Google AI Overviews for accountants: what changed and what to fix

Google AI Overviews now answer most "accountant near me" and "how much does an accountant cost" searches before a single blue link appears. If your firm's website isn't structured to be the source Google's generative answer pulls from, you're invisible at the exact moment a prospect is deciding who to call.

This is not a ranking-algorithm tweak. It's a change in what "showing up" means. A firm that used to rank #3 for "small business accountant Leeds" could now be entirely absent from the answer Google gives, even while still holding that same #3 position in the traditional results underneath. AI Overviews synthesize an answer from a handful of sources it trusts enough to cite or paraphrase — and most accounting-firm websites aren't built to be one of those sources.

This guide covers what actually changed, why keyword-focused SEO tactics from 2020 don't move the needle anymore, and the specific technical and content fixes to run over the next two weeks.

How is Google AI Overviews different from classic SEO for accounting searches?

AI Overviews compress multiple web sources into a single synthesized answer, which means a firm can rank well in classic search and still get zero visibility in the answer box. Classic SEO rewarded a page for matching a keyword closely enough and having enough authority to rank in position 1 through 10. AI Overviews reward a page for being extractable — written in a way a language model can lift a clean, self-contained answer from without needing to interpret intent, resolve ambiguity, or stitch together information scattered across the page.

For accounting firms, this distinction matters because most firm websites were built around the old model. Pages target a keyword like "accountant in Bristol," repeat it in the H1, page title, and body copy, then spend three paragraphs on the firm's history and values before ever answering what a prospective client actually wants to know: what do you do, what does it cost, and are you the right accountant for a business like mine.

AI Overviews are built from passages, not pages. Google's system is pulling a specific paragraph or two — often no more than 40-60 words — that directly answers a specific query. A page can be well-optimized in the traditional sense and still contain no single passage clean enough to extract. This is why some firms with strong domain authority and years of backlinks are seeing their AI Overview visibility lag behind smaller, newer firms whose pages happen to be structured in direct-answer blocks.

The second big shift is corroboration. AI Overviews tend to draw from multiple sources that agree with each other, not just the single highest-authority page. If your firm's pricing, service descriptions, or claimed specialisms aren't consistent across your website, your Google Business Profile, and third-party listings (Companies House filings, review platforms, professional body directories like ICAEW or ACCA "find a firm" tools), the model has less confidence citing you — because there's no corroborating signal that the information is accurate and current.

The third shift is that local intent now resolves inside the AI Overview itself for many queries, rather than triggering a separate local pack. A search like "bookkeeper for a limousine company in Manchester" might once have surfaced a local 3-pack plus organic results. Now it can generate a synthesized answer describing what such a bookkeeper should offer, sometimes naming firms, sometimes not — and if your site doesn't explicitly address that niche-plus-location combination anywhere, you don't exist in that answer even if you serve exactly that client.

Why do answer-ready pages beat keyword density now?

A service page written as a direct answer to a client's actual question gets cited; a service page written to satisfy a keyword tool does not. The practical test: pick any service page on your site and ask whether the first 100 words answer a real question a prospect would type or say into a search bar or a ChatGPT prompt. If the first 100 words are about your firm's mission, founding story, or a stock description of "comprehensive accounting solutions," you've written a page for a 2015 crawler, not a 2026 answer engine.

Take a page titled "Payroll Services." The old-SEO version opens with something like: "At [Firm Name], we understand payroll can be complex. Our experienced team offers comprehensive payroll solutions tailored to your business needs." That's 30 words and contains zero extractable facts — no number of employees handled, no turnaround time, no pricing range, no compliance detail. There's nothing here a model can lift and present as an answer.

The answer-ready version opens: "We run payroll for businesses with 1 to 250 employees, submitting RTI to HMRC on every pay run and handling auto-enrolment pension contributions automatically. Payroll for a 10-person business typically costs £75–£150 per month depending on frequency (weekly vs monthly) and whether we handle pension administration." That's an extractable passage. It has a number range, a specific compliance reference (RTI — Real Time Information, the HMRC payroll reporting requirement), and a direct cost anchor. A model synthesizing an answer to "how much does payroll cost for a small business" has something concrete to pull from.

This is the core fix: every service page should have at least one paragraph, ideally near the top, that could be lifted verbatim as a complete answer to the query that page is meant to serve. Write it as if you were the last two sentences of a Google featured snippet — because functionally, that's what it needs to become.

What technical checks should an accounting firm run first?

Before touching content, confirm Google's systems can actually crawl, parse, and trust your site's structure — this is a five-part technical audit you can run in under a day.

Crawlability. Check Google Search Console's Page Indexing report for any service or location pages excluded from the index. Accounting firm sites frequently have this problem because they were built on templated platforms (industry-specific website builders sold to accountants) that generate near-duplicate location pages — "Accountant in Leeds," "Accountant in Bradford," "Accountant in Wakefield" — with 90% identical boilerplate. Google may be treating these as duplicate content and refusing to index most of them. If you have more than three location pages that are structurally identical apart from the town name, you need to rewrite each with genuinely distinct local detail (client examples from that area, local business registration nuances, travel radius) or consolidate them into one strong regional page.

Schema markup. Implement Accountant or ProfessionalService schema (schema.org's structured data vocabulary) on your homepage and service pages, with AccountingService as the more specific type where your CMS supports it. Add LocalBusiness schema with accurate areaServed properties for every location you genuinely serve — not every location you'd like to rank in. Add FAQPage schema to any page carrying a genuine FAQ section. This markup doesn't guarantee an AI Overview citation, but it removes ambiguity about what your business is, what it does, and where it operates — ambiguity that otherwise has to be inferred, and inference reduces citation confidence.

Author and entity markup. If your site publishes articles or guides (tax deadline explainers, R&D tax credit breakdowns, Making Tax Digital updates), each piece needs a named, credentialed author — not "Admin" or the firm name as byline. Use Person schema linking the author to their professional credentials (ACCA, ACA, CTA, CPA, EA) and to a bio page. Google's systems weight expertise signals more heavily post the various Helpful Content and E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) updates, and this weighting carries into what AI Overviews are willing to cite. An unattributed "Understanding VAT thresholds" article competes against HMRC's own guidance and established tax publishers with named, credentialed writers. You will not win that competition anonymously.

FAQ visibility. Google increasingly pulls AI Overview content from genuine FAQ sections rather than manufacturing the Q&A itself. Audit your top five service pages: does each have a real FAQ section with 4-8 questions phrased the way a client would actually ask them ("Do I need an accountant if I'm a sole trader?" not "Sole Trader Accounting Requirements")? Add them if missing, using FAQPage schema, and keep each answer to 2-4 sentences — long enough to be complete, short enough to be extractable.

Internal linking. Check whether your highest-value service pages (the ones tied to your best-margin work — say, R&D tax credit claims or outsourced CFO services) are reachable within two clicks from the homepage and are linked to from at least three other relevant pages on the site. Google's crawlers and AI systems use internal link structure as a signal of what you consider important. A firm that buries its most profitable service three levels deep in a "Services" dropdown, while its blog links prolifically to a low-value "About Us" page, is telling Google the wrong page matters.

What content checks matter beyond the technical layer?

Technical fixes make your content extractable; content fixes make it worth extracting — and that means proving niche fit, location fit, and pricing transparency in specific, checkable language rather than generic claims. Five checks, run against every core service page:

Niche proof. "We work with small businesses" is not a niche claim — it's the absence of one. If your firm has real depth serving, say, dental practices, e-commerce sellers on Amazon FBA, or construction companies navigating CIS (Construction Industry Scheme) deductions, that needs to be stated explicitly, with specifics: how many clients in that niche, what specific problems you solve that a generalist wouldn't (CIS verification and monthly returns, VAT margin schemes for pre-owned goods, R&D claim documentation for software companies). AI Overviews responding to "accountant for [specific business type]" queries need a source that names that business type explicitly. A page that only says "we serve businesses of all sizes and industries" gives the model nothing to match against a niche query, no matter how good your actual expertise is.

Location proof. Beyond a town name in a title tag, location proof means specifics that couldn't be copy-pasted onto a competitor's site in a different city: which local authority area you're registered to advise within for business rates queries, whether you attend a specific local business network or chamber of commerce, named local clients (with permission) or anonymized local case studies ("a 12-person manufacturing business in the Trafford Park area"), and your actual office address with LocalBusiness schema matching your Google Business Profile exactly. Mismatched address formatting between your website, Google Business Profile, and Companies House filing is a small thing that erodes the corroboration signal AI systems look for.

Pricing and process clarity. Refusing to publish any pricing information used to be conventional wisdom in professional services; it's now a visibility cost. You don't need exact quotes, but you do need ranges and the variables that move them: "Annual accounts for a limited company with turnover under £200k typically range from £800 to £1,800 depending on transaction volume and bookkeeping quality." Pair this with a plain description of your onboarding process — what happens in week one, what documents you need, how long until first deliverable. AI Overviews answering "how much does an accountant cost" and "what happens when you switch accountants" need extractable numbers and steps. If you don't provide them, a competitor or a generic listicle site will, and that's whose language becomes the answer.

Review language. Star ratings alone don't feed AI Overviews — the actual text of reviews does, when Google can access and parse it. Encourage clients to write reviews that mention specifics: the service used, the industry, the outcome ("switched from a generalist to [Firm] for our R&D claim and recovered £34,000 — process took six weeks"). Generic five-star reviews ("Great service, highly recommend") carry sentiment but no extractable fact. Where you have permission, feature the fuller review text on your own site as well as on Google Business Profile and Trustpilot, so the same corroborating detail appears in multiple places Google can cross-reference.

Direct-answer sections. Every core page needs a section — ideally with a clear subheading phrased as a question — that answers the single most common question a prospect asks before becoming a client. For a tax return service page, that's usually "How much does a self-assessment tax return cost?" or "What happens if I miss the January 31st deadline?" (HMRC applies an automatic £100 penalty even at one minute past midnight, rising with continued lateness). Answer it directly, in the first two sentences of the section, before adding nuance. This is the single highest-leverage content change most firm websites can make, because it directly mirrors how AI Overviews are built: question in, extractable answer out.

What should a small firm with limited marketing time do in the next 14 days?

Run this as a sequenced list — technical fixes first because they unlock indexing and trust, content fixes second because they're worthless if the page isn't crawlable or trusted. This assumes one person spending roughly an hour a day, not a marketing department.

Days 1-2: Audit. Pull your Search Console Page Indexing report and list every service and location page that's not indexed. Separately, list your five highest-margin services and check whether each has its own page, reachable in two clicks from the homepage.

Days 3-4: Schema. Add or fix ProfessionalService/Accountant schema on your homepage. Add LocalBusiness schema with accurate areaServed on any page targeting a specific location. If your CMS makes this hard, a plugin (for WordPress: Rank Math or Yoast both support custom schema types) usually solves it in under an hour.

Days 5-6: Rewrite your top three service pages' opening paragraphs. For each, write a 2-3 sentence direct-answer opener with a specific number (cost range, turnaround time, or client volume) in it. Delete the "we understand your business needs" boilerplate entirely.

Days 7-8: Add or fix FAQ sections. Pick your three highest-traffic pages (check Search Console's Pages report). Write 4-6 real client questions and 2-4 sentence answers for each. Mark up with FAQPage schema.

Days 9-10: Fix author attribution. If you or your team publish any articles, add named author bylines linking to a bio page listing credentials (ACCA, ACA, CTA, CPA, or equivalent). Add Person schema to author bios.

Days 11-12: Niche and location proof. Pick your single strongest niche (the industry you have the most genuine depth in) and rewrite that service page to name the niche explicitly, with a specific number of clients served and one anonymized case example. Do the same for your primary location page, adding a specific local detail no competitor could copy-paste.

Day 13: Pricing. Add a pricing range with the variables that move it to your three top service pages, even if it's a wide range. If you genuinely cannot publish any figure, publish the process instead — what happens in a first consultation and what's needed to get a quote within 48 hours.

Day 14: Review and internal links. Send a review request to your three most recent satisfied clients with a specific prompt ("mention what service we helped with and the outcome, if you're comfortable"). Add three internal links from your blog or resources section into your highest-margin service page.

None of this requires new content volume or a redesign. It requires making the content you already have extractable, specific, and corroborated — which is what AI Overviews are built to reward, and what most accounting firm websites, built for a keyword-matching era, currently aren't.