Internal linking is one of the few search improvements you can apply to an existing site without waiting for a new backlink, a new platform, or a new content campaign. The durable method is simple: use real performance data to choose pages, add links that help a reader continue the task, log exactly what changed, and compare the same evidence set afterward.
This guide adapts the useful workflow in Borja F. Autiero's internal-linking article and checks its claims against Google's current documentation. The workflow is useful. Some of the examples are observations and heuristics, not proven ranking effects.
What Google actually documents
Google says it generally discovers pages through links from pages it already crawled. It recommends crawlable <a href> links and says anchor text helps people and Google understand the destination. Good anchor text is descriptive, concise, and relevant to both the source page and the page it links to.
Source: Google link best practices.
Google also documents Search Console's Performance report as a way to examine search traffic by queries, pages, countries, and other dimensions. Its position metric is the average topmost position for your site across impressions, so it should be interpreted at the relevant query and page level rather than treated as one exact rank.
Source: getting started with Search Console, Performance report documentation, and click, impression, and position definitions.
What the source article observed — and what it did not prove
The article reviewed 11 public Mailchimp articles. All 11 linked to educational pages, seven linked to product pages in the article body, and all 11 exposed product links in the footer. That is a useful example of one site's linking pattern. It is not a controlled test of ranking changes, and it does not establish a universal link count or template.
The recommendation to prioritize pages with average positions 11-20 is also a triage heuristic. It can surface pages that already receive impressions and may deserve attention, but Google does not document “page two” as a special internal-link threshold. Diagnose the page, query intent, content quality, competing results, and conversion value before changing it.
The measurable workflow
1. Freeze the comparison frame
In Search Console, choose the property, search type, date range, country, and device view you will use. Export the baseline. If you later compare periods, keep those filters consistent and note seasonality, site releases, and other changes that could affect the result.
A practical review starts with the latest three months for orientation, then uses a like-for-like comparison such as the latest complete 28 days versus the previous 28 days. The exact window matters less than preserving the same definitions.
2. Group pages by the reader's job
Cluster pages by the question or decision they help a visitor complete, not by a shared word alone. Planning labels can help:
- Pillar: the broad guide that organizes the topic.
- Support: a focused answer, example, checklist, or technical explanation.
- Money page: a product, service, audit, demo, or other commercial next step.
These are editorial labels, not Google-defined page types. A link belongs when the destination is the next useful step for the reader.
3. Build source-to-destination opportunities
For every proposed link, record five things:
- The source URL.
- The destination URL.
- The existing or proposed sentence that will contain the link.
- The descriptive anchor text.
- Why the destination helps the reader at that moment.
Verify that both URLs are real, canonical, indexable where appropriate, and not already connected by the same contextual link. Do not manufacture a sentence solely to hold a keyword-rich anchor.
4. Write anchors for understanding
Use words that tell the reader what they will get after clicking: “Google's link best practices” is clearer than “read more.” Exact-match repetition is not the goal. Natural variation comes from different sentences and reader needs; stable navigation labels can remain consistent.
5. Keep global navigation selective
Headers and footers can expose a small set of genuinely important destinations, but they do not replace contextual links inside useful content. Check both desktop and mobile rendering, and avoid turning the footer into an undifferentiated directory.
6. Log the release
Record the source page, destination, anchor, date, deployed revision, and baseline evidence. Confirm the exact links render in production. A committed change, preview, or successful crawler request is not proof that search performance improved.
7. Re-measure without claiming false causality
After enough complete-period data exists, repeat the same Search Console filters. Review relevant page-query pairs, impressions, clicks, click-through rate, and average position. Also inspect qualified conversions so a traffic gain does not hide worse buyer fit.
A before/after movement is an observation, not automatic proof that the links caused it. Algorithm updates, new backlinks, content edits, demand, competitors, and seasonality may all contribute.
How this connects to AI search
Descriptive internal links strengthen the site's navigable content graph and help retrieval systems reach related evidence. They also help a person move from an answer to a comparison or next step. See how that retrieval layer fits the broader modern search optimization stack.
No major platform documents internal linking as a guaranteed AI-citation lever, however. Treat AI visibility as a separate observation: capture the engine, prompt, date, mode, cited URL, and downstream referral or conversion.
Discovery is not ranking. Ranking is not citation. Citation is not recommendation. Recommendation is not conversion. Internal links can support the path; they do not prove the outcome.
A compact operating checklist
- Export a fixed Search Console baseline.
- Map canonical pages by reader intent and business value.
- Identify contextual source-to-destination opportunities.
- Write descriptive anchors inside useful sentences.
- Verify the links and mobile experience on the deployed revision.
- Log every applied change.
- Compare the same filters after a complete measurement window.
- Report observed movement, confounders, and unknowns separately.
Bottom line
The “hack” is disciplined information architecture. Use performance data to prioritize, reader intent to decide, descriptive links to connect, and a change log to learn. That makes internal linking useful even when the result is less dramatic than a viral promise.