In 2023, I published a small experiment about a surprisingly practical SEO question: when one page links to the same destination several times, which anchor text does Google keep? The results suggested that Google selected the first text anchor and, separately, an image anchor. This site documents my attempt to repeat that work with tighter controls, a navigation comparison, and two small tests for source order and URL fragments.
Why selective link priority deserves another look
Links do several jobs at once. They help people move through a site, help crawlers discover URLs, distribute internal authority, and provide descriptive context through anchor text. The complication appears when a source page links to one target URL more than once. A header, a card image, an introductory sentence, and a call to action may all point to the same place while using different words.
Historically, SEOs called this “first link priority,” but that label is too confident. Google has said that its extraction systems can select one, several, or all links, and that the behavior can change. I prefer “selective link priority” because it describes the observation without turning a limited test into a universal law.
The original study produced three simple patterns: image then text; text, text, then image; and text, text, image, then another text link. Google Search Console appeared to preserve one text anchor and one image anchor. I want to see whether the text-link portion of that result still reproduces and whether modern reporting lets us observe the image portion at all.
A deliberately small replication
I am keeping the test on six useful pages rather than manufacturing a large collection of thin targets. Four pages preserve the original replication chain, one page acts as a measurement canary, and one page explains HTML order and fragment links. Every page is independently useful; the controlled links sit inside content I would publish even without the experiment.
The companion copper-anchor implementation notes explain why linked images require separate accessibility and SEO decisions. This is the second—and final—link from this page to that destination.
The navigation question is important, but secondary
The new extension asks whether a repeated header link and a contextual body link to the same homepage are reported differently. I am not letting that question dominate the study. Three configurations repeat the original work; one configuration explores navigation. On each participating interior page, the navigation link appears before the article, while the body link appears naturally in the prose. Both anchor phrases occur the same number of times across the site, so frequency is not confused with page region.
This does not answer every navigation question. It does not compare breadcrumbs, sidebars, footers, or mobile clones. A separate condition now isolates one narrower issue: whether HTML source order or CSS visual order better predicts the reported anchor.
The fragment crossover starts here with two links to the same document. The indigo document-route reference uses the base URL first. Later in the same paragraph, the willow section-route reference adds a fragment identifier. A second source page reverses that order.
The measurement gate
The original experiment relied on Search Console’s Top Linking Text report. Current documentation creates two reasons for caution. First, the report is sampled and not comprehensive. Second, image links are now documented as “empty” in the report, even though Google’s link guidance says an image’s alt attribute can function as anchor text. That means a modern replication may be able to observe text-anchor selection while remaining unable to identify which image alt text Google processed.
Before I treat any result as evidence, I will use the canary page to check whether links from sibling URL-prefix properties appear as external links and how image links are represented. The canary receives one text link from this page: Search Console canary text signal. A separate linked image points to the same canary from another source page.
What would change my mind
I am preregistering a narrow expectation: the first text anchor will be the text anchor most likely to appear for each replicated configuration. I am less confident that image alt text will be observable through the current report. For the navigation extension, I am genuinely uncertain. The header anchor may win because it appears first, the contextual anchor may win because it is embedded in the main article, or both may appear.
A failure to reproduce the older pattern would be useful. It could mean the underlying extraction behavior changed, the reporting layer changed, or the site-level context matters more than the original design captured. The correct response would be to separate those explanations—not to force a clean conclusion from ambiguous data.
How I will report the outcome
I will preserve the deployed HTML, record Googlebot requests from server logs, verify canonical selection, and export Search Console data on a fixed schedule. An anchor counts as observed only after it appears in two consecutive exports. “Not observed” will remain distinct from “ignored.” If the implementation changes after launch, the change will be dated publicly.
The practical goal is modest: give site owners better evidence for choosing the first useful anchor on a page, writing descriptive alt text, and avoiding accidental duplicate navigation markup. Selective link priority is rarely the biggest SEO problem on a site, but it can be a meaningful tie-breaker when templates create many competing links.
Background: the original Zyppy selective link priority study; Google’s current link best practices; and Search Console’s Links report documentation.