Viveka When someone needs a LinkedIn strategist or a visibility consultant now, there’s a decent chance her first stop is ChatGPT or Perplexity instead of Google, and the answer she gets back has actual names in it. LinkedIn content shows up in about 11% of AI answers, second among all cited domains, and what the engines cite is mostly profiles and long articles rather than whatever went viral this week. I pointed the machines at my own name, and they confidently introduced a version of me from years ago. Here’s the data, what I found, and how to fix what they say about you.
Do people really ask AI who to hire?
They do, and there’s finally real data on it. Semrush analyzed 325,000 prompts and the 89,000 LinkedIn URLs cited in the answers across ChatGPT Search, Google AI Mode, and Perplexity, and found LinkedIn showing up in about 11% of AI responses on average, which makes it the second most-cited domain in the whole study. (Semrush sells SEO software, so a study concluding that search visibility matters is also marketing for Semrush. The dataset is real and enormous, though, and nobody else has measured this at that scale yet, so it’s the best flashlight we currently have.)
When the question is “who should I hire,” the engines assemble an answer out of whatever they can find and cite: profiles, articles, posts, podcast pages, conference bios, old team pages. That last one matters more than you’d think, and I’m about to show you why.
What did the machines say when I checked my own name?
They introduced me as a person I stopped being years ago. I had Claude run the check for me, which is exactly the method I teach, so you’re watching it work: search my name the way a prospective client would, then search the kind of hire my actual clients want to make, and report back what surfaced.
Searching my name plus “LinkedIn expert” returned seven main sources. Exactly one of them was my current LinkedIn profile. The rest were my old company’s team page, a bio from the training company I started back in 2006, podcast interviews from years ago, and my book LinkedIn Marketing: An Hour a Day. The AI summary then confidently introduced me by a title I haven’t held in years, at a company I left, as if no time had passed at all. (Cool cool cool.)
And the website with my current offers and my current thinking, the one I redesigned this very month? Nowhere in the answer. The one bright spot besides the profile: when Claude ran the hiring question my real clients would ask, about LinkedIn help for women founders over 50, my Substack surfaced. The essays are getting picked up. The site isn’t yet.
If you’ve been building expertise for twenty or thirty years, expect your version of the same result: the machines have plenty to cite about you, and most of it points at who you used to be. That’s not a punishment, it’s just how data ages. The internet remembers your loudest years, and if your loudest years happened two positionings ago, that’s the version of you it introduces to your next client. I talk about reinvention as reassembly, taking the puzzle of your expertise apart and building a different picture from the same pieces, and what I found is that the machines are still showing everyone the old picture. (I made this worse for myself, honestly. After I left the LinkedIn training world I ignored that whole side of my expertise for about a year, which means I stopped generating anything new for the machines to find. The longer version of that story is here, and I don’t recommend the ignoring-it strategy.)
Why do posts with 20 reactions get cited when viral ones aren’t?
Because the engines reach for whatever answers a question cleanly, and by that standard the winning material is modest: the median LinkedIn post cited by an AI engine has 15 to 25 total reactions and no more than one comment. The format they cite most is the long-form LinkedIn article of 500 to 2,000 words, which almost nobody writes anymore because articles stopped producing big feed numbers years ago, and I want more women to know both of those facts before they grade another week of content by its reaction count.
On ChatGPT Search and Google AI Mode, 59% of LinkedIn citations go to individual people rather than company pages. (Perplexity flips it, favoring company pages at 59%, so if you sell through a company brand, that’s the engine to watch.) For a solo expert, the assets in play are your personal profile and your articles, both of which you control completely.
I have watched women delete posts because they “only” got 12 reactions, and I want to reframe that number for you: reach is one game, and being part of the answer when a machine gets asked a hiring question is a different game that never shows up in your analytics at all. Some of your least applauded work may be doing the most patient client-getting work you own.
What should you do about it?
Four moves, in this order: read your profile the way a machine would, write long-form articles again, put your current positioning somewhere you own, and keep receipts. Here’s each one.
1. Read your profile the way a machine would. Your About section should say what you do now, for whom, in the words a client would actually type into a chat box. If your headline still leads with a role you’ve moved past, that’s the first fix, it takes an afternoon, and rebuilding profiles is literally what I do, so I can tell you the afternoon pays for itself.
2. Write articles again. 500 to 2,000 words, one clear question as the title, the answer delivered in the first two sentences under each heading. (You’re reading that structure right now. The headings in this piece are questions on purpose.) You don’t need fifty of them. You need a handful that answer the exact questions your best clients ask in a first call.
3. Put your current positioning somewhere you own. A site, an about page, a body of essays that are true this year. My Substack surfacing in that hiring search while my brand-new site sat invisible tells you the machines reward accumulated, consistent writing over freshly launched anything, so the sooner you start the pile, the better.
4. Keep receipts. Ideas I was writing about two and three years ago now come back to me in AI answers with nobody’s name attached, which is irritating in a way I’ve decided to treat as motivating. Your thinking, written down, with your name on it, time-stamped, is the only claim you have. The women I work with have decades of it sitting in talks, client calls, and half-finished drafts, and none of that counts until it’s published somewhere a machine can read.
Then run the check on yourself. Open ChatGPT or Perplexity, ask who you are, and then ask the hiring question your best client would ask. Five minutes, and you’ll know whether the machines are introducing the current you or the 2014 you. (If you want to see how I set up Claude to run checks like this, I compared the two big work assistants here.)
Is this just SEO hype with a new name?
Some of it absolutely is. The AI-visibility industry is going to sell a great many audits this year, and most of the panic content exists (as always) to raise your cortisol. I’m neither a hype-bro nor a doomer on this one, and the honest middle looks like this: one very large vendor study, a real shift in how people ask for recommendations, and a free check you can run on yourself in five minutes. You don’t need to buy anything to act on it. You need an accurate profile, a few real articles, and a pile of receipts with your name on them, which is work worth doing even if the machines all vanished tomorrow.
So go ask the robots who you are, and then go make the answer true. Keep your voice, and keep your freaking receipts.