Steady Rankings, Dropping Enquiries: What Google AI Search Changed for Local Business Websites in 2026

This scenario has come up repeatedly in customer conversations since February. The details vary. The underlying problem does not.

Here is a typical case. A small accountancy firm in Limerick. Google Business Profile fully completed, photos updated monthly, reviews coming in steadily. They sat in the local map pack for their key search terms. Enquiries through the website contact form had been reliable for three years.

In February, without any change on their end, the enquiries started thinning. Not dramatically. Just a slow, steady reduction over six weeks. When they investigated, their map pack position was unchanged. But when they searched for their own services, many queries returned an AI-generated summary at the top of the page. Their business was not in it. Their nearest competitor was.

The problem was not their Google Business Profile. It was their website.

What Changed in How Google Handles Local Queries

During our operations review this week, a data point in Whitespark's 2026 Local Search Ranking Factors report stopped me. The study, which is the most comprehensive annual analysis of local SEO signals published, introduced a new category this year: AI Search Visibility. The signal weighting is materially different from anything that came before it.

To understand why that matters, a brief operational note on what has happened to Google Search in 2026. AI Overviews — the AI-generated summaries that appear above traditional results — now show for around 68% of all local queries, according to Whitespark's 2026 research. Traditional local packs appear for approximately 39% of queries. ^[1] That means for most local searches today, the AI-generated answer is what customers encounter first. The local pack is secondary.

Google AI Mode extends this further. It is an end-to-end AI search experience powered by Gemini that constructs a comprehensive response rather than presenting a ranked list. For local queries, it does not show ten businesses and let the customer choose. It surfaces one or two — sometimes none — depending on how confidently it can assess the options.

AI local visibility is significantly harder to achieve than traditional local rankings. According to Search Engine Land's 2026 report on AI local visibility, achieving visibility in AI-generated local results is up to 30 times harder than ranking well in traditional local search. ^[2] That compression means the businesses that appear in AI results receive outsized visibility. The ones that do not appear receive none.

The Signal Reversal Most Businesses Have Not Noticed

This is the part that matters most operationally.

In traditional Google local pack rankings, Whitespark's data shows GBP signals carry the heaviest weight at around 32% of ranking decisions. On-page website signals — the content and structure of your actual website — contribute approximately 15%.

In AI search visibility, those proportions invert. On-page signals carry approximately 24% of weight. GBP signals carry only 12%. Whitespark's summary of this finding is direct: for AI search visibility, the website is more important than the Google Business Profile. ^[1]

The accountancy firm in Limerick had excellent GBP scores and underperformed in AI results precisely because of this. Their website was slow. Service descriptions were brief and generic. There was no structured data. The website had been treated as a secondary asset for years, because traditional local SEO made it secondary. In AI search, it is not.

A business can be fully optimised for the ranking signals that determined local visibility in 2023 and still be invisible in the search results most customers are now seeing. That is not a hypothetical risk. It is what we are observing in customer data today.

What On-Page Signals Actually Mean for a Local Business

When Whitespark refers to on-page signals at 24% of AI search visibility weight, they mean the content, structure, and technical quality of your actual website. For a local business, the most relevant factors are:

Page loading speed. AI systems access and assess websites directly when compiling answers. A site that loads slowly gives Gemini less confidence in the data it is reading. The infrastructure differences between managed WordPress hosting platforms translate directly into how reliably AI systems can access and process your site under normal query loads.

Specific service content. Pages with clear, precise descriptions of what you do, where you operate, and who you serve give AI systems something concrete to cite. A homepage with vague copy and a contact form provides little. Dedicated service pages, with location-specific information and direct answers to the questions customers actually ask, give AI systems the material they need to feature your business confidently.

Structured data. FAQPage schema and LocalBusiness schema provide AI systems with a machine-readable version of your business information. Research from early 2026 consistently shows pages with FAQ structured data are surfaced significantly more often in AI-generated results. WordPress makes this straightforward to implement through standard plugins. Aligning your content with Google's E-E-A-T standards provides the credibility layer that makes structured data meaningful rather than just technically present.

Review velocity and recency. Reviews still carry significant weight in AI search, representing around 16% of AI visibility signals per Whitespark's weighting. A business with a steady flow of recent reviews will be cited more reliably than a competitor with older or fewer reviews, regardless of GBP completeness.

One operational caveat worth stating clearly: none of these signals work in isolation. Website speed alone does not produce AI visibility if the content is thin. Structured data alone does not compensate for a weak review profile. The signals operate together. The combination that consistently performs in AI search is: fast-loading site, specific service content, structured data implemented, active review profile, and a complete GBP. Remove any single element and you give AI systems a reason to surface a competitor instead.

What the Limerick Firm Actually Changed

They were not starting from scratch. Over three weeks, they made targeted changes. Service pages were expanded with specific descriptions of each service area and the types of clients served. FAQPage schema was added through a standard WordPress plugin. A review request email went to recent clients, which produced several new verified reviews. Their GBP was already solid; the work was almost entirely on the website.

Their site speed was the issue that took the most time to resolve. They were on shared hosting with a slow shared server environment. Moving to managed WordPress changed their measured TTFB noticeably — somewhere in the range of 30% to 50% faster on Irish mobile connections, though that estimate reflects a range across the site configurations we have seen rather than a precise single figure.

By week six, their appearances in AI-generated results had increased measurably. By week eight, enquiry volume was close to where it had been. The changes were not technically complex. What was missing was understanding that the criteria had changed.

Web60's managed WordPress hosting runs on Nginx with Redis object caching on sovereign Irish cloud infrastructure, optimised for the fast, reliable load times that both AI systems and customers require. The AI website builder generates properly structured service pages from your business description, giving AI search something concrete and well-organised to evaluate from the start.

The Pattern That Will Repeat

A business that has invested heavily in GBP optimisation and traditional local SEO while treating its website as an afterthought is now in a structurally weaker position than it was twelve months ago. That is not a prediction. It is what the 2026 signal weighting data shows.

One scenario where this matters less: businesses that operate entirely through referral and repeat clients, with no interest in generating new business through search. That is a legitimate operating model. It applies to a narrow and shrinking fraction of Irish businesses.

For the majority — service providers, retailers, professional practices, and hospitality businesses trying to be found when someone searches for what they offer — the operational question has shifted. It is no longer only "is my GBP optimised?" It is "does my website give AI search systems enough structured, fast-loading, credible information to recommend me over a competitor?"

The GBP still matters. It represents around 12% of AI search visibility signals and remains important for traditional local pack appearances. The website now carries more weight. That is a significant change in where to focus attention.

Conclusion

The accountancy firm in Limerick now reviews its AI search visibility data quarterly, alongside its traditional ranking positions. Not because the situation is fragile, but because the signals that determine visibility are not static. They want to know when things shift before the enquiries reflect it.

That is the operational discipline AI search rewards. Not perfection. Awareness of what has changed, and a website configured to meet those changed criteria.

Your Google Business Profile is not obsolete. But it is no longer sufficient on its own. Your website — its speed, its content quality, its structured data — has become the primary input into whether AI search features your business or your competitor when a customer in your area searches for what you do.

Sources