You are staring at a draft, cursor blinking, wondering how many internal links per page is the right call. Three feels thin. Twelve feels desperate. Nobody has ever given you a number you trust.
Here is the good news: there is no number to memorize, so you cannot get it wrong the way you fear. Google has never published an ideal count. What exists instead is a judgment call, and judgment is something you can learn in about ten minutes.
This piece gives you that judgment. You will see how internal link dilution actually works, what too many internal links looks like on a real page, what under-linking looks like, and a practical starting range you can apply to the next article you publish. No quotas. No density rules. Just a way to decide whether the next link earns its place.
Take a breath. You are closer to getting this right than you think.
Is there a maximum for how many internal links per page?
No. Google states plainly that there is no magical ideal number of links a page should contain.
That single sentence retires a lot of anxiety. There is no ceiling to stay under and no floor to reach. You may have heard that 100 links per page is the hard limit. Do not carry that around anymore. It is not current guidance, and treating it as a rule will make you cut links that were helping readers.
What Google does offer is a heuristic you can actually use: if a page feels like it has too many links, it probably does. That sounds soft. It is not. It hands the decision back to the only judge that matters, which is a person reading the page.
The rest of Google's link guidance is just as practical:
- Link to other resources on your site when they help a reader understand the page, and do it in context.
- Every page you care about should have a link from at least one other page on your site.
- Do not chain links next to one another. Two links side by side are hard to tell apart and lose the words that gave each one meaning.
- Use ordinary HTML anchors with an href. Links a crawler cannot parse do not count, no matter how many you add.
Notice what is missing. No count per 1,000 words. No maximum. No required density. The guidance is about usefulness, and that is a much easier standard to meet than a number you were guessing at.
Your next step: open your most recent article and read it once as a reader, not as a marketer. If the links felt like noise, you have your answer already.
How internal link dilution actually works
Internal link dilution is what happens when one page points to more destinations than it can meaningfully prioritize. The page's ability to pass value gets spread thinner across every extra link, and no single destination stands out as the one that matters.
The classic way to picture it comes from the original PageRank model. A page has some amount of value to pass along. It splits that value across its outgoing links. Point to four useful pages and each one receives roughly a quarter of what this page has to give. Add six more low-value destinations and the same contribution now spreads across ten. Each destination gets a smaller slice.
That is link equity dilution in one picture, and it is worth holding onto because it explains the direction of the effect.
It is also a teaching model, not a calculator. The real PageRank algorithm is a recursive graph calculation with a random-surfer interpretation and a damping factor, and modern Google ranking is not documented as a simple one-over-the-number-of-links formula. So use the picture to reason about priority. Do not use it to do arithmetic on your rankings.
Here is the part most people miss.
Two different things get called "internal links," and they behave in opposite directions:
- Links on a page are the outgoing ones. They navigate, they add context, and in the dilution model they divide what the source page has to give.
- Links to a page are the incoming ones. They help readers and crawlers find that destination, and they signal that the site considers it important.
A page can be stuffed with outgoing links and still be starving as a destination, because nothing points at it. A page can also collect a healthy number of inbound links and still be over-linked itself, because its own body copy plus its menu plus its footer plus its related-posts module all point everywhere at once.
Once you separate those two, link equity dilution stops being scary. It is not a penalty waiting to fire. It is a budget question. Every extra destination you expose competes with the ones you actually care about, for the reader's attention first and for your page's contribution second.
One more thing to keep straight: dilution does not mean every extra relevant link hurts you. An extra link can create real value by helping a reader, exposing a page that was hard to find, or making a relationship on your site clear. The only question worth asking is about the marginal link. Does this next one help enough to justify one more destination competing for attention?
Your next step: count the contextual links in your body copy separately from your template links. They are different budgets and mixing them is what makes the number look alarming.
What is a practical starting range?
For a typical editorial article, start with roughly 3 to 5 contextual internal links in the body, on top of whatever your header, navigation, footer, and related-content modules already do.
Be clear about where that range comes from. It is a practitioner starting point reported in guidance from Ahrefs and Semrush. It is not a Google threshold, and it did not come out of a study that found an optimum. Treat it as a place to begin, not a target to hit.
Then adjust by what the page actually is:
| Page situation | Where to start | How to adjust |
|---|---|---|
| Short, tightly focused article | 2 to 4 contextual links | Go lower if there are only a couple of genuinely useful next steps. |
| Typical editorial article | 3 to 5 contextual links | Add only destinations that answer a likely follow-up question. |
| Long, reference-style article | More than 5 can be right | Add section by section, and only where the link is genuinely useful. |
| Conversion or landing page | Often fewer | Protect the path to the action. Keep only links that remove a real barrier. |
| Directory, category, or navigation template | Potentially many | The count may be structurally necessary. Organize it so priorities still read. |
Please do not turn 3 to 5 into the optimal number of internal links for everything you publish. There is no such thing. A long reference page that answers twelve subquestions legitimately needs more paths than a 700-word answer piece. A product category page needs more still. The standard is not the count, it is whether a reader can see what matters.
You may also run into advice to place 1 to 2 links in every short paragraph. The source that offers it says outright that this is not an exact science. Applied mechanically, it will over-link your article fast. Use it as a loose sense of spacing, never as a quota.
The honest definition of the right number: the smallest set of links that covers the paths a reader is likely to take, while keeping your important pages reachable.
Your next step: on your next draft, write the article first with zero links. Then add only the ones you would miss.
The signs you have too many internal links
You have crossed the line when the page stops being easy to read, or stops making its priorities obvious. The count itself is never the tell. These eight signals are.
1. The page is hard to read. Links interrupt sentences, half the lines look clickable, and the reader cannot tell which paths matter.
2. The hierarchy disappears. The page points at so many destinations that neither a reader nor a search engine can infer what your site actually considers important.
3. A link exists without a reader reason. The destination is loosely related, repeats what is already on the page, or is there mostly because you wanted another internal link.
4. The same destination is linked over and over. Repeating a link to one article in a single page will not necessarily help that target. If you need it twice, space the two out and make sure each one serves a different moment.
5. Menus, footers, and sidebars point to everything. Sitewide links spread attention and the source page's contribution across low-priority destinations. Keep persistent navigation focused on the pages readers need and the business needs to expose.
6. Links are chained together. Two or three links sitting next to each other lose the surrounding words that explained them. Google calls this out specifically.
7. The page was built for link count, not usefulness. Hidden links, or thin pages made mainly to place links, are a policy problem. Ordinary relevant internal links are not spam just because there are a lot of them.
8. The intended action is unclear. On a landing page, even a technically relevant link competes with the one thing you wanted the reader to do.
Hundreds or thousands of links on a page do make it hard for people and search engines to see which links matter. That is a real observation. It is not a universal cutoff, and you should not read it as one. A directory or a product index may need exactly that many.
If your article trips two or three of these signals, the fix takes fifteen minutes. Delete the links that are repetitive, weakly related, adjacent without context, or present only because a count target told you to add them. That is usually most of them.
Your next step: pick one over-linked article this week. Cut it to the links you can defend out loud. One article. That is the whole assignment.
The signs you are under-linking
Under-linking is not having fewer links than some target. It is failing to give readers and crawlers a route to a page that matters.
Here is what it looks like:
- A page the business cares about has no inbound internal link from anywhere on the site. That is an orphan page, and it is harder for search engines to find and index. It also misses the internal value other pages could pass to it.
- You publish something new and never link to it from the existing articles it relates to. Nobody has a route in.
- An important page sits several clicks deep when it could be reached directly. Practitioners often use three clicks from the homepage as a working target for priority pages. That is an architecture heuristic, not a Google rule, and it is still a useful one.
- An article makes a claim that another page on your site explains in full, and gives the reader no way to get there.
- The only real route to important pages is your site search box. Google does not generally type queries into a search box while crawling, so those pages may as well not be linked at all.
Under-linking hurts in a quieter way than over-linking, which is exactly why it survives so long. Nobody complains about a link that is not there.
The fix is small and specific. You do not need to sprinkle links across every article. Add one intentional inbound path from a relevant page that is already easy to find. A page with existing traffic and authority makes a good source for a newer page that has neither.
Your next step: list the five pages your business actually needs people to reach. Check that each one has at least one inbound internal link from a relevant article. Fix whichever is missing.
Link count and click depth are two different problems
Link count answers "how many destinations does this page expose?" Click depth answers "how many steps does it take to get there?"
Those come apart more often than you would expect. A page can carry plenty of links and still leave your priority pages buried, because those links lead into long winding paths. A page can carry very few links and still make everything easy to reach, because the paths are short and direct.
So when a priority page feels invisible, check depth before you add links. Adding twenty more destinations to an article will not shorten the route to the page you care about. One well-placed link from a page people already land on will.
Your next step: click from your homepage to your most important page. Count the clicks. If it takes more than three, that is a structure problem, not a link-count problem.
What internal links do for AI extraction
There is no AI-specific link count. No documented optimal number of internal links earns you a citation in ChatGPT, Perplexity, or Google's AI features.
Google's own position is that the normal best practices still apply for AI Overviews and AI Mode, with no additional requirements and no special optimizations needed. So anyone selling you a magic number for AI is selling you something they cannot back.
What does help is different from what you might expect, and it is mostly about the page, not the links:
- Put the important answer where it is easy to find, in clear and concise sections.
- Use descriptive headings, tables, and FAQ material where they genuinely aid understanding.
- Support your claims with examples, data, and sources.
- Keep the content accurate and current.
- Make the page crawlable, and make sure important pages are reachable through ordinary links.
- Use internal links to expose supporting detail and clarify relationships, not to build a dense web inside every paragraph.
The practical rule is short. Make the answer self-contained enough to lift out on its own, then offer a small number of high-value paths to the deeper material. More links do not make a page more extractable. A cluttered page makes the main answer harder to find, for a person and for a machine.
One caution about the dashboards. AI citation reporting, including Bing's AI Performance data, describes aggregated citation activity rather than a full log of every answer. Bing's own documentation says citation counts show how often pages are cited, not their importance, ranking, or role in a response, and that the data does not explain why a page was referenced. Use it for trends and comparisons. Do not use it to reverse-engineer a link-count threshold that was never there.
Your next step: open one page you want cited. Can a stranger read the first two paragraphs and get the whole answer? If not, that is your work, not your link count.
Where this leaves you
The right number of internal links is the smallest set that covers the useful paths a reader might take while keeping your important pages easy to reach. That is it. That is the rule.
You do not need a density calculator. You need to ask one question of each link: would a reader logically want this next? If yes, keep it. If you are arguing with yourself about it, cut it.
If the manual side is what wears you down, that part is worth systematizing. Doing this by hand means holding your whole site in your head while you write, and that is the step most teams skip when they are behind. In DeepSmith, internal linking happens inside the writing pipeline rather than after it: the Content Map keeps a current picture of every page on your site, and Content Studio uses it to place internal links while the article is being written. Nothing there tells you a universal correct count, because no tool can. It removes the cross-referencing, so your judgment is the only part left to apply.
Start with one page this week. Cut what does not help, add one inbound link to something important, and move on. Momentum matters more than perfection here.
Ready to stop doing the cross-referencing by hand? Start a free DeepSmith trial and see what a linked, publish-ready draft looks like on your own site.
