Why Linking Is the Hard Part
When pages are generated from data, nobody places each one by hand. Without a system, new pages ship with nothing pointing at them. An orphan page is hard for search engines to find and receives no authority from the rest of the site. Linking has to be designed once and enforced by the build.
Step 1: Give Every Page a Parent
Each page belongs to a category, and each category has a hub page that lists its pages. The hub is the guaranteed route in. Link the hubs from somewhere on every page, such as the footer, so they sit one click from anywhere.
Step 2: Add a Middle Layer When Hubs Get Large
A hub that lists everything stops working as it grows. Our research hub once rendered one card per page: 639 cards, 681 links and 1.68 MB of HTML on a single page. Each page it linked to received a tiny share of its authority, and the page was large enough to risk being cut short by crawlers.
We added topic pages between the hub and the individual pages. The hub now links to eleven topics, and each topic lists its own pages. The addresses of the individual pages did not change. The structure is hub, then topic, then page.
Step 3: Add Related Links to Every Page
Each page links to a handful of closely related pages in its own category, and to a few in other categories. Choose them in this order.
- Hand-picked links where you know the relationship.
- Pages that link to this one, to return the link.
- Pages with overlapping keywords.
- Any sibling, so no page has an empty block.
Cap the block. Eight related links is plenty. More dilutes each one.
Step 4: Link New Pages From Established Ones
A batch of new pages that link only to each other is an island. For each new page, add a link from two to four established pages on closely related subjects. Keep the list short and relevant. Mass changes to pages that already rank create noise in your results and can unsettle pages that were doing well.
We avoid touching pages currently ranking just off the first page, since they are the most sensitive to change, and choose sources that have room in their related block.
Step 5: Check What Actually Renders
This is the mistake we made. A link can exist in your data and never appear on the page. Our related block shows eight links, hand-picked ones first. We added new pages to the related lists of pages that already had eight, so the new links were stored and never shown. An audit of the rendered pages found it.
Test the output. For each new page, count the pages whose rendered HTML links to it.
Step 6: Make the Build Enforce It
| Check | What it prevents |
|---|---|
| Every hub is linked from the footer | A whole category becoming unreachable |
| Every page is in the sitemap | Pages that exist and are never submitted |
| Every page has a category and a parent | Pages created outside the system |
| No two pages share an address | Two pages competing for one URL |
| Related links point to pages that exist | Broken links after a page is removed |
If a check fails, the site does not publish. That is more reliable than a periodic audit, because it catches the problem on the day it is introduced.
Step 7: Keep the Marketing Pages Separate
You do not need to link thousands of pages from your main navigation or homepage. Keep those for visitors deciding whether to buy. The generated pages are linked through hubs, topics, related blocks and the footer.
Is Crawl Budget the Problem?
Usually not. Google says its crawl budget guidance is for sites with roughly a million or more pages, or ten thousand or more that change daily, and that other sites do not need to read it. If your pages are not being found, look at links first.
If You Want This Done for You
Presenc AI runs a team of AI agents that does this work, and a person reviews every page before it ships. Request access.
Sources
The structure and both problems described are from our own site. Crawl budget thresholds are from Google's crawl budget guide, read November 13, 2026.