What this workflow does
An internal linking audit checks whether people and crawlers can move through the site using clear HTML links. It looks beyond link counts. The audit asks whether an important page is discoverable, whether the link appears in a useful context and whether the anchor prepares the reader for the destination.
Google says it can generally crawl a link when it is an <a> element with an href that resolves to a web address. Google also uses links to discover pages and understand relevance. Those facts make implementation and wording part of the same review.
When to use it
Run this audit after a content inventory, during a navigation or taxonomy change, before a consolidation project, or when important pages have weak discovery paths. It is also useful after a migration because internal links often keep pointing at redirects or old canonical variants.
A focused audit can cover one section or page type. Define the scope rather than mixing storefront, documentation, editorial and account areas into one average.
Prepare the URL inventory
Start with the URLs the site intends to keep. For each URL, record:
| Field | Purpose |
|---|---|
| Canonical URL | Provides the preferred destination for analysis |
| Indexability | Prevents recommendations to blocked or noindex pages |
| HTTP status | Finds broken links, redirects and soft errors |
| Page type and topic | Supports meaningful source-to-destination matching |
| Business or reader priority | Separates critical paths from incidental pages |
| Organic landing evidence | Shows pages that users already enter from search |
| Existing inlinks and outlinks | Establishes the current graph |
Normalize protocol, host, case, trailing slash and known parameters according to the site’s URL rules. Keep raw URLs in the export so that a redirect or inconsistent link can still be traced to its source.
Build the crawlable link graph
Use rendered HTML when the site adds links with JavaScript. Count only links that a browser and crawler can treat as anchors with usable href values. Record the source URL, destination URL, anchor text, surrounding section and placement type.
Separate these placements because they do different jobs:
- primary and secondary navigation;
- breadcrumbs;
- section hubs, category pages and indexes;
- contextual links inside the main content;
- related-content modules;
- utility links in headers and footers.
Do not treat 100 repeated footer links as equivalent to 100 editorial recommendations. Keep sitewide and contextual links in separate columns.
Find discovery problems
Orphaned pages
Call a page orphaned only after comparing the crawler graph with other sources such as XML sitemaps, analytics landing pages, Search Console data and the CMS inventory. A page absent from one crawl may simply be outside the crawl scope or reachable from an untested state.
For each confirmed orphan, decide whether the page should be linked, consolidated, redirected, marked noindex or removed. Adding a link is not always the right answer.
Weak paths
Review priority pages with very few meaningful inlinks, excessive click depth or links only from low-value utility areas. Trace the actual path from relevant hubs and parent topics. A page can have many inlinks and still be poorly connected if they all come from one repeated template.
Broken and indirect links
Find internal links that lead to 4xx or 5xx responses, redirect chains, loops, temporary redirects and non-canonical variants. Update source links to the final preferred destination when that destination is stable and appropriate.
Do not replace a broken destination with the home page by default. Use the closest useful equivalent or remove the link when no equivalent exists.
Review anchor text
Read the anchor in its sentence and page section. A useful anchor is descriptive enough to set expectations, concise enough to scan and natural in the surrounding copy.
Flag anchors that are:
- empty or image-only without useful alternative text;
- generic when the destination could be named;
- misleading about the page or action;
- unnaturally repetitive across unrelated contexts;
- overloaded with keyword variants;
- longer than needed to identify the destination.
“Learn more” can be understandable when nearby text names the destination, but the link becomes harder to scan out of context. Prefer wording that makes sense in both places. Do not rewrite every anchor into the same exact-match phrase.
Audit page relationships
Links into priority pages
Identify authoritative and contextually relevant source pages that can help a reader reach the priority destination. The source should discuss a question that the destination answers in more depth or provides as a next step.
Links out to supporting pages
Priority pages should also help readers verify claims, learn prerequisites and continue related tasks. An audit that only pushes links inward creates dead ends and misses user journeys.
Hubs, breadcrumbs and navigation
Check whether the hierarchy is reflected in visible navigation and breadcrumbs. A hub should describe and link to its children, while child pages should offer a clear route back to the broader section. Avoid creating empty hubs solely to pass links.
Duplicate and competing pages
When several pages target the same reader task, more cross-linking may reinforce the ambiguity. Flag the overlap for consolidation or clearer differentiation before prescribing additional links.
Recommend contextual links
Each recommendation should specify:
| Field | Requirement |
|---|---|
| Source page | A canonical page that contains a real linking context |
| Destination | The final canonical URL, not a redirect |
| Placement | Existing paragraph or section where the link helps |
| Anchor direction | Natural wording, not a mandatory exact phrase |
| Reader reason | What the destination lets the reader do next |
| Editorial change | Sentence or context needed to make the link honest |
| Priority | Based on the destination and the strength of the relationship |
Do not drop links into unrelated paragraphs. If the source page needs a new claim just to justify the link, verify and edit that claim through the normal editorial process.
Prioritize the work
Fix links that block a critical journey or point to errors first. Next, address canonical inconsistencies and weak paths to important pages. Contextual opportunities come after those defects because they require editorial judgment.
Use link count as a diagnostic, not a target. Google’s guidance does not prescribe a magic number of links per page. The page should contain the links readers need, presented without making the main content hard to use.
Validation plan
After release:
- recrawl the changed source pages and destinations;
- verify status, canonical and robots directives;
- inspect rendered anchors and their
hrefvalues; - confirm mobile navigation exposes the intended paths;
- compare the updated graph with the approved recommendations;
- monitor relevant Search Console evidence without expecting immediate processing.
Keep a list of implemented, rejected and deferred recommendations. Otherwise the same suggestions will reappear in the next audit without context.
Quality checks
- The canonical and indexability inventory is validated before graph analysis.
- Sitewide links and contextual links are measured separately.
- Every reported orphan is checked against at least one source beyond the crawler.
- Every new link has a reader reason and a valid source context.
- Anchors describe destinations without forced repetition.
- New links point directly to successful canonical URLs.
- Broken-link replacements preserve user intent.
- The report contains a post-release crawl and manual sample.
Limitations
Authenticated areas, forms, internal search and interaction-dependent states may not appear in the crawl. Link graphs also change whenever templates or navigation change. State the user agent, rendering mode, crawl limits and date with the report.
Internal links can improve discovery and understanding, but an audit cannot guarantee indexing or rankings.