Internal Links: The Cheapest SEO Work You Skip

Almost every website we are asked to look at has the same gap. The pages exist. The writing is fine. Someone has paid for content, maybe paid for links, certainly paid for the build. And then the pages sit there like strangers at a wedding, barely acknowledging each other.

Internal linking — links from one page of your site to another page of your own site — is the only meaningful SEO work that costs nothing but attention. No tools to buy, no outreach, no monthly retainer strictly required. You can do a lot of it with a spreadsheet and an afternoon. It is also the work that almost nobody does, because it is unglamorous and because no one sells it as a product.

What an internal link actually does

Three things, and it helps to keep them separate in your head.

Discovery. A search engine finds pages by following links. If a page on your site has no link pointing to it from anywhere else on your site, and it is not in your sitemap, the only way it gets found is luck. These are called orphan pages. They are more common than you would think, especially on sites where someone built a landing page for a campaign two years ago and nobody ever wired it into the menu.

Importance. Links are how a site tells a crawler what it thinks matters. A page linked from the main navigation, from the footer, and from six blog posts is clearly central. A page reachable only by clicking through three sub-menus is clearly not. Google has never published the arithmetic behind this and we should not pretend to know it. But the direction of travel is not really disputed by anyone who works on large sites: the internal link graph is one of the signals of what a site considers its own important pages.

Meaning. The words in the link — the anchor text — describe the destination. This is the part most sites throw away, and it is the part with the clearest documented value. Google's own guidance on link text asks publishers to write anchor text that describes the page being linked to, and specifically discourages generic text like "click here" and "page". That is not a theory. It is in the documentation.

Anchor text is the description, not the decoration

Open your most recent blog post. Find the links out to your own service pages. What do they say?

If they say read more, click here, our services, or — the most common one — the bare URL, you have written a link that carries no information. A reader deciding whether to click gets nothing. A crawler building a picture of what that destination page is about gets nothing.

Compare these three ways of pointing at the same page:

• To know more about our services, click here.
• We also offer web development.
• If the site itself is the bottleneck, our website design and development work is where to start.

The third one tells a human what they will get and tells a machine what the page is. It also reads like a sentence a person wrote, which matters more than it used to.

The over-optimisation trap

The moment people learn anchor text matters, they overdo it. Every internal link becomes "SEO services in Delhi", repeated forty times in identical form. Do not do this. Vary the phrasing the way you would in normal speech. Sometimes the service name, sometimes a description of the problem it solves, sometimes the page title. Write the sentence first, then make the natural noun phrase into the link.

There is a simple test. Read the sentence out loud with the link words in it. If it sounds like something you would say to a customer on the phone, it is fine. If it sounds like a directory listing, rewrite it.

Hierarchy: how far is your money page from the front door

Click depth is the number of clicks from your home page to any given page, taking the shortest route. Your home page is depth zero. Anything in the main menu is depth one. A blog post reachable only from page four of your blog archive might be depth five or six.

Nobody outside Google can tell you the exact depth at which a page starts being neglected, and anyone who quotes you a number is guessing. What is observable is simpler: pages buried deep get crawled less often and tend to update more slowly in search results. If your highest-margin service sits four clicks deep behind a mega-menu, that is a choice you made without meaning to.

Some practical consequences:

• Paginated archives are a trap. A blog with 200 posts and ten per page means post number 150 is nine clicks from the front door. Category pages and a decent related-posts block fix most of this.
• The footer is not a filing cabinet. A footer with sixty links makes every one of them less meaningful. Keep it to the things a visitor genuinely needs from any page.
• Menus that only expand on hover do not exist on a phone the way they do on a laptop. Check what your navigation actually renders as on a mobile device. Sometimes the whole second level quietly disappears.

The pages that deserve the links

Here is the decision that actually matters, and it is a business decision rather than a technical one.

List your pages in three buckets.

1. Pages that make money. Service pages, product categories, the pages where a form or a phone number does the work. Usually between five and twenty pages.
2. Pages that build trust. About, team, pricing if you publish it, case studies, contact.
3. Pages that attract strangers. Blog posts, guides, answers to questions. There may be hundreds.

Almost all internal linking effort should run from bucket three into bucket one. A blog post about GST invoicing that gets steady traffic and links to nothing is a waiting room with no doors. It brought someone in and then let them leave.

The reverse is rarely worth much. A service page linking out to nine blog posts sends people away from the point of decision. One or two, where the post genuinely answers a hesitation a buyer would have, is enough.

Where the links should physically sit

A link inside a paragraph of body copy, written as part of a real sentence, is doing more work than the same link in an auto-generated "related posts" strip at the bottom. The automated strip is not useless, but it is boilerplate, and boilerplate repeated across every page carries less weight than editorial text. Put the important links in the prose.

What not to buy

Several things get sold under the internal linking heading that we would not pay for.

Automatic keyword-to-link plugins. These scan your text and turn every instance of a chosen phrase into a link. They produce pages with the same link repeated five times, links inside headings, links in places no editor would put them, and occasionally a link from a page to itself. They save an hour and create a mess that takes a day to clean. If you already have one running, turn it off and look at what it did.

Footer link blocks for every city in the NCR. "SEO in Noida, SEO in Gurgaon, SEO in Ghaziabad" across all 400 pages. This is a pattern from 2012 and it fools nobody. If you genuinely serve Noida and want a page for it, write a page about working in Noida with something real on it. Otherwise leave it out.

"Silo architecture" as a packaged deliverable. The underlying idea — group related pages, link within the group — is sensible. Sold as a rigid structure that forbids linking across groups, it makes sites worse and makes editing painful. Link where it helps a reader.

Adding nofollow to internal links to "sculpt" flow. This was a technique a very long time ago. It has not worked in the way people describe for well over a decade, and the plugins that still offer it are selling nostalgia.

A two-hour audit you can do yourself

You do not need to hire anyone for the first pass.

1. List your money pages. Five to twenty URLs in a spreadsheet.
2. For each one, search Google for site:yourdomain.in "the topic". This shows you which of your own pages mention that topic. Those are your candidate linking pages.
3. Open each candidate and check whether it links to the money page. Most will not.
4. Add one link, in the body text, with descriptive anchor text. One good link per page beats four stuffed ones.
5. Check for broken internal links. Any link returning a 404 wastes a crawl and irritates a visitor. Links pointing at old URLs that 301 redirect elsewhere still work, but chains of two or three redirects are worth cleaning up — point the link at the final destination.
6. Find the orphans. Compare the URLs in your sitemap against the pages you can actually reach by clicking. Anything unreachable either needs a link or needs deleting.

Step five and six are where a tool helps, and where ongoing technical maintenance earns its keep, because broken internal links accumulate every time someone renames a page. But steps one to four are editorial judgement, and you know your business better than any crawler does.

Does any of this matter for AI search?

Honest answer: partly, and nobody outside the model providers knows how much.

What is reasonable to say is that the assistants which cite sources are reading pages that were crawled, and crawling still follows links. A page nothing links to is less likely to be in the index that these systems draw on. Beyond that — whether anchor text influences how a language model summarises your business — we would be guessing, and so is anyone who tells you otherwise with confidence. We write more about what is and is not knowable in AI visibility work, and we try to flag the uncertain parts as uncertain.

Where to start this week

Pick your single most profitable service page. Search your own site for the three phrases a customer would use to describe that service. Open every page that comes back and add one sentence with one link, written the way you would say it. That is it. Twenty minutes, no budget, and it is more than most sites have ever done.

Do that for each money page over a few weeks and you will have rebuilt your site's internal structure without touching the design. If the site is large enough that this is genuinely impractical by hand, or the structure itself is the problem rather than the links, that is a conversation worth having — our SEO work usually starts here rather than with content, and you can tell us what you are working with.

Questions we get asked