AI SEO in San Francisco, the city that built the answer engines and forgot to put its own businesses inside them.
ChatGPT was born a few blocks from your office. It still can't name you when a customer asks. We fix that.
Organic traffic growth for a hardware manufacturer competing against retail giants.
of businesses invisible in AI search — AI Visibility Index
digital-PR campaigns run through PressForge
Manufacturing client — real, from our case studies
books on AI + growth, Barnes & Noble, 5.0★
The strangest fact in California search: the metro that ships the models is where local businesses are most invisible inside them. When a San Francisco buyer asks ChatGPT for a FiDi tax attorney, asks Perplexity which Marina med-spa to trust, or reads Google's AI Overview before scrolling, the model names two or three businesses and stops. The AI boom pulled the city's agencies upmarket toward venture accounts and left everyone else — law, healthcare, hospitality, home services, e-commerce — with nobody engineering their AI visibility. That is the gap we walked into. We measure which businesses the engines recommend in your category, then build the entities, citations, and coverage that put you in the answer. It is generative engine optimization, and right now the San Francisco field is wide open.
How AI engines answer a San Francisco search
A language model does not rank San Francisco businesses the way Google's ten links did. When someone asks ChatGPT for a "Sunset District pediatric dentist" or asks Perplexity for a "Financial District litigation boutique," the model assembles an answer from the sources it already trusts — and in San Francisco those sources are specific. It pulls from the San Francisco Business Times and The San Francisco Standard, from the aggregators that own page one today (Yelp, Avvo, Thumbtack, Healthgrades), from your own structured data, from TechCrunch when the tech layer genuinely is your market, and increasingly from what people say about you on Reddit and in the forums the crawlers scrape.
Then it names a short list and stops. There is no page two in an AI answer. The city's loudest industry doesn't even use the head term — founders ask their investor Slack and X feeds, not Google — which is exactly why the classic SERP went soft and why the AI answer is the real battlefield now. The buyers who do search are the abandoned layer: practice managers in the Richmond comparing options at lunch, FiDi partners validating a referral at nine at night, hospitality GMs looking ahead of Dreamforce, home-services owners on mobile across the Peninsula. Each of those prompts resolves to a different short list built from different sources. The MentionLayer AI Visibility Index found 65.9% of businesses are invisible in AI search entirely. In the birthplace of the answer engine, that irony is your opening.
- Neighborhood-qualified prompts ("med spa Marina", "HVAC Daly City", "accountant Financial District") each resolve to a different short list — one Bay Area page can't win them all
- Professional-services answers lean on credential and practice-area language the model expects: fee-only, board-certified, fiduciary, of-counsel
- San Francisco verticals get cited through San Francisco sources — the San Francisco Business Times, The San Francisco Standard, and your vertical's trade press
- E-commerce and DTC brands compete on national AI answers where the city qualifier never mattered — they need category authority, not a Bay Area badge
- Consistent entity data across your GBP, site schema, and directories is what lets a model corroborate you and quote you by name
What we build for San Francisco businesses
AI search optimization in San Francisco is not a homepage rewrite. It is the deliberate construction of an entity the models can find, understand, and trust — across every source a San Francisco buyer's prompt touches. We start by measuring where you actually stand: which questions your buyers ask the engines, which competitors get named instead of you, and which sources those answers are built from. Then we close the gaps one at a time, prioritising the neighborhood and vertical sub-markets you can win in 90 days over the citywide terms. The classic SERP is unusually soft here; the AI answer is where the durable prize sits.
AI visibility baseline
We map the real prompts San Francisco buyers use in your category — FiDi professional services, Marina and Sunset practices, SoMa e-commerce — and record who the engines name today. The honest starting line, measured, not guessed.
Entity + citation engineering
Consistent entity data across your site schema, Google Business Profile, and the directories San Francisco models corroborate against — so ChatGPT can describe your practice or firm without hedging.
Answer-ready content
We restructure your money pages so a model can extract a clean, quotable answer — condition-plus-neighborhood queries for healthcare, service-plus-suburb queries across the Peninsula, category queries with national intent for e-commerce.
Digital PR into cited sources
Earned placements in the publications San Francisco engines pull from — the San Francisco Business Times, The San Francisco Standard, your vertical's trade press, and TechCrunch only when the tech layer is genuinely your buyer — the same PressForge engine behind 300+ campaigns.
Local + map-pack signal
Google Business Profile and review-velocity work tuned for the neighborhood map packs where three positions take most of the calls, feeding both the local pack and the AI Overview above it.
Monitoring with MentionLayer
Ongoing tracking of who the engines cite for your San Francisco prompts, so we prove movement in the answers themselves — not in a rankings report nobody reads.
Why Xpand Digital, not another San Francisco agency
Most agencies discovered generative engine optimization the week you did. We built the tooling for it. MentionLayer is our own AI-visibility platform — it measures which businesses the models name and why — and it is how we produced the AI Visibility Index that found 65.9% of businesses invisible in AI search. PressForge is our digital-PR engine, 300+ campaigns of earned coverage in exactly the kind of publications San Francisco's models cite. And our founder, Joel House, wrote the book on this shift, literally: "AI for Revenue" and "The Growth Architecture," both on Barnes & Noble at 5.0 stars. No other agency fighting for San Francisco AI search can say it built the measurement layer, the PR engine, and the playbook. We also refuse the CAC-narrative pitch the city's surviving agencies default to. If you hold a term sheet and want a growth-hacking shop, this is not your page. We work for the layer the AI boom abandoned — professional services, healthcare, hospitality, home services, e-commerce — and we take one client per industry per sub-market, so we are never optimising two competitors into the same AI answer. Judge the receipts — 2,414% organic growth for an e-commerce client, $600K reactivated from a dormant database in 90 days, a 5.0 rating on Google — not a rented FiDi address we would never pretend to have.
Common questions.
What is AI SEO, and how is it different from regular SEO in San Francisco?
Regular SEO fights for a position in Google's ten blue links. AI SEO — also called generative engine optimization — is about getting your business named inside the answer ChatGPT, Perplexity, Gemini, or Google's AI Overview gives a San Francisco buyer. There is no page two in an AI answer; the model names two or three businesses and stops. Getting into that short list depends on entity data, citations, and coverage across the exact sources the model trusts, which in San Francisco means outlets like the San Francisco Business Times and The San Francisco Standard alongside your own structured data. It overlaps with classic SEO, but the target is the answer, not the ranking.
Do San Francisco buyers actually use AI to find local businesses?
Increasingly, yes. The city's tech founders ask their networks instead of Google, but the buyers who run the rest of San Francisco — practice managers, hospitality GMs, FiDi partners, home-services owners — increasingly ask ChatGPT or Perplexity for a recommendation and get a named short list back. Google now places an AI Overview above the local pack for a growing share of queries, so even buyers who never open ChatGPT are reading an AI answer first. Our own AI Visibility Index found 65.9% of businesses are invisible in AI search, which means most of your San Francisco competitors are not there yet either. That gap is the opportunity.
How much does AI SEO cost in San Francisco?
San Francisco pricing is a barbell: startup-and-enterprise shops open at $8,000 to $15,000+ because their model only works on venture-funded clients, and national template mills charge $500 to $1,500 for reporting theatre. Serious SMB and mid-market AI visibility work realistically runs $2,500 to $6,000 a month depending on category and how far behind the engines you start. Any agency quoting a flat rate is guessing. We scope from your actual AI answers and SERPs, bill month-to-month with no lock-in, and if there is no measurable movement by day 90, that month is free.
My company is in AI. Why am I not already visible inside these models?
Because building or funding a model has nothing to do with whether that model names your company when a buyer asks. Visibility in an AI answer comes from entity consistency, third-party corroboration, and coverage across the sources the engine trusts — work that happens on your site, your Google Business Profile, and the publications San Francisco's models cite, not in a training run. It is one of the ironies of this market: the businesses closest to the technology are often the least optimized for appearing in it. We close that specific gap.
Which AI engines do you optimize for?
The ones built a few blocks from your office — ChatGPT, Perplexity, Google's AI Overviews and AI Mode, and Gemini — and no, being in the city that ships them earns you nothing automatically. A FiDi tax attorney and a Marina med-spa surface through different source sets: Perplexity leans on cited links, AI Overviews blend ranking signals with a generated summary, and ChatGPT weights entity recognition and corroboration. We build the entity, citation, and coverage foundation all of them reward, then track each engine separately so you can see which answers moved.
Do I need a San Francisco office or local clients for this to work?
No, and we will not pretend otherwise — no rented FiDi address, no borrowed local logos. A language model does not cite your business because your agency has a SoMa suite; a vendor's address is not a ranking or citation factor. What moves AI answers is whether your entity data is consistent, whether authoritative San Francisco sources describe you, and whether your content answers the query cleanly. We serve San Francisco remotely and founder-led, work your neighborhood and Peninsula sub-markets daily, and Joel reads every audit himself. Judge the work and the receipts, not the zip code.
We're not a startup. Will you take a practice, firm, or local-services business?
That is the entire point of this page. The San Francisco AI visibility programs we build are for professional-services firms, healthcare and dental practices, hospitality operators, home-services companies, and e-commerce brands — the layer the venture-chasing agencies stopped serving. The work looks like entity and citation engineering, neighborhood and Peninsula answer pages, review systems for the map pack, and category authority for e-commerce. No CAC decks, no growth-hacking vocabulary, no pretending you are a startup.
How do you prove AI SEO is working?
We measure the answers, not just the rankings. MentionLayer — the AI-visibility platform we built — captures which firms the engines name for your San Francisco prompts at the start, then tracks that list as the work lands. You see the exact queries where you went from unnamed to cited, alongside organic traffic and rankings. In the city that ships these models, that is the fitting standard: movement shown in the answers themselves, not something we ask you to take on faith.
How generative engine optimization works, end to end.
The engine-agnostic play for getting cited across AI search.
Classic organic SEO for San Francisco — the other half of getting found.
Take a San Francisco AI answer nobody is defending
Start with the audit. It is free, and Joel reads it personally — an actual look at which San Francisco competitors the engines name in your category, which sources those answers are built from, and where your 90-day wins sit. If we work together it is month-to-month, because the work should retain you, not a contract. We take one client per industry per sub-market, so we are never optimising two rivals into the same AI answer, and we open a limited number of new builds a month so nothing ships from a content mill. The guarantee is in writing: measurable movement by day 90 — in rankings, traffic, or the answers themselves — or that month is free. A top-five metro this exposed inside AI search is a mispricing, and mispricings get corrected. Take the seat while it is this cheap.