Internal Linking
Build an internal linking system that connects clusters, fixes orphan pages, controls link depth, and routes readers from useful content to commercial pages.
Internal linking is the practice of connecting pages on the same domain. It is the part of authority flow, discovery, and reader routing that the site owner fully controls: no outreach, publisher approval, or algorithm change is required to add a useful path between two pages. That control is why internal links are one of the most under-used levers in otherwise mature SEO programs.
Phase: P12, Internal Linking. Stage: C — Build. Timebox: 2–4 working days for the first site-wide pass, then 15–30 minutes per new or materially refreshed page. Owner: SEO lead, with content owners implementing contextual links and engineering owning template or navigation changes.
The phase turns a planned graph into live, crawlable routes. It does not mean inserting a fixed number of links into every article. Each link must have a reason: orient the reader, delegate detail, supply evidence, continue a task, or bridge useful information to a relevant commercial destination.
Why this phase, and why here
P12 consumes the approved topical map and information architecture from P8, the production inventory from P10, the pages created in P11, and the canonical URLs confirmed during technical work. Those earlier phases decide page ownership, cluster membership, hierarchy, and destination. This phase implements the relationships they specified.
The order matters because links are directional decisions. Adding links before canonical destinations are settled can send readers and crawlers to pages that will be merged or redirected. Adding them before content exists encourages authors to link to approximate matches. Running the audit before production is finished creates an orphan report that becomes obsolete as soon as the next batch publishes.
Skipping the phase leaves three separate failures. New pages may remain orphans: live URLs with no internal links pointing to them. Topic clusters may exist only in a planning sheet, so the site never expresses their relationships. Commercial pages may receive navigation links but no contextual routes from the informational pages where readers develop intent and trust.
This phase also follows on-page optimization because the destination page’s job and primary subject must be stable before anchor language can describe it accurately. It precedes entity and structured-data work because live link relationships provide evidence about hierarchy and relatedness; the next phase should model the site that actually exists, not the map that was intended.
Inputs and outputs
An input is accepted only when it identifies live or approved canonical URLs. An output is accepted only when another owner can verify the change without reconstructing the audit.
| Direction | Item | Acceptance condition |
|---|---|---|
| Input | Approved topical map | Names each page’s canonical URL, cluster, pillar or spoke role, intent, and required relationships. |
| Input | Current crawl and URL inventory | Includes every indexable HTML page, status code, canonical target, inbound internal-link count, source pages, and click depth where available. |
| Input | Production and migration log | Identifies newly published pages, pages awaiting release, merge survivors, redirects, and retired destinations. |
| Input | Commercial destination list | Names the product, service, category, use-case, lead, or conversion pages the business can genuinely support. |
| Output | Implemented link graph | Records every added, changed, and removed link as source URL, destination URL, anchor, placement, reason, and status. |
| Output | Resolved orphan register | Every indexable orphan is linked, redirected, consolidated, intentionally noindexed, or documented as an approved exception. |
| Output | Depth exceptions report | Lists priority pages deeper than the operating threshold, the cause, owner, action, and approved exception where relevant. |
| Output | Cluster contract check | Confirms spoke-to-pillar, pillar-to-spoke, and justified lateral routes for every in-scope cluster. |
| Output | Commercial bridge map | Identifies the informational sources that route qualified readers to each priority money page and explains the transition. |
The outputs form a contract with P13: canonical pages and their live relationships are known, so entity and schema decisions can describe a stable structure.
Ready to Monitor Your AI Visibility?
Track how AI chatbots mention your brand across ChatGPT, Perplexity, and other platforms.
The checklist
The sequence starts with destination safety, moves through cluster structure and reader routes, then closes with a fresh crawl. Every item has a done-when condition because “add more links” is not auditable.
1. Freeze canonical destinations
What: Confirm the final URL and indexability of every page in scope, especially merge survivors and newly published destinations.
Why: A contextually perfect link still wastes attention and authority if it lands on a redirect, duplicate, error, or page scheduled for retirement. Destination stability must come before placement.
How: Join the topical map, crawl, sitemap, production log, and redirect plan. Replace proposed URLs with canonical live URLs. Exclude pages blocked from indexing unless the link serves a deliberate user-only function.
Tool: Crawl export, migration log, and the AmICited Pages report.
Done when: Every destination returns the intended successful response, resolves to itself as canonical, matches the approved intent, and has no unresolved retirement or merge instruction.
2. Enforce the cluster contract
What: Make every spoke link up to its pillar page , make every pillar link down to all active spokes, and add lateral links only between genuinely related spokes.
Why: The up-and-down pattern explains the cluster without forcing every page into a complete mesh. A spoke routes the reader to broader orientation; a pillar delegates narrower questions. Lateral links remain meaningful because they appear only when one spoke is a plausible next step from another.
How: For each cluster, compare the approved node list with live source pages. Add a contextual route from each spoke to the pillar. Add a descriptive route from the pillar to every active spoke, ideally near the section that introduces that subtopic. For a lateral link, write the reader’s next question in the change log; if the destination does not answer it, omit the link.
Tool: Semantic Map, topical-map graph, rendered pages, and crawl link data.
Done when: Every active spoke has a live contextual link to its pillar, the pillar links to every active spoke, and every lateral edge has a recorded reader reason rather than “same cluster.”
3. Rewrite weak anchor text
What: Replace vague or misleading anchor text with concise language that describes the destination or the task completed there.
Why: The clickable words are the strongest descriptive signal the site owner chooses for a link. “Read more,” “click here,” and a bare “learn more” spend that signal without naming what waits on the other side. They also make scanning harder for readers using assistive technology or reading links out of context.
How: Read the sentence around the link, not the anchor alone. Prefer natural phrases such as “compare enterprise plans” or “diagnose crawl-depth problems.” Keep the anchor specific enough to set an expectation and broad enough to remain truthful. Avoid stuffing an exact keyword into every source; repeated mechanical anchors make prose worse and can misrepresent slightly different destinations.
Tool: Crawl anchor export, content editor, and rendered-page review.
Done when: No priority contextual link uses “click here,” “read more,” or an unqualified “learn more”; duplicate destinations are not linked repeatedly from the same section; and every anchor still describes the destination when read without its surrounding sentence.
4. Audit homepage distance and priority-page depth
What: Measure crawl depth as the shortest discoverable link path from the homepage to each indexable page, then review priority pages deeper than three clicks.
Why: Depth is a routing diagnostic. A valuable page buried behind five or six choices is harder for a new visitor to find and depends on more pages remaining linked and crawlable. The three-click threshold is an operating review trigger, not a claim that every site must be flat.
How: Use a crawl that follows internal HTML links. Separate global navigation, breadcrumbs, pagination, and contextual links so the shortest path is understandable. For every priority page at depth four or more, find the nearest relevant pillar, category, directory, or high-traffic page that could offer a useful shorter route. Do not add homepage links merely to improve the number.
Tool: Crawl depth report and Directory View.
Done when: Every priority commercial page, pillar, and active campaign page is within three clicks of the homepage or has a documented reason, owner, and dated plan; no shortcut is unrelated to the source page.
5. Resolve every orphan
What: Find indexable pages with zero discoverable inbound internal links and assign each one an action.
Why: A sitemap can reveal an orphan’s URL, but it does not explain where the page belongs or give a reader a path to it. Orphans often expose a deeper decision that was never made: integrate, consolidate, redirect, noindex, or retire.
How: Compare crawl-discovered URLs with sitemaps, analytics landing pages, search performance, CMS exports, and the production log. Validate true orphans manually because crawler scope, authentication, JavaScript, or pagination can create false positives. If the page is valuable and unique, link it from a relevant live source. If it duplicates another page, follow the approved consolidation route rather than manufacturing a link to preserve it.
Tool: Pages report, crawl comparison, sitemap export, and CMS inventory.
Done when: The count of unexplained indexable orphans is zero; every detected URL has evidence, a disposition, an owner, and verification in the follow-up crawl.
6. Build informational-to-commercial bridges
What: Connect informational pages to relevant money pages—pages intended to generate a purchase, subscription, enquiry, booking, or another business outcome—at the point where the reader’s task naturally advances.
Why: Informational content can earn discovery and trust while leaving the business destination isolated. A bridge passes the reader and internal authority toward the page that can satisfy the next commercial need. The transition must be useful; forcing a product link into an unrelated definition damages trust.
How: For each priority money page, identify informational pages whose readers share the problem, audience, or decision. Add a sentence that explains why the destination is the next step, then use a descriptive link. Work in both directions where helpful: the money page can link back to implementation detail, evidence, or definitions rather than repeating them.
Tool: Pages report for source-page performance, Semantic Map for subject adjacency, conversion-path inventory, and content review.
Done when: Every priority money page has at least two relevant contextual inbound-link opportunities evaluated, at least one live contextual inbound link where a valid source exists, and a documented content gap where no honest bridge is currently possible.
7. Rebalance links from strong and over-linked sources
What: Review pages with substantial visibility, traffic, or inbound links, alongside templates and hubs that send unusually many links.
Why: Internal links distribute link equity and attention. A strong page that never routes onward traps opportunity; a template that links to hundreds of low-priority destinations makes every route less selective and harder to scan.
How: Inspect high-performing sources for missing routes to priority destinations. Review sitewide footer, tag, related-content, and faceted links separately from editorial links. Remove duplicates and obsolete destinations before adding new ones. Preserve links that users need even when they are not SEO priorities.
Tool: Pages report, crawl inlink/outlink counts, analytics, and template inventory.
Done when: The highest-value source pages have useful onward routes, no priority page relies only on a sitewide template link, and any page with more than 150 crawlable internal links has been reviewed for duplicates, utility, and unnecessary template expansion.
8. Validate the rendered graph
What: Crawl the released site again and inspect a sample of rendered pages on desktop and mobile.
Why: A CMS entry or ticket status does not prove that a crawler can follow the link. Links can disappear behind scripts, point to staging hosts, inherit nofollow, be hidden by templates, or break during deployment.
How: Re-crawl from the homepage with the same scope used for the baseline. Compare orphans, depth, status codes, canonical targets, anchors, and source-to-destination edges. Manually inspect every template changed and a sample from each cluster. Record the crawl time and release identifier.
Tool: Crawl comparison, browser inspection, AmICited reports, and the implementation log.
Done when: Every planned edge appears in rendered HTML or an intentionally crawlable equivalent, all destinations resolve correctly, unexplained orphans equal zero, priority depth exceptions are recorded, and the owner signs the final graph export.
Tools in AmICited
AmICited helps choose and verify routes; it does not replace the crawl needed to prove that a specific HTML link is live.
| Product view | Use in this phase | Deep link | Completion evidence |
|---|---|---|---|
| Organic vs Paid Pages | Find high-visibility source pages, weak commercial destinations, and newly changed landing pages that need link review. | Open Pages | Export the source and destination URLs used in the bridge map. |
| Directory View | See site-section shape, descend through directories, and identify deep or isolated sections before reviewing page-level routes. | Open Directory View | Capture the relevant directory level and record the route selected for each depth fix. |
| Semantic Map | Test whether two subjects are genuinely adjacent before creating a lateral or informational-to-commercial link. | Open Semantic Map | Record the cluster reviewed and validate the proposed relationship against the actual page content. |
Stay Updated on AI Visibility Trends
Get the latest insights on AI mentions, brand monitoring, and optimization strategies.
Decision rules
These are operating thresholds for consistent review, not universal ranking-factor claims. Record justified exceptions instead of quietly changing the rule.
| Signal | Bad looks like | Required action |
|---|---|---|
| Indexable orphans | More than 0 unexplained pages with no crawl-discoverable inbound internal link. | Integrate, consolidate, redirect, noindex, or document an approved exception; then re-crawl. |
| Priority-page depth | A priority commercial page, pillar, or current campaign page is 4 or more clicks from the homepage. | Find a relevant shorter route or record why the depth is intentional. |
| Cluster contract | Any active spoke has 0 contextual links to its pillar, or the pillar omits any active spoke. | Implement both directions before closing the cluster. |
| Contextual support | A priority page has 0 contextual inbound links and relies only on navigation, footer, sitemap, or breadcrumb discovery. | Add a useful editorial source or document the missing source-content gap. |
| Weak anchors | Any priority contextual link says only “click here,” “read more,” or “learn more.” | Rewrite it to describe the destination or task. |
| Same-page duplication | The same destination is linked more than once within one short section without a distinct user need. | Keep the clearest placement and remove the duplicate. |
| Lateral density | A spoke links to every other spoke merely because they share a cluster. | Remove links that cannot be justified by a next reader question. |
| Commercial bridge | A priority money page has no relevant informational source evaluated. | Review at least two candidate sources or log the content gap. |
| Outlink volume | A page exposes more than 150 crawlable internal links. | Review templates, facets, duplicates, and utility; the count triggers review, not automatic deletion. |
| Destination safety | Any new link resolves through a redirect, error, noncanonical duplicate, or retired URL. | Point directly to the approved canonical destination before release. |
Deliverable: the internal-link implementation register
The authoritative deliverable is a versioned table plus the post-release crawl export. Use one row per link decision, including rejected links when the rejection records an important boundary.
Cluster ID | Source URL | Destination URL | Source role | Destination role
Anchor text | Placement/section | Link purpose | Status | Implemented date
Owner | Validation crawl | HTTP result | Canonical result | Depth before/after
Commercial bridge? | Exception/rejection reason | Evidence
Attach three supporting views: an orphan disposition register, a priority-depth exception list, and a cluster graph. The graph is useful for inspection, but the row-level register remains authoritative because it records anchor, placement, owner, and verification.
The package is complete when a reviewer can select any priority page and answer: how is it reached, what does it link to, why do those routes exist, which cluster owns it, and when were the links last verified?
What goes wrong
- Links are added by keyword matching. A tool sees the same phrase on two pages and creates a link even though the reader tasks differ. Validate the destination against the sentence and the next question.
- Every page links to every sibling. The cluster becomes a dense list with no hierarchy. Restore the pillar-up, pillar-down contract and keep only useful lateral edges.
- “Read more” becomes the default anchor. The link gives no destination clue and repeated cards become indistinguishable. Name the destination or action.
- The audit counts sitemap URLs as connected. Crawlers may discover the URLs, but readers cannot navigate to them and relationships remain unstated. Require a crawl-followable inbound link for connection.
- Depth is flattened with irrelevant homepage links. The metric improves while navigation gets worse. Shorten routes through relevant hubs, categories, pillars, or strong contextual sources.
- Old URLs remain in body copy after a migration. Redirects hide the error and add unnecessary hops. Update internal sources to the final canonical URL.
- Commercial links are forced into unrelated education. The call to action interrupts the task and erodes trust. Bridge only where the product, service, or category genuinely advances the reader.
- Only low-performing pages are used as sources. The team links from pages it is already editing and ignores established pages with visibility and inbound authority. Review strong sources explicitly.
- Template links are treated as editorial endorsements. A footer or related-content widget creates reach but little contextual explanation. Give priority pages at least one relevant in-copy route.
- The spreadsheet closes before deployment. Suggested and approved links are mistaken for live links. A post-release crawl and rendered-page sample are mandatory.
Next phase
P13, Structured Data and Entities, needs the canonical inventory, cluster ownership, live hierarchy, and source-to-destination relationships produced here. It uses that evidence to describe what each page is about, which entity it represents, and how page types relate without contradicting the visible site.
Hand over the implementation register, final crawl, orphan dispositions, depth exceptions, and cluster graph. The next owner accepts the package only when canonical destinations are stable, unexplained orphans are zero, and schema decisions will not be built on proposed links that never shipped.
FAQ
How many internal links should a page have?
There is no useful universal count. Add every link that helps the reader continue a relevant task, then remove repeated, weak, or unrelated links. For this phase, every priority page needs at least one contextual inbound link and one useful contextual outbound link.
Should every spoke link to every other spoke in its cluster?
No. Every spoke should link to its pillar, while lateral spoke-to-spoke links belong only where the destination resolves the reader’s next question. A complete mesh creates noise and weakens the meaning of each link.
What anchor text should internal links use?
Use concise, natural language that names the destination or the task it completes. Vary wording when context varies, but do not force exact-match phrases or use vague labels such as “read more” when a descriptive phrase is available.
What is an orphan page?
An orphan page is an indexable page with no discoverable internal link pointing to it from another live page. A sitemap entry can expose the URL to a crawler, but it does not give a reader a route or explain the page’s relationship to the site.
How close should important pages be to the homepage?
Treat three clicks from the homepage as a review threshold for priority pages, not a law for every URL. Any priority commercial, pillar, or current campaign page deeper than three clicks needs a documented reason or a shorter useful route.
More tutorials in this section
Ready to put it into practice?
Free check · 7-day trial · no credit card