Author Signals That Still Work and the Ones Google Retired

Author markup occupies a strange place in SEO memory. Half the industry remembers it as a ranking signal Google killed; the other half treats "author authority" as a live optimization target. Both halves are working from a blurred version of what actually happened.

The history is worth getting right, because the retired system and the signals that plausibly work today are different mechanisms, and conflating them leads to cargo-cult markup.

What rel=author actually was

In 2011 Google began supporting authorship markup: a rel="author" link from an article to an author page, chained to a Google+ profile that linked back. When the loop verified, Google could show the author's photo and byline next to the result.

Two things about that system are commonly misremembered. First, it was primarily a display feature — a rich snippet — not a documented ranking input. Second, the verification depended on Google+, which is why the system's fate was tied to that product's.

What Google retired, and what it said

The shutdown came in two steps, both publicly announced. In June 2014 Google removed author photos from search results. In August 2014 it dropped authorship display entirely, with John Mueller stating the markup was no longer being processed. Google's stated reasons were low and error-prone adoption by webmasters and little observed benefit to searchers from the feature.

So what was retired was: the rel=author verification chain, the profile-photo rich result, and Google's processing of that specific markup. The rel="author" attribute still exists in HTML; Google has simply said it does not use it. Adding it today is inert.

What "author rank" was, and never became

Alongside authorship ran a persistent idea called "author rank" — that Google scored authors and ranked their content accordingly. This came partly from a Google patent commonly referred to as Agent Rank, which describes scoring content by the reputation of the "agents" who signed it, and partly from executives musing publicly about identity and content quality.

Here the documented-versus-inferred line matters. The patent is documented. That the described system was ever deployed as a ranking signal is not — Google patents things it never ships, and Google representatives repeatedly denied that authorship markup carried ranking weight. The reasonable reading of the record is that Google explored per-author scoring, found the identity-verification problem hard, and retired the visible experiment.

Which author signals plausibly work now

Several author-adjacent mechanisms are alive, with varying evidence behind them.

Structured data. The author property in Article schema is documented and current. Google's documentation asks for it, recommends filling out author.name and author.url, and uses it to understand who wrote what. That is an understanding function, not a promised ranking function — but it is markup Google actively consumes, unlike rel=author.

Entity reconciliation. Google's Knowledge Graph contains people, and some authors exist in it as entities. An author page that links consistently to the same profiles (via sameAs markup and plain links), with a stable name used across publications, gives Google what it needs to reconcile mentions into one entity rather than several ambiguous ones. That reconciliation is documented behavior for entities generally; that it moves rankings for the author's articles is inference.

Quality rater guidance. Google's Search Quality Rater Guidelines explicitly instruct raters to research the reputation of content creators, not just websites, when assessing E-E-A-T. Raters do not set rankings directly — Google has documented that their judgments train and evaluate systems rather than act on individual sites — but the guidelines are the clearest statement available of what Google wants its systems to reward. Author reputation is named in them.

Bylines in news. Google's publisher guidance recommends clear bylines, author information, and dates on news content. For publishers, this is documented best practice from Google itself, even though it is framed as transparency rather than as a ranking lever.

What this does not support

The live mechanisms above do not add up to "build author pages and rankings follow." No documented signal guarantees that. They also do not support buying bylines from recognized names as a shortcut: to whatever extent author signals operate, they plausibly operate through entity reconciliation and reputation earned elsewhere, not through a name string appearing on a page. A famous byline atop weak content leaves you with weak content.

Treat author investment the way you would treat any contested signal: the cost is low, the downside is near zero, the upside is plausible but unproven, and some of the benefit — reader trust, and legibility to AI search systems that cite sources — accrues regardless of what Google's ranking systems do with it.

What to do

1. Remove or ignore rel=author. It is not processed. Effort spent there is wasted.
2. Mark up authors in Article schema. Include author.name and author.url pointing to a real author page, per Google's current documentation.
3. Build one canonical author page per writer. Bio, credentials, links to their work, and sameAs links to their profiles elsewhere. Link every byline to it.
4. Keep names consistent. One spelling, one form, across your site and ideally across the web. Entity reconciliation works on consistency.
5. Put real expertise signals on the page. Raters are told to look for who wrote the content and why they should be trusted. Make that answer findable in seconds, because the systems trained on rater judgments are aimed at the same target.
6. Do not report author markup as a ranking initiative. Frame it internally as machine-readability and trust work with a possible search upside — that is what the evidence supports.