How do you show up in ChatGPT answers as a consultant? You build citable material on the four surfaces AI checks before recommending a person: content under your name, third-party mentions of you, directory and community presence, and quotable one-liners tied to your name. AI cannot recommend someone it cannot quote. That is the whole game, and almost nobody is playing it at the individual level. Every GEO guide I have read optimizes a company website. Meanwhile the buyer is typing "who should I hire to fix my B2B pricing" into ChatGPT, and the answer names people.
Buyers now ask AI who to shortlist. AI assembles those answers from citable public material, and per Profound's 2026 citation reports (1.4M citations analyzed), LinkedIn is the #1 cited domain for professional questions. This post gives you the person-level playbook: the four surfaces framework, a what-to-do table, and my own accidental proof, one LinkedIn comment that became a cited source in AI search.
The shortlist moved inside the chat window
A buying decision used to start with "ask three friends, Google the rest". Now a growing share of it starts with a 20-word question typed into an AI assistant. I first took this seriously listening to Ethan Smith on Lenny's podcast talking about answer engine optimization. Two things stuck. First, for an early-stage brand he recommends investing in AEO before SEO, because AI queries can be 20+ words long and specific enough for a small brand to win. Second, the channels that get indexed by LLMs are not the ones marketers obsess over: YouTube, Reddit, and notably, LinkedIn articles get indexed better than posts.
Everything in that conversation was framed around companies. But the logic applies harder to consultants, because a consultant IS the brand. And the data says the raw material for those answers already lives where you post. Per Profound's 2026 citation reports, built on 1.4 million analyzed citations, LinkedIn is the number one cited domain for professional questions across major AI platforms. Inside LinkedIn, the mix is shifting toward what you say rather than who you say you are: feed-post citations rose from 20.9% to 26%, while profile citations fell from 33.9% to 14.5%. Your resume is losing citation share. Your takes are gaining it.
The four surfaces AI checks before recommending a person
When an AI assembles "here are consultants worth talking to", it is pattern-matching across public, attributable material. I think of it as four surfaces. Miss one and you are still findable. Miss three and you do not exist.
- Surface 1: content under your name. LinkedIn posts and articles where you make specific, quotable claims about your niche. Not "5 tips" filler. Positions with numbers attached. This is the surface the Profound data covers, and it is the one you control completely.
- Surface 2: third-party mentions. Other people's posts naming you, podcast appearances, "best fractional CFOs" listicles, newsletter shout-outs. AI weighs what others say about you more than what you say about yourself, exactly like a human reference check.
- Surface 3: directory and community presence. Reddit and Quora threads in your niche, industry directories, review platforms. Ethan Smith's rule here is the right one: use your real name, add genuinely thoughtful answers, and do not shill your own brand. Fake accounts and drop-your-link comments are how you get pattern-matched as spam.
- Surface 4: quotable one-liners with your name attached. The smallest unit of citability: a sharp, specific sentence that answers a real question, in public, under your byline. Comments qualify. Which brings me to the thing I found by accident.
The comment that got cited
I discovered, by accident, that one LinkedIn comment I published through my own pipeline became a cited source in AI search. Not a post. Not an article. A comment.
It should not have surprised me. A comment is a named, public, timestamped statement answering a specific question. That is precisely the shape AI answers are built from. And comments are the cheapest surface-4 asset you can produce: I get roughly 30K impressions a week from about 5 comments a day, at 10-15 minutes total.
Comments do double duty as outbound too. When I suggested to an SDR team that they comment on prospects' posts before cold outreach, the answer was blunt: "We are measured on outbound, not inbound." Two weeks later one of them tried it anyway. His response rate doubled. A cold email from a name the prospect saw in their comments yesterday is not a cold email. I have had prospects recognize me from my comments before intro calls, before I said a word about what I do. That recognition and AI citability are the same asset viewed from two angles: a public record of you being useful, in your own words, on the questions your buyers ask.
The playbook, surface by surface
| Surface | What to do | Effort |
|---|---|---|
| 1. Content under your name | 2-3 LinkedIn posts a week with specific claims and numbers; turn your best posts into LinkedIn articles (they index better in LLMs) | ~2 hrs/week with a capture system |
| 2. Third-party mentions | Pitch 1 podcast or newsletter a month; make your frameworks nameable so others can cite them | ~1 hr/month |
| 3. Directories & communities | Answer real questions on Reddit/Quora in your niche under your real name; claim relevant directory listings once | ~30 min/week |
| 4. Quotable one-liners | ~5 thoughtful comments a day on posts your buyers read; one sharp claim per comment, no pitch | 10-15 min/day |
Effort estimates are from my own weekly routine, not aspirational.
What not to do: flood the zone with AI slop
The obvious shortcut is to have AI mass-produce surface-1 content. The data says the shortcut backfires. Ethan Smith's team actually tested it: there is now more AI-generated content online than original content, and the AI-generated pages produced inferior search results, not better ones. On LinkedIn specifically, Originality.AI's 2025 study of 3,368 posts from 99 influential profiles classified 53.7% of long posts as likely-AI, and found likely-AI posts underperform human-written ones in most professional sectors, by up to roughly 80% in strategy and innovation topics.
I watched the same movie in early SEO. People duplicated millions of keyword-stuffed pages, and Google cut the middlemen out in a day. The same correction is coming for generic AI content. My position: AI-assisted wins, AI-generated loses. Use AI to polish and scale your actual thinking, your client stories, your contrarian takes. The originality has to come from you, because originality is what gets cited.
The honest part: this compounds in months, not days
There is no ad budget for getting a person into AI answers. The material accumulates, the citations follow. My cited comment surfaced long after I wrote it, and I only found it by accident. What shows up early is the human layer: prospects recognizing your name within weeks of consistent commenting, warmer replies, shorter intro calls. One of my customers watched his LinkedIn SSI climb from 45 to 57 in two weeks of commenting. The AI layer builds on top of that, quietly, while the average feed gets worse around you: AuthoredUp's analysis of 3M+ LinkedIn posts found median impressions fell 47% year over year (1,211 in June 2024 to 636 in May 2025). Reach is decaying. Citability is not. Play the second game.
Run the four surfaces without the grind
You can execute this playbook with a notes app and discipline. The failure mode is volume: four surfaces, every week, in your own voice, forever. That is the part I productized. Liftli runs inside the AI you already use (Claude today, ChatGPT and Cursor next), learns your extracted voice, and drafts posts and comments from your voice notes and calls. Nothing publishes without your one-tap approval. And because LinkedIn's User Agreement prohibits extensions and bots acting on your account, Liftli publishes and schedules through official APIs, gated by your one-tap approval. Free tier, no card, from $29.