I searched my name recently and my Google Knowledge Panel was gone. Not smaller. Gone. My entity had split into a bunch of scattered, conflicting versions of me, with really low confidence scores, 24 in the case of the one that should have been me. For reference, I had been sitting above 200. And nothing on my end had changed. If anything I was doing better than ever, repurposing all my videos into articles.
The fix came from a Facebook post
That’s when I came across a Facebook post from Dennis Yu about fixing your Wikidata to improve your Google Knowledge Panel. He wrote a full article on the fix as well. I implemented it, and I’m back up to a 208 confidence score. The highest I think I’ve ever been is around 250, so that’s pretty much where I was before everything fell apart. The panel itself is back too, with the overview, my age, and my social profiles. It’s still missing a little bit, but it’s in a way better place.
What Wikidata actually is
Wikidata is a giant database of entities and facts. Google and the other search engines pull from it when they build Knowledge Panels. Every entity gets its own Q number; mine is Q138557395. Every fact on an entity is stored as a subject, a property, and a value, which is what makes the whole thing machine readable.
Going through every single claim
If you don’t have an entity, you create one. I had to create mine originally, and that’s the case for a lot of the people we work with. If you do have one, you go through every statement on it and make sure each claim points to the right thing and has a reference backing it up. One example from mine: the Minnesota Dunk Squad, the dunk squad I created, is on my entity as a sports team, and the claim references the Dunk Squad website.
Then you fill in the missing properties. I’m a human male in the United States. My name is Dylan Haugen. My birthday is on there. My occupations are on there. I actually speak Spanish, which I don’t think I had said publicly before this video. The goal is to get every fact about you into that machine readable format, each one with a reference showing it’s real. And the cool thing is you can have Claude do essentially all of it.
Person schema ties it to your website
Once the Wikidata entity is solid, the last piece is on your own site. We added Person schema to my personal brand site with a sameAs property pointing to my Wikidata URL, my LinkedIn, my Instagram, and the rest of my profiles. Claude handled that through Rank Math on my site, and there are multiple ways to do it. The point is that Google can now see the website, the Wikidata entry, and the social profiles as one entity instead of fragments.
I didn’t change anything else, which is what makes this a clean result. The Wikidata repair and the schema are the whole difference between a 24 and a 208. We’ve implemented the same thing for a lot of our clients, giving them Wikidata and applying it as schema on their websites, and seen some very good improvements, like the builds we documented for Trenton Sandler and Brooke Lance. The full case study version of this fix is on BlitzMetrics.
Look yourself up
We built a tool called the Knowledge Graph Explorer that searches Google’s Knowledge Graph and shows you the entities on your name, with the confidence score and your KGMID, which is kind of like your Social Security number on Google. If your panel has ever disappeared, or never showed up at all, start there.
From there, you can literally hand this article, or the video above, to Claude and say: check if I have a Wikidata entity, optimize it if I do, create one if I don’t, and then apply it as schema on my site. That prompt is basically the whole fix. It’s the same reason my whole article pipeline runs through Claude now. This stuff used to take weeks of manual work, and now it doesn’t.
About The Author
