How to run a local GEO baseline audit

Ask ten local business owners how their brand is performing in AI-driven search, and nine of them will instinctively point to their Google Business Profile. Historically, this was the correct instinct. For over a decade, optimizing your Google Business Profile (GBP) was the single most effective lever for local search engine optimization. Today, however, looking solely at your GBP to measure AI visibility is looking in the wrong place.

The gap between traditional map visibility and generative AI recommendations is staggering. According to SOCi’s 2026 Local Visibility Index, which analyzed nearly 350,000 business locations, ChatGPT recommended just 1.2% of the locations in the database. Contrast this with the 35.9% appearance rate those exact same brands achieved in Google’s traditional local 3-pack. That represents a roughly 30-fold drop-off in visibility when users shift from a standard search engine to an AI assistant.

The numbers vary across other engines, but the trend remains highly competitive. Gemini recommended 11% of the analyzed locations, benefiting significantly from its native integration with Google’s ecosystem. Perplexity recommended 7.4%. Furthermore, the data underlying these recommendations is often highly unstable. Business profile information across the web was found to be only about 68% accurate on ChatGPT and Perplexity. Gemini achieved 100% accuracy, but only because it draws its data directly from Google Maps.

This means a business can easily dominate the local map pack in its zip code and still completely disappear the moment a consumer asks an AI assistant for a recommendation. Most local businesses have never actually audited what generative engines say about them. Consequently, they continue to invest heavily in standard content, citations, and backlink strategies without knowing if any of those efforts are registering where it now matters most. A local Generative Engine Optimization (GEO) baseline audit solves this problem. It provides a structured, repeatable framework to benchmark how AI platforms describe, recommend, or overlook a business before you allocate budget to optimize for them.

Why the Baseline Audit Must Come First

In digital marketing, attempting to optimize without a baseline is like stepping onto a scale for the first time three weeks into a new diet. Without a starting number, there is no reliable way to determine if your tactics are actually producing results. A structured local GEO baseline audit gives you tangible, quantifiable metrics that you can track over time: your brand’s share of voice, citation rates, and factual accuracy across different large language models (LLMs).

Beyond benchmarking performance, a baseline audit answers a fundamental technical question: Can AI crawlers even access, interpret, and trust your website? If an LLM cannot crawl your site or struggles to make sense of your data, any creative content strategy you build will fail. You must identify and resolve these underlying eligibility and indexation issues before moving on to advanced content creation.

It is also crucial to understand that generative AI engines evaluate local ranking signals very differently than traditional search engines do. In traditional local search, physical proximity is often the dominant ranking factor. The business physically closest to the user’s GPS coordinates or stated location typically wins a spot in the local pack. AI assistants do not prioritize proximity in the same way. Instead, they prioritize data confidence, brand authority, and cross-web consistency.

Generative models look for third-party validation, structured structured data, and identical business details across multiple independent web sources. Proximity is treated as just one variable among many. Because AI relies on this broad web of data—weighted differently than Google’s map algorithms—your current map-pack rankings are no longer a reliable indicator of your visibility in conversational search.

Step 1: Assemble Your Audit Inputs

Before you begin prompting different AI models, you must organize your methodology. Running random queries will yield inconsistent results. Start by setting up a dedicated tracking spreadsheet to categorize your audit queries. To get a complete picture of your AI visibility, you need to test four specific query categories, each designed to uncover a different operational or contextual weakness:

• Discovery Queries: These are high-funnel, non-branded searches designed to see if you appear when users look for local solutions. Examples include “best [service] near me” or “top-rated [service] in [city].”
• Comparison Queries: These queries measure your brand’s perceived authority against your direct market rivals. Examples include “[Your Brand] vs. [Competitor] in [city]” or “should I choose [Your Brand] or [Competitor] for [service]?”
• Trust Queries: These look at how the AI assesses your reputation and reliability. Examples include “[Your Brand] reviews” or “is [Your Brand] reliable and licensed?”
• Logistics Queries: These test the factual accuracy of the AI’s database. Examples include “what are the hours for [Your Brand] in [city]?”, “where can I park at [Your Brand]?”, or “what is the phone number for [Your Brand]?”

Once your queries are defined, you must test them across the core platforms your target audience uses: ChatGPT, Perplexity, Gemini, and Google AI Overviews. Because each of these systems uses different training data, live-search integrations, and retrieval-augmented generation (RAG) pipelines, appearing on one engine is no guarantee you will appear on the others.

To ensure your data is clean and actionable, you must control for external variables that can quietly distort your results. AI responses are highly personalized and context-dependent. Always test from a clearly defined location, and explicitly note the city or ZIP code you are targeting in your tracking sheets.

Additionally, perform your searches in both logged-out, private browsing sessions (to establish a clean baseline) and logged-in accounts (to observe how personal search history might impact the output). Always date-stamp every query session. LLMs and their underlying search indexes are updated continuously; a baseline record is only useful if you know exactly when the snapshot was captured.

Step 2: Run the Prompts and Record the Results

With your query list and platforms prepared, begin executing your prompts. For every single interaction across each platform, you need to capture and record five core metrics:

• Mention: Did the AI explicitly name your business in its response? (Yes/No)
• Mention Order: Where did your business appear in the response? Was it listed first, buried in the middle, placed last, or omitted entirely?
• Sentiment and Framing: How did the model describe your business? Was the tone positive, strictly neutral, or did it highlight negative customer feedback?
• Factual Accuracy: Are the hours of operation, physical address, list of services, and pricing details correct, or is the model hallucinating outdated information?
• Cited Sources: Which specific source URLs, local directories, or review platforms did the model cite to support its answer?

These five data points provide a much clearer picture of your conversational search footprint than standard keyword tracking tools can offer.

To simplify the recording process, build your spreadsheet with dedicated columns for the Prompt, Platform, Mention (Y/N), Position, Accuracy Score (0-100%), Sentiment (Positive/Neutral/Negative), Citation Count, and Top Sources. Once your data is entered, calculate two primary high-level metrics to summarize your performance:

• Visibility Percentage: The frequency with which your business is recommended or mentioned across all tested prompts.
• Accuracy Percentage: The frequency with which the AI provides entirely correct factual details about your operations.

If you prefer not to build this tracking system from scratch, you can use this free GEO audit Google Sheets template, which includes pre-formatted sheets for logging responses, tracking competitors, generating scorecards, and identifying visibility gaps.

While conducting your audits, make sure to document your competitors’ performance as well. Note which of your local rivals appear in the responses, where they rank in the recommendation hierarchy, and what sources the AI cites to validate them. This will show you exactly who is winning the AI share of voice in your market and help you identify the specific platforms and directories driving their visibility.

Step 3: Diagnose Your Visibility Gaps

Once your audit data is logged, you will begin to notice patterns. Every performance gap you identify will generally fall into one of three distinct categories. Diagnosing which bucket a gap belongs to is essential for determining your next steps:

The Invisible Gap

This occurs when your business simply does not appear in response to highly relevant local queries. This is the most common issue for local brands that have never optimized for generative search. Typically, an Invisible Gap is caused by technical blockages preventing AI agents from crawling your website, a lack of structured, machine-readable content on your pages, or a lack of brand mentions on authoritative third-party websites.

The Inaccurate Gap

In this scenario, the AI does recommend your business, but the information it provides is incorrect or outdated. It might display an old phone number, list services you no longer offer, or show incorrect business hours. Because LLMs prioritize data confidence, inconsistent business data acts as a negative trust signal. If an engine detects conflicting information across the web, it is highly likely to omit your business from future recommendations to avoid giving users incorrect details. This issue is almost always caused by outdated information on your own site or inconsistent Name, Address, and Phone (NAP) data across major directories.

The Misframed Gap

Here, your business is mentioned, but it is presented unfavorably. It might be ranked below weaker competitors, framed as a budget-only option when you offer premium services, or accompanied by a warning about mixed customer reviews. A Misframed Gap is usually the result of a weak or neglected online review profile, low review volume, or a lack of authoritative local PR and brand mentions to counter historical negative press.

Step 4: Fix Your Gaps in the Right Order

When resolving the issues identified in your audit, the order of operations is critical. Skipping straight to content creation while your site has underlying technical issues is a waste of time and resources. To get the best results, address your gaps in this specific sequence:

Phase 1: Technical Eligibility

Your first priority is ensuring that search engine and LLM crawlers can easily access and understand your website. If your technical foundation is broken, your content will remain invisible to AI engines.

• Verify Crawler Access: Review your robots.txt file and web server configurations to confirm you are not blocking user-agents like GPTBot, PerplexityBot, or ClaudeBot. Be particularly careful with firewalls and security suites. For example, Cloudflare introduced default settings designed to help site owners easily block AI scrapers, which some webmasters enabled without fully realizing it would also exclude them from conversational search indexes. Read more about these options on the Cloudflare Blog.
• Standardize Your NAP Data: Ensure your business name, physical address, and local phone number are formatted identically across every page of your site, your GBP, and major data aggregators.
• Deploy Structured Schema Markup: Implement comprehensive JSON-LD schema on your website. Use specific schemas like LocalBusiness, Organization, FAQPage, and Service to supply search engines with clean, structured data that can be parsed without risk of misinterpretation.

Phase 2: Establish Trust Signals

Once you are technically accessible, focus on building the trust signals that AI engines require to confidently recommend your brand to users.

• Cultivate a Robust Review Ecosystem: Actively solicit fresh, detailed reviews across Google, Yelp, and industry-specific platforms (such as TripAdvisor, Houzz, or Avvo). LLMs look at review velocity, diversity, and overall sentiment to gauge real-world customer satisfaction.
• Engage with Your Customers: Respond consistently to both positive and negative reviews on your business profiles. Generative engines track active profile management as a strong signal of operational reliability.
• Build Consistent Citations: Audit and clean up your brand profiles on major social media platforms, local chambers of commerce, and local directories. Consistent business descriptions across multiple independent websites confirm your legitimacy to AI models.

Phase 3: Optimize for Relevance

Only after addressing eligibility and trust should you focus on expanding your website’s content to capture niche search intent.

• Develop Detailed Location Pages: Replace thin, templated city pages with high-value local resources. Include actual photos of completed local projects, case studies of local clients, and specific neighborhood-level details.
• Create Structured FAQ Hubs: Write comprehensive FAQ pages that address common customer questions in a natural Q&A format. Structuring your content this way makes it much easier for RAG pipelines to extract your answers and present them directly in conversational search results.
• Avoid Low-Value Content: Do not publish thin, generic content that simply swaps city names in a template. AI engines are increasingly adept at filtering out repetitive, low-effort pages, and relying on them can hurt your site’s overall quality score.

The logic behind this sequence is simple: if crawlers cannot access your website, or if your core business data is inconsistent across the web, your content optimization efforts will have little impact. Ensuring technical eligibility must always be your first step.

Step 5: Make Your GEO Audit Repeatable

A single GEO audit only provides a temporary snapshot of your performance. Because generative search engines, training databases, and algorithmic weights are constantly changing, you must run these audits on a regular schedule to maintain visibility.

For most local and regional businesses, a quarterly audit cadence strikes the right balance. This schedule allows you to monitor how algorithm updates affect your visibility and measure the impact of your optimizations without turning the audit process into a full-time job.

As you transition to tracking your performance in generative search, you will also need to adjust your key performance indicators (KPIs). Traditional SEO metrics, such as organic clicks, do not translate perfectly to generative environments, where AI models often answer queries directly on the search results page.

Instead of focusing solely on website clicks, look at broader business indicators: rises in branded search volume, increased phone call volume, and more directions requests on your local listings. These real-world actions are strong indicators that AI recommendations are driving actual customers to your business, even when those interactions do not result in a traditional website visit.

When reviewing your quarterly audit data, look for changes in your visibility and accuracy scores. If your visibility has increased but your average recommendation placement remains low, you likely need to focus on building stronger trust signals and reviews. If your accuracy score drops, you need to audit your directory profiles for outdated or conflicting business information.

Additionally, monitor your competitors’ share of voice over time. If a competitor is steadily gaining ground across your target keywords, analyze their digital footprint to see what changes they are making. Often, you will find they are earning new local press mentions, generating a higher volume of positive reviews, or simply keeping their business profiles more active and up to date.

Start with a Diagnostic, Not with Guesswork

Succeeding in the era of generative search does not require a complete overhaul of your marketing strategy. It starts with a simple, structured diagnostic. By running a local GEO baseline audit, you can understand exactly how AI engines perceive your business, resolve critical access and accuracy issues, and build an authoritative digital footprint that models can confidently recommend.

The alternative is operating in the dark. Without a clear diagnostic, you risk missing out on valuable customer leads to competitors who have optimized their profiles for AI discovery. If you have not yet audited how generative engines present your business, setting up a baseline tracking sheet is the most valuable first step you can take.