenue. List every page that links to it. If the list is the footer and the header, you have found your first job. Send the page and what it should rank for to hello@aervio.io, or call +1 438-498-3913, weekdays 9–17 ET from our Montreal studio. We will tell you what is linking to it today and what should.
Questions
How many internal links should a page have?
There is no magic number. A page should link to what a reader would plausibly need next, with descriptive text. A long guide may carry a dozen contextual links; a short service page may carry three. If a link would not help a person, it should not be there.
Do automatic related-post plugins count as internal linking?
They produce links, but they are driven by tags or recency, not by your commercial priorities. They can fill the bottom of an article. They do not replace deciding which pages should send readers to your service and contact pages.
Does anchor text matter?
Yes, as a description. Google uses link text to understand what the destination is about. Use the name of the thing, such as Hands-off IT, rather than click here or read more.
Read next
Let's write the next step.