Why ranking in Perplexity and ChatGPT is a different game
Google ranks pages. Perplexity and ChatGPT rank answers, and then attribute those answers to sources. That distinction changes everything. A page that ranks #3 on Google can still get zero traffic from an AI Overview if the model does not trust it enough to cite. Meanwhile a page ranked #14 with tight, well-structured, entity-grounded content can end up as the single citation ChatGPT hands the user. Generative engine optimization — GEO — is the discipline of engineering that outcome on purpose.
For local service businesses this is a bigger deal than most owners realize. When someone asks Perplexity 'who is the best emergency plumber in Bellevue,' the model returns two or three named companies with links. If you are not one of them, you do not exist in that conversation. There is no page two.
Traditional SEO vs. GEO: the signals actually differ
Classic SEO optimizes for a ranking algorithm that scores pages against a query. GEO optimizes for a language model that has to decide, in real time, which sources to trust enough to quote. The two overlap, but the winning signals are not the same.
- SEO rewards keyword coverage. GEO rewards clean, self-contained factual answers a model can lift verbatim.
- SEO rewards backlinks. GEO rewards being consistently described the same way across independent sources — a strong entity graph.
- SEO rewards long dwell time. GEO rewards structured data that unambiguously states who you are, where you serve, and what you do.
- SEO tolerates fluff intros. GEO punishes them — the model skips past padding and cites competitors who lead with the answer.
The four pillars of GEO for local businesses
1. Entity grounding
LLMs need to be confident about who you are before they will cite you. That means one canonical name, one address, one phone number, one service description — repeated identically across your site, Google Business Profile, Bing Places, Yelp, Apple Business Connect, industry directories, and every citation site that matters in your vertical. Inconsistency is the single fastest way to get filtered out of AI answers, because the model resolves ambiguity by picking someone else.
2. Structured data that leaves no room for interpretation
Schema is how you hand the model a machine-readable version of your business. For a local service brand the baseline is LocalBusiness (or the closest subtype — Plumber, HVACBusiness, Attorney, RoofingContractor), plus Service schema for each offer, FAQPage for question-led pages, and BreadcrumbList for site structure. Every entity needs sameAs links pointing at your GBP, LinkedIn, and top citation sources. This is what lets Perplexity and Google's AI Overview stitch your identity together instead of guessing.
3. Answer-first content architecture
Every important page should open with a direct, quotable answer to the question that page targets — one to three sentences, before any storytelling. Then expand. Models scan for the tightest passage that answers the query and cite that. If your first 60 words are a brand introduction, you have already lost. Write the way a Wikipedia editor would: definition first, evidence second, nuance third.
4. Earned citations from trusted local sources
LLMs weight sources they have already learned to trust. For local businesses that means local newspapers, chamber of commerce pages, industry associations, .gov and .edu mentions, well-established review platforms, and niche vertical publications. One mention in the Seattle Times or a state contractor board carries more weight in the model's decision than fifty low-quality directory links. Chase the mentions, not the DoFollow.
A step-by-step 90-day plan
Weeks 1–2: audit and fix your entity
- Pick one canonical NAP (name, address, phone) and correct every mismatch across the top 30 citation sources for your vertical.
- Rewrite your GBP description, categories, and services to match your website exactly.
- Add or upgrade LocalBusiness schema on your homepage with complete sameAs links.
Weeks 3–6: rebuild your top pages answer-first
- List the 20 questions your customers actually ask a model ('best roofer in Kirkland,' 'how much does a heat pump cost in Seattle,' 'do I need a permit for a deck in Bellevue').
- Give each question its own URL. Open with a 40–80 word direct answer. Follow with detail, local context, and one honest tradeoff paragraph.
- Wrap each page in FAQPage schema for the sub-questions and Service schema for the offer.
Weeks 7–12: earn citations and measure
- Pitch one local publication per week with a genuinely useful angle — data from your jobs, a neighborhood trend, a seasonal warning.
- Get listed on the top three vertical-specific authority sources for your industry (e.g., state licensing boards, trade associations).
- Track citation share weekly by prompting ChatGPT, Perplexity, Gemini, and Claude with your 20 target questions and logging who gets named. That is your real GEO scoreboard.
What to stop doing immediately
- Stop writing 2,000-word intros before the answer. Models skip them and cite whoever answered first.
- Stop chasing directory-farm backlinks. They do not move the citation needle and often hurt entity clarity.
- Stop publishing AI-slop blog posts with no local specificity. Perplexity and ChatGPT are trained to discount them.
- Stop obsessing over keyword density. Obsess over factual density instead — verifiable claims per paragraph.
The bottom line
Ranking in Perplexity and ChatGPT is not luck and it is not mystical. It is entity grounding, structured data, answer-first writing, and trusted citations, executed consistently for a couple of quarters. Local service businesses that get this right in 2026 will look, three years from now, like the ones who registered short domains in 2003 — early, obvious in hindsight, and impossible to catch. If you want us to run this playbook for you, that is exactly what Aliens Digital is built to do.