Hey, it's Amit here.

I ran an audit on myself recently and it told me something I didn't want to hear.

Google has an "Amit Raj" in its books. It's mine. But it sits at rank 8 of 10 results for that name, with a score of 24, and not a single fact bound to it  -  no description, no link, no role. The top result for "Amit Raj" is Amit Thackeray, the Indian politician, at a score of 565.

In classical SEO terms that's nothing. I rank fine for "Amit Raj link building" and "The Links Guy" and the queries that actually bring people to me. The thing the audit was checking sits a layer below that. It used to be the kind of subject you'd hear an entity nerd talk about at a conference and quietly file under "later."

To be clear, I haven't been ignoring this work either. 

The pieces are there. They just weren't stitched together at the structured layer Google reads from. That's the gap — and it's the reminder that this isn't one fix. It's a lot of pieces filling in a larger picture.

There's a layer of Google that isn't pages and isn't links. It's a machine-readable catalogue of who and what exists, with facts attached. People, organisations, places, products. Each "thing" has attributes (occupation, founded date, headquarters, social profiles) and relationships (founder-of, employed-by, located-in).

Picture it as the shift from "string" to "thing."

If Google sees the string "Amit Raj" on a page, that's just text. If Google has a thing for me — with my occupation, my employer, my field of work, the topics I'm bound to — then that string resolves to a specific person it knows things about. Two different worlds. The page above me in the SERP isn't necessarily "better content." It might be that Google can resolve the author into a recognised thing and can't resolve mine.

The Knowledge Graph and the Knowledge Panel get used interchangeably and they shouldn't be. The graph is the database underneath. The panel — that card on the right of a branded search — is the visible output.

They're correlated, not identical. Some brands get a panel from a strong Google Business Profile alone, without ever doing entity work. Others have solid graph entries and never get a panel rendered. Working on the graph is the most reliable path to triggering a panel. Not a guarantee.

Craig Campbell did a podcast with Bogdan Bratis where they get into exactly this — about 49 minutes in. Bogdan talks about getting his own Knowledge Panel to render, with what he describes as a few small tweaks. The reason it landed so quickly was that his graph was already strong. He'd been on X Factor Romania as a singer. Press, public information, his name bound to a real career across the web. The panel was sitting under the surface waiting for a nudge. He just nudged it. If your graph is thin to begin with, the same tweaks don't trigger anything. Graph first, then panel.

Up to about 2024, getting a panel was the main visible benefit of being in the graph. Nice to have. Not critical. You could rank, traffic could flow, business could happen, without ever appearing in it.

Three things happened more or less at once:

One. AI search systems — ChatGPT Search, Perplexity, Gemini AI Overviews — became the front door for a real share of decision queries. 

Not all queries. Specifically the "who should I hire / which tool should I use / who's an expert in X" ones. The high-intent ones.

Two. Strip those systems back to first principles and they're doing the same job classical search has always done, just with more confident summarisation: resolve names and concepts into structured entities, then reason about them. 

ChatGPT, Perplexity, Gemini, Google's own search — they store things at different certainty levels, but underneath the surface they all need to know who you are before they'll surface you. A loose page reference isn't enough.

Three. Entity trust became the gating signal for being cited at all. 

If the system can't resolve you to a known thing with facts attached, it can't risk naming you. The hallucination cost is too high. So you don't get cited. You don't get named. You don't exist in the answer.

This is why AI search visibility audits are catching people a bit off-guard this year. They look at classical rankings, see strong positions, assume things are fine. Lets say an agency asked Perplexity for the "best link building agencies in the UK" and they're not in the answer. Their content is being read. Their pages are indexed. Their author entity isn't recognised.

I built the audit tool because I wanted to see this clearly on myself before talking about it. Then I bundled it so anyone could run the same check on their own brand.

There's no clean playbook here either. Same as the engagement issue I wrote a few weeks back  -  it's a stack of different bits that compound. 

It’s nothing fancy and all of it could take months to give you the final end benefit.  Most of it overlaps with the off-page work you're already doing — it just demands a structure on top.

The acid test is simple. Re-run the audit ninety days later. If you've done the work, the resultScore for your entity should move from the 20s into potentially triple digits, and you should appear higher in the result list. If Perplexity starts naming you in answers it wasn't naming you in before, the binding is taking hold.

When did you last check your own recognition status — yours or your authors' — with a tool, not a vibe? If you're like I was a month ago, the answer is "I haven't."

Amit Raj

The Links Guy

P.S. AI search isn't replacing classical SEO. It's reading the same web we always optimised, then summarising it through a different lens. The entity work is the part that scores well in both lenses at once. That's why it's worth doing properly, not as a side-quest.