Google recently highlighted how AI agents interpret websites differently than humans, leaning on a stripped-down view of each page called the accessibility tree. The discussion, shared in the Whitespark Local Insider recap for July 27, 2026, also covered the role of customer reviews in AI answers, a Google Maps loophole affecting Service Area Businesses, a workaround for getting drone video onto a Google Business Profile, and why missing pricing can hurt AI visibility.<\/p>\n
The accessibility tree is a structured representation of a web page that exposes headings, labels, links, text, and element roles to machines rather than to sighted users. According to the Whitespark recap, AI agents use this view to understand what a page is about, which means a page that looks clean to a human can still confuse an AI if the underlying HTML is messy.<\/p>\n
To inspect the accessibility tree in Chrome, right-click any page, choose Inspect, click the double-arrow icon in the Elements panel, select Accessibility, and toggle Show Accessibility Tree on. A well-built tree clearly displays headings, text, buttons, links, and structure. A poorly built one buries or omits that information, making content harder for AI to interpret.<\/p>\n
The practical takeaway is to use proper HTML headings, descriptive link text, form labels, and alt text on images. The easier a site is for AI agents to parse, the more confidently it can be surfaced in AI-generated results.<\/p>\n
Reviews on Google and other platforms do more than reassure a human shopper. They feed the language AI systems use to describe a business. The recap points to a Gemini example where specific details about a business were lifted directly from Google Maps reviews.<\/p>\n
Reviews tend to teach AI what the business actually does, which services it is known for, what problems it solves, and why customers recommend it. The recap suggests asking customers to naturally mention the service they received, the problem that was solved, and what made the experience great when soliciting reviews. More specific review content gives AI more material to confidently recommend the business.<\/p>\n
Local SEO practitioner Mika Lepisto found that Google hides the address and map-pin editing options on Service Area Business listings, but users can still reach those editing controls by modifying the Google Maps URL. Whitespark describes the discovery and notes that Google’s bug bounty team told Lepisto the behavior is “intended functionality.”<\/p>\n
The upside is that a business owner could potentially use the same path to fix a Service Area Business that is ranking in the wrong location. The downside is that, until Google adds proper controls for business owners, the loophole is easy for bad actors to abuse. The recap recommends monitoring GBP listings for unexpected changes, including map pin moves, until Google addresses the gap.<\/p>\n
Google’s Immersive View is gated behind an invitation program that businesses cannot join on demand. The recap shares a workaround attributed to Aaron Weiche: capture a simple drone video of the business and upload it to the Google Business Profile. The result is not the official Google Immersive View, but it can produce a similar visual effect and make a profile stand out compared with the average listing.<\/p>\n
Kevin Indig’s article “Where AI agents get stuck on your site” argues that AI agents routinely search for pricing on behalf of users, even when the user did not ask about price. The recap, drawing on a podcast discussion with Claire Carlile and Julian Hooks, frames two possible outcomes when a business does not publish its pricing: the AI excludes the business entirely from its answer, or the AI pulls pricing from third-party sites, affiliates, or aggregators and surfaces potentially inaccurate or incomplete information. Both outcomes favor listing a price directly on the site, the listings, and other owned channels.<\/p>\n
The recap also referenced several pieces of local SEO coverage from the same week: Joy Hawkins on the State of Local SEO in 2026, Sterling Sky and Brad Wetherall on disappearing Google reviews, Barry Schwartz on a new bug showing businesses as having no reviews yet, Julian Hooks on building an AI search visibility tracker for under $100 per month, and Elizabeth Linder on an AI Prioritization Framework for SEOs.<\/p>\n
It also noted that the Whitespark Local Platform can monitor GBP listings for changes, including map pin moves, and send alerts when something shifts, framing the monitoring as a safeguard against the kind of listing tampering the Maps loophole enables.<\/p>\n
The accessibility tree is a structured, machine-readable view of a web page that exposes headings, labels, links, text, and element roles. AI agents use it to understand page content, so a clean HTML structure helps AI systems interpret a site accurately.<\/p>\n
Right-click any page, choose Inspect, click the double-arrow icon in the Elements panel, select Accessibility, and toggle Show Accessibility Tree on. A good tree surfaces headings, text, buttons, links, and structure clearly; a bad one hides or scrambles that information.<\/p>\n
AI agents often look up pricing on behalf of users even when price is not explicitly part of the question. If a business does not publish its price, the AI may exclude it from the answer or surface potentially inaccurate pricing pulled from third-party sites, affiliates, or aggregators. Publishing pricing on owned channels reduces both risks.<\/p>\n