Clients used to Google "CPA near me" and scroll through a map pack. That still happens, but a growing share of people are asking ChatGPT or Perplexity directly: "who's the best accountant for small business taxes in [city]?" and booking whoever gets mentioned. If your firm isn't showing up in those answers, you're losing work to competitors who figured this out first.
This isn't a future problem. A January 2026 Deloitte study found that 63% of finance organizations have already deployed AI in their operations, and the clients of those organizations are increasingly using AI tools to find the professionals they hire. The window to get ahead of this is still open. But it won't stay open.
Why AI search works differently from Google
Google ranks pages. AI tools recommend entities. That's the core difference, and it changes everything about how you should think about visibility.
When someone asks ChatGPT to recommend a tax preparer for a small business owner, ChatGPT isn't crawling the web live and returning a list of links. It's drawing on everything it's been trained on: directories, articles, review sites, professional databases, forum discussions. If your firm appears consistently across those sources with the same name, address, description, and service list, you become a recognizable entity. If your information is scattered and inconsistent, you don't exist in any meaningful way to the model.
Perplexity works differently. It does run a live crawl per query, which means freshness matters. Pages that haven't been updated in months lose ground to competitors refreshing their content every two weeks. If you have a page on small business bookkeeping or Section 179 deductions, update it regularly with new numbers or examples. Not a full rewrite. Just enough to signal that the page reflects current information.
Gemini sits between the two. It favors pages that are logically structured, use clear headings, and directly answer the question being asked. It responds to schema markup more explicitly than ChatGPT does, and it gives more weight to pages where the intent is unambiguous.
The schema markup your firm actually needs
Schema markup is machine-readable code added to your website that tells AI systems what your business is, who it serves, and what it does. Most accounting firm websites have none of it. That's a significant gap.
Start with LocalBusiness schema using the FinancialService subtype. This signals to AI systems that your firm is a location-dependent professional service, which matters when someone searches "accountant near me." Include serviceType, areaServed, and hasOfferCatalog properties so the schema actually describes your work rather than just confirming you exist. JSON-LD is the right format for this. It lives in the <head> of your page and doesn't interfere with anything your visitors see.
Add Person schema to your team pages. List each CPA's name, credentials, and specific focus areas (tax planning, bookkeeping, audit, whatever applies). When someone asks an AI tool for a "CPA who specializes in small business taxes," that Person schema is how the AI connects your team member's expertise to the query. Without it, the AI is guessing.
For any page that answers a client question, wrap it in FAQPage and Article JSON-LD schemas. Add BreadcrumbList schema so AI systems understand where each page sits within your site. This sounds like a lot, but in practice it's a handful of JSON blocks added once and maintained occasionally.
Citations and directories that actually move the needle
AI systems, especially ChatGPT, build trust by seeing your firm mentioned consistently across authoritative sources. Not everywhere. The right places, with accurate information.
For accountants, the directories that carry weight are:
- Google Business Profile (the foundation; every other signal amplifies this one)
- AICPA directory
- State CPA society listings
- Yelp and Better Business Bureau
- Clutch (for B2B-focused firms)
- LinkedIn company page with complete service descriptions
- Local Chamber of Commerce listings
The research is pretty clear on this: businesses with 10 accurate, high-authority citations consistently outperform competitors with 200 low-quality or inconsistent ones. Inconsistency is the real killer. If your firm is "Smith & Associates CPA" on Google, "Smith and Associates Accounting" on Yelp, and "Smith Associates" on LinkedIn, AI systems treat those as potentially different entities. Standardize your name, address, phone number, and service descriptions across every listing.
Building content AI tools will actually cite
The content strategy that works for AI search is different from what worked for Google in 2018. Broad keyword-stuffed pages don't get cited. Narrow, specific pages that answer one question well do.
Identify 5 to 10 questions your clients actually ask. Not "small business accounting" but "how do I depreciate equipment under Section 179" or "what records do I need to give my accountant for tax prep." Each question gets its own dedicated page. The first two sentences of that page should give a direct, quotable answer with specific numbers or thresholds. That's what AI tools pull when they cite a source.
For a firm that does tax prep and bookkeeping, the highest-priority pages are usually: a service page for small business tax preparation with local specificity, a page explaining your bookkeeping process and what software you work with (QuickBooks, Xero, and similar), and a FAQ page covering the questions that come up on every initial client call. Those three alone, built correctly with schema markup and updated regularly, will outperform a 20-page site with no structure.
Perplexity favors original research, expert opinions with credentials, and content that has been updated recently. If your firm publishes a short post every few weeks covering a tax change or a bookkeeping question relevant to your clients, that content becomes citation material. It doesn't need to be long. It needs to be specific, accurate, and current.
Fixing the entity recognition gap
The biggest reason accounting firms don't appear in AI search results isn't a missing blog post. It's absent entity recognition. AI systems build models of trusted businesses from structured, consistent signals across authoritative sources. If those signals are thin or contradictory, the firm simply doesn't register as a credible entity.
Fixing this starts with your Google Business Profile. Fill every field. Use the exact same business name you use everywhere else. Write a description that explicitly names your services (tax preparation, bookkeeping, small business accounting) and the types of clients you serve. Add photos, respond to reviews, and post updates at least twice a month. GBP is still the anchor for local AI recommendations.
Beyond GBP, make sure your website has a clear, crawlable About page that lists your firm's history, team credentials, and service areas. Include your CPA license numbers and any professional affiliations. AI tools treat professional credentials as trust signals, and accountants have legitimate credentials most other service businesses don't. Use them.
According to research from Syntora and others tracking AI search behavior in 2026, accounting firms that get recommended by ChatGPT and Claude share a common profile: consistent entity data across high-authority directories, schema markup on service and team pages, and content that directly answers client questions with specificity. Firms missing even one of those three elements are significantly less likely to appear.
How to audit where you stand right now
Before spending time on any of this, it helps to know what AI tools currently know about your firm. Ask ChatGPT and Perplexity directly: "who are the best CPAs for small business tax prep in [your city]?" See if your firm comes up. Then ask follow-up questions about your specific firm name and see what information the AI returns. That tells you immediately whether your entity signals are working.
Check your schema markup using Google's Rich Results Test. Most accounting firm websites return zero structured data. That's fixable. Run a citation audit using a tool like BrightLocal or Whitespark to find inconsistencies across directories. Fixing those inconsistencies is tedious but it's also one of the highest-return things you can do for AI visibility.
If you want this done properly without spending weeks on it yourself, SuggestedByGPT runs a full AI visibility scan for your firm and identifies exactly where the gaps are, which directories are sending conflicting signals, and which schema markup is missing from your pages. It's built specifically for service businesses where trust and local visibility determine who gets hired.
The firms getting recommended by AI tools right now didn't get lucky. They made their information consistent, their expertise legible to machines, and their content specific enough to cite. Those are learnable, fixable things.
Run a free scan of your firm's current AI visibility at /start and see exactly what's working and what isn't.