How to Get Your Small Business Cited by Google AI Overviews, ChatGPT, and Perplexity

You used to fight for one of ten blue links on Google. If you ranked in the top three, customers found you. Now your best customer might never see those links at all.

In early 2026, 68% of Google searches ended without a single click, according to SparkToro's zero-click study. When a Google AI Overview appears — and it now does on more than one in four searches, and on 83% of "best product" queries — the top organic result loses 34.5% to 58% of its clicks. Pew Research watched 68,000 real searches and found the click-through rate fell from 15% to just 8% when an AI summary was present. Only about 1% of users clicked a source link inside the AI Overview itself.

At the same time, 45% of consumers used ChatGPT, Gemini, or Perplexity to find a local business in the past year, up from just 6% a year earlier, according to BrightLocal's 2026 Local Consumer Review Survey. Gartner forecasts traditional search volume will drop 25% by the end of 2026 as answer engines take over.

The question is no longer "Do you rank on Google?" It is "Does the AI mention you by name when it answers?"

This is not just an SEO update. It is a channel shift. If your business is invisible to answer engines, you are invisible to a growing share of customers who never scroll past the AI box. The good news: small businesses can win citations without outspending national brands — if you understand what these systems actually cite.

From SEO to GEO: What Changed​

Traditional SEO vs. Generative Engine Optimization​

Traditional SEO was about earning a position among ten blue links. Generative Engine Optimization (GEO) — also called Answer Engine Optimization (AEO) — is about earning a place among the two to seven sources an AI cites in a single answer.

The models do not rank pages the way Google's classic algorithm did. They synthesize. They pull a fact from your blog post, a price from your service page, a review count from Yelp, and a definition from Reddit, then stitch them into one conversational answer with citations.

Adobe's 2026 SEO overview puts it this way: modern generative systems evaluate meaning, relationships, and credibility signals rather than keyword frequency. The goal shifts from keyword-first to answer-first.

What matters now:

• Can the AI extract a clean fact from your page? Structured headings, direct answers, and tables are easier to lift than dense paragraphs.
• Can it trust you? Experience, Expertise, Authoritativeness, and Trust (E-E-A-T) signals are 96% more likely to get cited, yet ranking first organically does not guarantee a citation.
• Can it understand who you are as an entity? Structured data tells the model "this page is an article by a verifiable author about a defined service in a specific location."

Only 17% to 54% of citations in AI answers even come from the traditional top-10 results. You can rank fourth and get cited, or rank first and get ignored.

Why Local Search Was Hit First​

AI Overviews now answer most local "best [service] near me" queries. For a query like "best coffee shop in Raleigh" or "affordable CPA for startups in Austin," ChatGPT and Perplexity rarely browse ten links. They aggregate.

An analysis of 200+ local-service citations across ChatGPT, Perplexity, and Google AI Overviews between January and May 2026 found the sources break down roughly like:

• Google Business Profile: 42%
• Aggregator directories (Yelp, Angi, BBB, BrightLocal ecosystem): 28%
• The business's own website: 17%
• Local press and blog mentions: 8%
• Review platforms with substantial volume: 5%

A separate study of 28 million AI responses to local queries found Yelp alone received 512,680 citations in Q4 2025 — 3.4 times more than the second-ranked platform. For local businesses, your directory footprint matters more to AI than it did to classic Google.

What AI Actually Cites (And What It Ignores)​

1. Reddit and YouTube Dominate the General Web​

For non-local informational queries, Averi.ai's research shows Reddit is the single most-cited source in Google AI Overviews at 21% of all citations, with YouTube at 18.8%. LinkedIn articles are the fourth most-cited source. Community content and social platforms collectively drive a huge share of citations.

Why? AI models are trained to value conversational, experience-based content that answers a question directly. A Reddit thread titled "I tried three payroll providers for my 12-person company — here's what actually happened" is more extractable than a corporate landing page that says "We offer best-in-class payroll solutions."

Implication for you: You do not need to become a Redditor, but content that reads like a real answer — first-person, specific, with numbers — outperforms generic marketing copy. A how-to article with a clear "Cost: $X, Time: Y hours, Tools needed: Z" table is far more likely to be lifted than a vague overview.

Ahrefs found a 0.737 correlation between a brand's YouTube presence and its AI visibility across 75,000 brands. Video is not just for humans; it is a citation signal.

2. Directory and Entity Data Wins Local​

For "best accountant in [city]" or "emergency plumber near me," the AI does not start with your blog. It starts with entities it can resolve confidently.

The hierarchy:

1. Is the business a verified entity? Google Business Profile with consistent NAP (name, address, phone), hours, categories, and photos.
2. Is it corroborated elsewhere? Same NAP on Yelp, BBB, Angi, Apple Business Connect, and industry directories. Inconsistencies kill confidence.
3. Is it reviewed and recent? Recency and volume of reviews, not just star rating, drive citation likelihood.
4. Does the website confirm the entity facts? Service pages that repeat the same address, geo-coordinates, and service area in structured data close the loop.

Your website's 17% share is not small — it is where the AI verifies what directories claim. Without it, you are a listing without proof.

3. Being Cited Pays, But Not How You Think​

When you are cited inside an AI Overview, you get 35% more organic clicks than pages that appear below the box, and 91% more paid clicks, according to recent performance data. That sounds like a win — until you see the baseline drop.

Across 25.1 million tracked impressions, organic CTR fell from 1.76% to 0.61% when an AI Overview appeared — a 61% drop. Even cited brands saw that collapse; citations double impressions but clicks barely budged (from 398,000 to 400,000 clicks when impressions doubled from 15.8 million to 33.1 million in one month-over-month study). You earn visibility, but not the flood of clicks organic search once delivered. Seer Interactive found paid CTR fell from 1.31% to 0.70% when an AI Overview was present, too.

The takeaway: optimizing for citation is defensive — you must be cited to be considered — and offensive — you must then convert that smaller click into a customer more efficiently than before.

The 7-Step Playbook to Get Cited​

Step 1: Make Every Page Answer-First​

AI does not reward essays. It rewards modules it can extract.

Rewrite your core service and location pages to lead with a direct answer, then expand.

Before (hard to extract):

"At our firm we pride ourselves on providing comprehensive bookkeeping solutions tailored to the unique needs of small businesses across various industries..."

After (easy to cite):

"Bookkeeping for restaurants in Austin costs $300–$600 per month for a 20–40 seat restaurant with 50–150 daily covers. It includes daily sales reconciliation from Toast or Square, weekly accounts payable, and monthly P&L by location. Here's what affects the price..."

Tactics:

• Use question-based H2s and H3s that mirror how people ask AI: "How much does X cost?", "What is included in Y?", "How long does Z take in [city]?"
• Put the answer in the first 40–60 words after the heading, plainly, before the nuance.
• Add comparison tables, cost breakdowns, checklists, and pros/cons lists. These formats are scanned and lifted whole.

Step 2: Deploy Structured Data in Priority Order​

Think of schema markup as the language you speak directly to the AI. Without it, the model has to guess what your page is.

Deploy in this order:

P0 — Non-negotiable:

• Organization schema with name, logo, sameAs links, address, geo, and opening hours
• Person schema for every author (verifiable name, bio, sameAs to LinkedIn or profile)
• Product or Service schema for each offer, with price range and area served

P1 — High leverage:

• Article schema with author, date published, and topic linkage to defined entities
• FAQPage schema for every Q&A block — this is the most-cited schema for answer engines; Perplexity and Google lift FAQ content directly
• HowTo schema for step-by-step guides (pricing, filing, setup)

P2 — Local amplification:

• LocalBusiness subtype (e.g., AccountingService, Restaurant, HomeAndConstructionBusiness)
• Review and AggregateRating that match what is on your Google Business Profile
• BreadcrumbList so the AI understands site hierarchy

Test with Google's Rich Results Test and keep markup in sync with visible content. Mismatched markup erodes trust.

Step 3: Fix Your Entity Corroboration​

This is the least glamorous work and the highest ROI for local citations.

• Audit your NAP across Google Business Profile, Yelp, BBB, Angi, Apple Maps, Bing Places, Facebook, and two industry directories. Exact character match matters.
• Fill every field: services, attributes, hours including holiday hours, photos with geotags, and a business description that uses the same 3–5 core phrases your website targets.
• Add UTM-free sameAs links from your website footer to each directory profile and back again where allowed.
• Keep hours and closures updated. AI models penalize stale entities.

One multi-location service business that standardized its NAP and added LocalBusiness schema across 12 location pages went from zero ChatGPT mentions to cited in 3 of 8 test prompts within six weeks — without publishing a new blog post. The entity was finally resolvable.

Step 4: Build Topical Authority, Not Just One Post​

AI prefers to cite sources that demonstrate depth on a topic, not a single viral post.

Instead of one post on "sales tax for SaaS," build a cluster:

• Pillar: "Sales tax for SaaS founders selling to customers in 20+ states"
• Cluster: nexus thresholds by state, marketplace facilitator rules, how to read a sales tax audit notice, vendor compensation discounts for filing on time, and a calculator template

Interlink them and use consistent entity phrases. The model learns you are an authority on "SaaS sales tax" and pulls from you preferentially.

A portfolio of 6–10 interlinked pages on one problem outperforms 30 scattered posts on random topics for AI citation.

Step 5: Add Experience Signals Humans and Models Both Trust​

E-E-A-T is not a checkbox. It is evidence.

• Experience: Add first-person cases with real numbers: "We onboarded a 15-person landscaping company with $1.2M revenue and cut its monthly close from 18 days to 6 by..." Even anonymized stories carry weight.
• Expertise: Author bylines with credentials, not "Admin." Link to the author's professional profile.
• Authoritativeness: Earn mentions in local press, industry blogs, and partner sites. Those 8% of citations from local press punch above their weight for local queries.
• Trust: Display last-updated dates, cite primary sources (IRS publications, state statutes), and add corrections transparently. AI models are increasingly trained to downgrade undated or unsourced claims.

Step 6: Publish Where AI Looks, Not Just Your Blog​

Since Reddit and YouTube drive over 39% of general AI citations combined, repurpose your best answer-first content:

• Turn your "How to choose a payroll provider in 2026" guide into a 6-minute YouTube walkthrough with chapters and a transcript. Upload the transcript as captions — the AI reads them.
• Answer two genuine questions per month on Reddit or industry forums without self-promotion; link only when the answer requires it. The goal is not referral traffic; it is citation eligibility.
• Pitch a local business journal or niche publication with a data-driven op-ed that includes your business name as the source of the data.

You are not chasing virality. You are planting citable objects where the models graze.

Step 7: Track AI Visibility, Not Just Google Rankings​

Classic rank tracking will miss the shift. Add:

• AI Overview presence: What share of your target keywords trigger an AI Overview? (Tools like Semrush, Ahrefs, and AirOps now track this.)
• Citation share: For a set of 20 prompts like "best [your service] in [your city]" and "how much does [your service] cost," how often does each engine name you? Test monthly in ChatGPT, Perplexity, and Google AI Mode and log it.
• Referral quality: In analytics, segment traffic where the referrer or landing page indicates an AI source. Adobe Analytics saw AI referrals grow more than 10x between July 2024 and February 2025, and AI-referred visits had lower bounce rates and longer sessions — fewer clicks, but higher intent.
• Conversion rate on AI traffic: Because you get fewer clicks per impression, your page must convert better. Add a direct, low-friction next step above the fold: "Check availability for [city] this week" beats "Learn more about our story."

What to Stop Doing​

• Stop writing for keyword density. Models evaluate meaning and relationships, not repetition. One clear answer beats ten keyword insertions.
• Stop hiding pricing or process behind "contact us." If the AI cannot extract a price range, timeline, or scope, it will cite a competitor who does.
• Stop publishing undated content. AI models downgrade content without a clear publication or updated date.
• Stop treating your Google Business Profile as set-and-forget. Stale hours, missing services, or mismatched addresses erase your biggest citation source.
• Stop measuring success by traffic alone. With zero-click at 68%, traffic will be flat or down even when visibility is up. Measure citations, calls, and booked appointments instead.

A Simple 30-Day Sprint for a Small Business​

If you do nothing else, do this in the next four weeks:

Week 1 — Entity audit (4 hours): Claim and standardize NAP across Google Business Profile, Yelp, BBB, Apple Maps, and Bing. Add LocalBusiness schema to your homepage and top two service pages. Fix any mismatches.

Week 2 — Answer-first rewrite (6 hours): Pick your two highest-margin services. Rewrite each page to start with a 50-word direct answer to "How much does it cost, what is included, and how long does it take in [your city]?" Add a cost table and an FAQ block with FAQPage schema.

Week 3 — Authority cluster (6 hours): Publish one pillar guide that answers a real customer question end-to-end, plus two supporting posts that link to it. Add author bylines, updated dates, and two primary-source citations.

Week 4 — Distribution and measurement (4 hours): Turn the pillar into a YouTube video with transcript, share a distilled answer on one relevant community forum, and run your 20-prompt citation test. Log where you are and are not mentioned.

Repeat monthly. Citation share compounds.

How This Connects to Your Books​

When fewer searches become clicks and more answers come from AI, every click that does arrive is more expensive and more intentional. You cannot afford to guess at unit economics.

Three bookkeeping habits make the AI shift survivable:

1. Track customer acquisition cost by channel, including AI referrals. Create separate income or lead-source categories for "AI search," "Google organic," and "directory" so you know whether your GEO work is lowering acquisition cost or just shifting it.
2. Separate content and directory spend from general marketing. Book your schema implementation, video production, and directory management as distinct expense lines. At month-end, you can see the cost per citation, not just the cost per click.
3. Reconcile payouts to gross revenue correctly. If you sell through marketplaces or take payments via Stripe, PayPal, or BNPL providers, your 1099-K or settlement reports will show gross amounts that do not equal bank deposits after fees and reserves. Booking gross sales, fees, and reserves separately keeps your true margin visible — critical when you are deciding how much to invest in getting cited.

Clean books do not get you cited, but they tell you whether the citation work is actually paying back.

Simplify Your Financial Management​

As you adapt how customers find you — from ten blue links to a single AI answer — keeping clear financial records is what lets you invest confidently in the channels that work. Beancount.io provides plain-text accounting that gives you complete transparency and control over your financial data — no black boxes, no vendor lock-in. Get started for free and see why developers and finance professionals are switching to plain-text accounting.