Something has quietly changed in how prospects find an accounting firm. They are no longer scrolling through ten blue links and clicking around until something looks right. They are opening ChatGPT, Gemini, or Copilot and asking a question in plain language: what kind of firm should I use for this, or who should I hire to help me with that? What comes back is one synthesized answer, with maybe a handful of sources cited underneath.
If your firm is not part of that answer, you are effectively invisible for that question, no matter how good you are at the work.
The practice of getting into those answers has a name: Generative Engine Optimization, or GEO. It sounds technical. It is not. It comes down to one question you can ask about any page on your site: would an AI system feel comfortable using this page as the answer?
Most firm websites fail that test for the same reason. They were built as brochures. They describe the firm, its history, and its list of services. AI tools are not looking for a brochure. They are looking for a clear answer to a specific question. Here are five fixes that close that gap, in the order I would make them.
Open almost any accounting firm service page and you will find some version of the same first line: founded in 1998, we are a full-service firm serving clients across a range of industries. That sentence tells a prospect nothing they were looking for, and it gives an AI system nothing to match against a specific question.
An answer page works the other way around. The headline is the question itself. "Do I need a bookkeeper or a CPA for my small business?" is a far better page title than "Our Services." So, is "How much does an accountant cost for a small business?" These are the things people actually type.
One page, one question. Resist the urge to cover five topics on a single page. A page that tries to answer everything is difficult to cite, because it is unclear what it is the answer to.
This is the fix that feels most uncomfortable, because it runs against how most of us were trained to write. In a memo or a client letter, you build the context, walk through the considerations, and land on the conclusion at the end. On a web page, that structure buries the one thing the reader came for.
Put the direct answer at the top, before the detail, before the caveats, before your credentials. Then use the rest of the page to explain, qualify, and give examples.
Here is a simple test. If a visitor reads only your first paragraph and nothing else, would they have their answer? If the honest response is no, rewrite that paragraph. AI systems tend to lift the clearest, most complete short answer they can find on a page. Give them one worth lifting.
Structure is what makes a page quotable. Headings act as signposts that tell both a reader and an AI system what is inside each section and whether it is relevant.
The trap is writing headings in internal language. "Engagement Scope," "Advisory Solutions," and "Compliance Framework" are how we talk to each other. They are not how a small business owner talks. Compare those to headings like "How much will this cost?", "When should I make the switch?", and "What do I need to bring to our first meeting?"
A good way to build the structure is to write down the follow-up questions a client asks you after the first one. Those follow-ups, in their words, become your section headings.
An FAQ section near the bottom of an answer page is one of the fastest and highest-value additions you can make. Three to five related questions, each with a short and direct answer, gives an AI system clean, self-contained pieces of content that are easy to parse and easy to cite.
You do not have to invent these questions. They are already sitting in your inbox, your discovery calls, and your intake forms. Pull them straight from there and keep the phrasing your clients used. "Do I need to file in another state?" will serve you better than "multi-state nexus considerations."
One more thing worth noticing about this fix: if a client is asking you a question repeatedly, dozens of people who are not yet your clients are typing that same question into an AI tool this week. Check out this example FAQ page.
Vagueness is the most common and most expensive problem I see on firm websites. Language like "we serve businesses of all sizes across many industries" is meant to keep the door open to everyone. In practice, it gives an AI system nothing specific to match to a specific question, so you get left out of the answer entirely.
This is where smaller firms have a real advantage, and most of them underuse it. It is far easier for an AI tool to confidently recommend a firm known for tax strategy for multi-state professional services businesses than one that simply says it does tax. Specificity is not a limitation. It is what makes you the obvious answer to a particular question.
Carry the same clarity into your team bios, which should show real credentials, specialties, and locations rather than a paragraph of general praise. Then make sure that story matches everywhere else you appear: your Google Business Profile, your QuickBooks ProAdvisor or Xero listing, your association directory, and your LinkedIn page. When an AI system cross-checks you and everything lines up, that consistency reads as trust. When it finds an old address here and a different firm name there, that reads as doubt, and doubt is easy to skip over.
You do not need to rebuild your website to start. Here is a version of this that fits into a normal month:
Four or five pages built this way will do more for how you show up in AI answers than a full redesign that keeps the same brochure language. Be specific in context and don’t be everything to everyone at once because then you’ll end up sounding vague.
None of this requires a developer, a technical background, or a larger marketing budget. It requires being clear about who you help, answering the question directly, and keeping your story consistent everywhere you appear.
Those are things you already do well when you are sitting across the table from a client explaining something complicated in plain language. The work is bringing that same clarity to the page, so that when someone asks an AI tool who they should hire, your firm is a name it feels confident giving.