I audited the backlinks to my own two sites and verified one. It lists the business in the wrong city. This is part two of the series I started in part one, and this is the baseline measurement.
How many backlinks do I actually have?
One inbound link that I could independently verify, across two live sites. The honest caveat is in the method: I checked pages I already knew mentioned the brands, so this is a floor rather than a total. Nobody without a paid link index can enumerate the web, and I would rather publish a verified floor than an unverified estimate.
Here is what I actually did. I searched for pages mentioning either brand, fetched each one, parsed its anchors, and checked whether a real link to my domain existed in the served HTML. Pages that render entirely in JavaScript got marked unverifiable rather than counted as absent, because a missing anchor in the served HTML is not proof of a missing link.
The number is low mostly because the number is low. But some of it is coverage, and pretending otherwise would be the same overclaiming I am about to complain about.
What is the one link?
It is a business directory I never submitted to, which scraped my details from somewhere else. It lists the business in Miami. The business is in Atlanta. And it links to the bare http version of the domain, which is not the canonical the site declares.
So the single inbound link I have is teaching search engines the wrong location and pointing at the wrong URL. That is worse than nothing in one specific way: a wrong location signal for a business that sells local consistency is an argument against me, sitting on the open web, being crawled.
I did not know it was there. I had been thinking about what to publish next while the one signal I already had was quietly wrong.
What did I get wrong about where to start?
I assumed the work was acquiring new links. Most of it turned out to be fixing and claiming things that already exist and cost nothing: profiles I own that link nowhere, listings that describe me incorrectly, and declarations on my own site pointing at URLs that do not resolve.
Three examples from one afternoon.
- My GitHub organization links nowhere. The page is public, indexable, rendered on the server, and has an empty website field. That is a followed link from a domain search engines trust, available for free, and I had never filled in the box.
- One site declares a company profile that returns a 404. The structured data tells engines “this brand is also that page,” and that page does not exist. As I write this it is still broken, because the right replacement URL belongs to an account I have not yet confirmed is mine, and pointing it at the wrong company would be worse than leaving it dead.
- Nothing points at the best thing I have made. The most linkable asset across both sites is the AI Citation Index, which is original data with an open method and published counts. Zero verified links point at it. I had never pitched it to anyone.
None of that is link building. It is housekeeping, and it appears to be most of the available upside right now.
Why did I not have more links?
Because I had published a reasonable amount and asked for nothing. Publishing is necessary and not sufficient. Nobody links to a post they have not seen, and I had no mechanism for anyone to see one.
I had told myself that good work gets found. It is a comfortable belief because it requires nothing of me. The audit made it hard to keep holding.
Do backlinks even matter for AI search?
Less than mentions, on the best evidence I can find. Ahrefs studied 75,000 brands and found branded web mentions correlate with AI Overview visibility at 0.664 on a Spearman coefficient, against 0.218 for backlinks. Correlation is not causation, and both numbers rise with brand size, so treat the gap as a direction rather than a mechanism.
Directionally it matches how these systems answer. An engine assembling an answer pulls from what it retrieves about you across the web, and being discussed in many places gives it more to retrieve than being linked from pages nobody discusses. That is a claim about retrieval surface, not about the model remembering you.
It does not make links worthless. It changes the order. Fix the entity first, so that when a mention happens engines can tell it is about you and not about a similarly named company. Then earn mentions. Whether links follow is one of the things this series is supposed to find out.
What am I doing first?
All of it free, in order: fill in the website field on the profiles I already own, correct or kill the directory listing with the wrong city, make my structured data point at URLs that resolve, and open a free Bing Webmaster account. That last one takes the audit from one scored factor to four, because it supplies referring domains, anchor text and a competitor comparison that I currently have no source for at all.
Then pitch the research nobody has seen.
If you have not done this on your own site, go and look before you plan anything. The measurement was the part I had been avoiding and it took an afternoon. Next in the series: whether any of it moves the number. I will publish that either way.