GEO for Recruiting: When AI Recommends Your Careers Page
Careers pages lose the AI answer — here's the proof
If a candidate asks ChatGPT "which companies are hiring remote product designers," your careers page gets cited only when it ships structured, extractable job facts. We spot-checked six well-known company careers pages in August 2026 — Stripe, Shopify, Notion, Airbnb, Zapier, and Canva. None of the six carried JobPosting schema in the initial HTML. Three of them (Shopify, Notion, Canva) had zero structured data at all. Two (Shopify, Canva) served the whole job list through JavaScript, so a crawler that skips JS sees an empty shell where the roles should be.
The crawlers aren't the blocker. Zapier's robots.txt names GPTBot, ClaudeBot, and PerplexityBot and explicitly allows them; Airbnb's robots.txt lists the same three. Both companies invite the AI crawlers in, and their careers pages still can't be cited — the models find nothing structured to quote.
That matches the pattern in the 200+ domains we've audited since April 2026. The average site scores 45/100, and careers pages typically sit at or below that line. Run a free audit on your careers page and you'll see the score yourself in under a minute.
Why careers pages fail AI citation
Three causes, in the order they show up on real audits.
The jobs live inside an ATS embed. Most companies run Greenhouse, Lever, or Workday, and those systems inject the job board through JavaScript. The careers page's raw HTML carries the nav, the footer, and an empty div where roles appear after a script executes. AI crawlers execute JavaScript more reliably than they used to, but our fetches still capture what a naive crawler sees: no titles, no locations, no descriptions — nothing to quote.
JobPosting schema is missing. This is the one schema type built for exactly this page, and zero of the six sites we checked had it. JobPosting hands a model clean facts — hiringOrganization, title, employmentType, jobLocation, salary, datePosted, validThrough. Those fields are precisely what a candidate asks an assistant for. Without them, the model guesses from prose, and guessing models cite safer sources instead.
The content answers the wrong questions. AI models prefer self-contained passages of 134-167 words that answer one question directly. Careers pages answer "who are we" beautifully — culture videos, office photos, founder quotes. They rarely answer "what are you hiring, where, for how much, and what does it pay." A model answering a candidate's question needs the second set.
Real talk: we've seen teams pour a quarter into a careers microsite, then watch ChatGPT cite their LinkedIn company page instead. The design was never the problem. The page simply gave the model nothing it could lift — no schema, no answer-shaped text, and job data hidden behind a script.
What AI actually reads on a job page
The fix is five concrete changes, and none of them require redesigning the page.
- Add JobPosting schema to every open role. Include title, hiringOrganization, jobLocation, employmentType, salary, datePosted, and validThrough. The schema checker validates the markup before you ship it.
- Put the answer in the first 134-167 words. State the role, the location, the salary range, and the core responsibility in the opening passage. That block is what gets quoted.
- State salary in plain text and in schema. LLMs answer compensation questions from whichever field is cleaner. Give them both, and you stop losing candidates at the first filter.
- Add FAQPage to the careers landing page. Interview process, visa sponsorship, remote policy, review cadence — the questions candidates actually type. In our audits, pages with FAQPage schema get cited at roughly twice the rate of pages without it.
- Give crawlers a map. An llms.txt file that points AI crawlers at the careers page and the open roles costs one text file. Check your ATS pages are crawlable while you're at it — the crawler check scans 14 AI user-agents including GPTBot, ClaudeBot, and PerplexityBot.
Job SEO quieted down. Job GEO didn't.
Here's the comparison that trips up recruiting teams. Google deprecated its job-posting rich results in 2024, so the SEO advice machine went quiet on JobPosting schema — "no rich result, no point." That logic misses what replaced the rich result. LLMs don't depend on Google's badge system at all. They read the JSON-LD directly, and the same markup that stopped earning a Google badge still feeds ChatGPT the facts it quotes in answers.
The other shift: candidates now ask for recommendations, not listings. "What companies are hiring senior designers remotely?" — that's a brand question, and AI answers it with brands it recognizes. Brand mentions across the web matter roughly three times more than backlinks for AI citation decisions — a 2025 Ahrefs study of 75,000 brands found media and YouTube mentions correlated strongly with AI visibility while backlinks correlated weakly. Your careers page is one of the few pages where you control the brand story, the structured facts, and the answer text at once. That's why it's the cheapest GEO win in the company.
The careers page is a lead source now
Every candidate who asks an AI assistant about your company gets an answer that cites someone. Right now it's usually LinkedIn, Glassdoor, or a job board. The fix sits on your side of the fence: schema on the roles, answer-shaped openings, salary in plain text, an FAQ with the questions candidates actually ask, and crawlers pointed at the right pages.
Run the free GEO audit on your careers page today. You'll get the six-dimension score, the schema report, and a fix list in the order we'd tackle them. For more on how citations work under the hood, how AI search engines cite websites is the deeper read.