Frequently Asked Questions

Gumshoe is an AI search visibility platform. It shows you how AI models like ChatGPT, Claude, Gemini, and Perplexity describe your brand when people ask for recommendations in your category.

We simulate real buyer conversations using persona-driven prompts, then analyze the results for brand visibility, competitive positioning, citation sources, and sentiment. We track 11 model families across seven providers, and each Visibility Audit runs six of them by default.

Gumshoe tracks 11 model families across seven providers:

• OpenAI: ChatGPT, ChatGPT mini
• Anthropic: Claude Sonnet, Claude Opus
• Google: Gemini Flash, Gemini Pro
• Google Search: AI Overviews
• Perplexity: Sonar
• xAI: Grok
• DeepSeek: DeepSeek Chat, DeepSeek Reasoner

Each Visibility Audit runs six families by default: Claude Sonnet, ChatGPT mini, Google AI Overviews, Perplexity Sonar, Gemini Flash, and Grok. You can change the selection on either paid plan, and Pro has no cap on how many families a single audit runs.

Google appears twice on purpose. Search AI Overviews is a separate engine from the Gemini models and answers differently, so we measure it in its own right.

Several families run with web search on, which means the model retrieves live pages before answering and returns the sources it used. That is where citation data comes from. We add new models as they become relevant.

Sign up at app.gumshoe.ai/go with your work email. Your first audit is a free sample, and it does not need a credit card. Results land within minutes. Read the Gumshoe Playbook for a step-by-step guide.

A personal email address works too. The only addresses we turn away are disposable, throwaway domains.

A project is your brand plus a focus area, usually one product or service line. You pick a plan for each project you track, so three product lines is three projects, and they can sit on different plans. Most teams put Pro on the area that matters most and Starter on the rest.

There are no per-seat charges. Anyone you invite to a project can see it.

Two monthly plans, priced per project: Starter at $99 and Pro at $299. Every new customer gets one free sample audit first, with no credit card. No annual contracts, and you can cancel anytime.

Starter covers a monthly Visibility Audit, a weekly Snapshot Audit, 10 Technical Audit runs, and 3 content generation pieces. Pro adds the Sentiment, Citation, and Content Audits, moves the Snapshot to daily, and raises the allowances to 50 Technical Audit runs and 10 content pieces. See the full plan comparison.

Technical Audit runs and content generation pieces reset on your renewal date, not on the first of the month.

A conversation is one full prompt-to-response exchange with an AI model, asked as a specific buyer persona. Both paid plans include 480 conversations per billing period, which is what a standard Visibility Audit uses: 8 personas, 10 prompts each, across 6 models. Every conversation is analyzed for brand mentions, competitive position, citation sources, and sentiment.

Go past 480 in a period and the extra conversations bill at $0.10 each. You always see the conversation count before you confirm a run.

A Snapshot Audit is the lighter recurring run included with a paid subscription, weekly on Starter and daily on Pro. It runs at a smaller scope than the full monthly Visibility Audit: 4 personas, 5 prompts per persona, across 4 models, for 80 conversations.

It exists so you catch a sharp move in your visibility between monthly audits. The Snapshot tells you something moved. The full audit tells you what happened and to whom. See how monitoring works.

The Visibility Audit answers whether models mention you. The rest answer why, and what to do about it.

• Sentiment Audit shows how models talk about you, sorted into pros, cons, and caveats, at two buyer-journey stages. Pro.
• Citation Audit shows which domains models lean on when they answer, and who those sources talk about. Pro.
• Content Audit shows what your site is missing against the questions buyers actually ask, split into retrieval and generation gaps. Pro.
• Technical Audit scores whether a model can fetch each URL, read it, and pull an answer out of it. On every plan, metered by run count.

This is the core problem GEO measurement has to solve, and most tools ignore it. AI answers are non-deterministic. That is a fact, not a flaw.

Gumshoe captures variance by running hundreds of unique prompts, each submitted once per model, across multiple personas and topics. Visibility is the percentage of those conversations where your brand appears. Breadth across prompts, personas, and models is what makes the data reliable, not repeated sampling of the same question.

A single ChatGPT screenshot tells you nothing. Hundreds of structured conversations across multiple models and personas tell you a great deal. Read more about why this matters.

Three problems with testing by hand:

1. You are one person with one perspective. AI responses vary by how the question is framed, and you only frame it your way. Your buyers frame it differently.
2. You see one result. AI answers are probabilistic, so a single response says almost nothing about your actual visibility.
3. You cannot do it at scale. Meaningful measurement takes hundreds of conversations across models, personas, and prompt variations. That is what Gumshoe automates.

No. Gumshoe reaches every AI model through official APIs. Scraping breaks most providers' terms of service, and scraped data is unreliable because browser sessions carry personalization, caching, and interface quirks that pollute the result. API access is more consistent, auditable, and compliant.

It also means what we measure is the model, not one anonymous browser session's view of it. Read about our methodology for full details.

For a baseline, one full audit tells you where you stand. For ongoing work, the monthly Visibility Audit tracks the trend and the Snapshot Audit catches sharp moves in between. During an active campaign, run the full audit more often so you get a faster feedback loop.

Model updates, competitor moves, and third-party coverage all shift your visibility without you changing anything, which is the main argument for a schedule rather than one-off runs.

Yes, for Preferred Partners. Preferred Partners get agency branding in the Gumshoe interface and in system-generated report downloads. It is a paid perk of the Agency Partner Program rather than something bundled with the standard agency discount, so talk to sales about the monthly minimum.

A fully white-labeled client portal on your own domain is not available today. Agencies that need one build it on the API and the export options.

Most agencies start with a baseline audit for each client, then put the client on a plan so the monthly Visibility Audit and the Snapshot Audit run on a schedule. From there: review visibility and competitive position, use the Citation and Content Audits to find the gaps, recommend and ship the work, then measure the change on the next run.

Each client brand is its own project with its own plan, so you can put Pro on the accounts that need the full picture and Starter on the rest.

The skepticism is fair. The GEO space is new and some vendors are making wild claims. Here is what Gumshoe does not promise: we do not guarantee AI placements, we do not claim to "hack" AI models, and we do not sell magic.

We provide measurement. We show you what models say about your brand, how often, in what context, and which sources they drew on. What you do with that is up to you. That is the opposite of snake oil. See our methodology for full details.