AI Visibility Audit for B2B SaaS in 2026: What It Checks, What It Costs and Who Does It

The audit checklist: ten things a real AI visibility audit checks

A useful audit answers one question per line below, per engine. If a report you pay for skips half of them, you bought a score, not an audit.

The first three lines are technical, and you can check them against your own robots.txt and Google Search Console. Lines four to nine are where the findings are. Line ten connects the mentions to pipeline, so you can judge the work in revenue terms.

Crawler access: three kinds of bot, and the search crawler matters most

OpenAI and Anthropic each document three roles: a training crawler, a search-indexing crawler, and a fetcher that visits a page because a user asked a question. Perplexity documents the last two, and Google documents a robots.txt token for Gemini plus its own user-triggered fetchers. Each role is controlled separately, so a robots.txt rule written for one can catch another by accident.

The search-indexing crawler matters most for being found. OpenAI's own documentation says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. Perplexity describes PerplexityBot as the crawler that surfaces and links websites in its search results, and says it is not used to crawl content for AI foundation models.

The training crawlers are a separate decision. Blocking GPTBot tells OpenAI not to use your content for training. Blocking ClaudeBot does the same for Anthropic. Neither removes you from search answers on its own.

Google-Extended works differently from the others. Google's documentation says it has no separate HTTP user agent, and that it manages whether content Google crawls may be used to train future Gemini models and to ground answers in Gemini Apps and Grounding with Google Search on Vertex AI. It does not affect inclusion in Google Search and is not a ranking signal. Blocking it does not take you out of AI Overviews, and allowing it does not put you in.

The user-triggered fetchers behave differently again. OpenAI says robots.txt rules may not apply to ChatGPT-User, because a person initiated the visit, and Perplexity and Google say their user-triggered fetchers generally ignore robots.txt for the same reason. Anthropic is the exception: its bots, Claude-User included, honor robots.txt, and it says disabling Claude-User may reduce your site's visibility for user-directed web search. An audit should say which of these visits you can see in your server logs, because that is the closest thing to ground truth you have.

Google AI Overviews: no special markup is needed

Google is unusually direct about this. Its AI features documentation says a page needs to be indexed and eligible to be shown in Google Search with a snippet to appear as a supporting link in AI Overviews or AI Mode, and that there are no additional technical requirements. It adds that you do not need new machine-readable files, AI text files or special schema markup.

That sentence should change how you read any audit. If the headline recommendation is a file or a markup type, it is solving a problem Google says you do not have. What moves you is ordinary work. Get the page indexed, answer the question in its first lines, and get onto the other pages the engine retrieves.

Google also names the mechanism that makes coverage matter. AI Overviews and AI Mode may use a query fan-out technique, issuing multiple related searches across subtopics to build one answer. So write pages that answer the sub-questions a buyer's question breaks into, as well as the question itself. We explain how that works in fan-out queries.

The numbers an audit should report, and what each one means

Four measures come up in every serious audit. They move independently, so a single blended score hides more than it shows.

Visibility is the percentage of AI responses that mention your brand, in Peec AI's definition. It is a naming rate across sampled answers, measured apart from rank and from citations.

Position is the average rank across the responses where your brand appears. You can be named rarely but early, or often but last.

Sentiment is the tone of those mentions, which Peec scores on a scale from 0 to 100. An engine that names you as the budget pick is naming you without helping you.

Sources are the pages each engine cites. This is where the useful findings hide, because being cited and being named are different events. An engine can pull your comparison page into its sources and still recommend only the agencies listed on it. Closing that gap means getting your brand onto the third-party pages it pulls from.

Two cautions. All of these are sampled readings from outside: a tool asks the engines a set of questions and counts what comes back. They are estimates, and they move with the prompt set, so compare the same prompts month to month. And none of them is revenue. An audit that stops at visibility has measured the top of the funnel and nothing else.

Who runs an AI visibility audit, and what it costs

Every price below comes from the provider's own site, checked on 23 September 2026. "Custom" means the provider quotes the price on a call. The providers fall into three types: audits run for you, free self-serve audit tools, and tracking tools.

Audits run for you: LoudFace, Derivatex and Loamly

These providers run the buyer prompts for you and send back the findings, with the gaps to fix first. They suit teams that want someone else to read the answers and explain what they mean.

LoudFace

Best for: B2B SaaS teams that want the audit read by someone who fixes what it finds.

The LoudFace audit checks where your brand stands across ChatGPT, Claude, Gemini and Perplexity against your top competitors. The audit uses data from Peec AI. You get an AI search presence score, a side-by-side competitor comparison across AI platforms, one fix you can implement within a week, and a personal Loom from Arnel on the gaps. It is free, and it measures one thing: whether your brand gets cited when your buyers search for solutions.

The team behind it runs SEO, AEO and GEO as one program for B2B SaaS and fintech. It took Toku to 97.8% visibility on "best stablecoin payroll providers", a 30-day reading ending 19 August 2026.

Where it falls short: it is a first read, built to show the gaps and the first fix. Daily tracking across hundreds of prompts is a tool's job.

Derivatex

Best for: a quick outside read on a fixed prompt set.

Derivatex offers a free AI visibility audit that tests 20 prompts your buyers ask across ChatGPT, Claude, Gemini and Perplexity, and returns a visibility score and your top 3 citation gaps. It also sells a standalone $3,500 diagnostic, covering the audit, a query map and a content strategy, for teams that want that before committing to a retainer.

Where it falls short: 20 prompts is a sample. Ask which 20, and whether they match how your buyers phrase the question.

Loamly

Best for: a paid, one-time diagnosis of why the engines recommend your competitors.

Loamly sells two one-time audit tiers. The Category Snapshot costs $990 and covers 20 buyer queries across 2 to 3 platforms, with a competitive positioning matrix and the top 5 prioritized recommendations, delivered in 3 business days. The Full Intelligence Report is custom scope: 50+ buyer queries across 6 AI platforms, a root-cause analysis and a 90-day playbook. Loamly also sells a separate monitoring subscription, which is a different product.

Where it falls short: the full report has no published price, so you need a call to scope it. And a one-time audit is a snapshot: the readings move month to month, so you will need something to track after it.

Free self-serve audit tools: Amplitude, HubSpot, AEO Engine and Geoptie

You enter a brand name or a URL and get a score or a list of gaps back for free. These tools suit a first check before you spend anything. Some read your own site for citation readiness, and others sample what the engines say about you.

Amplitude

Best for: product-led teams that already live in Amplitude.

Amplitude's free report gives a visibility score and sentiment across top AI platforms, including ChatGPT and Google AI Overview, plus competitive rankings, critical topics and personalized recommendations. You start by entering your company name, and by submitting it you consent to Amplitude processing your personal data.

Where it falls short: the page names two engines, ChatGPT and Google AI Overview, so ask which others the report covers. To tie the results to conversions, you connect the report to Amplitude Analytics.

HubSpot AI Search Grader and HubSpot AEO

Best for: a free baseline, then low-cost prompt tracking.

HubSpot's own AEO audit guide points readers to HubSpot AI Search Grader as a free diagnostic across major answer engines, before any paid software. The AEO product page lists the same free tool as AEO Grader, a one-time snapshot of your visibility across ChatGPT, Perplexity and Gemini. The paid product, HubSpot AEO, costs $50 a month, or $45 a month billed annually, and includes 25 prompts tracked on those same three engines. You can try those 25 prompts free with no credit card, it needs no other HubSpot subscription, and Marketing Hub Pro and Enterprise include it with more prompts.

Where it falls short: the grader is a one-time snapshot, and the $50 plan covers 25 prompts on three engines. More prompts means buying more or moving up to Marketing Hub.

AEO Engine

Best for: a free technical readiness check, with a managed program behind it.

AEO Engine's free audit checks five dimensions: AI crawler access, structured data and entity clarity, answer-ready content coverage, internal linking and canonical hygiene, and citation and proof signals. Its managed plans are published: $1,597 a month for Growth, $2,997 for Scale, and $5,000 to $10,000 for its Dominate tiers.

Where it falls short: its five published checks all read your site. They tell you whether you are ready to be cited. Whether the engines already cite you takes a separate read of the answers.

Geoptie

Best for: a no-signup page check.

Geoptie's free GEO audit takes a URL and scores six metrics: citation readiness, answer alignment, knowledge graph optimization, content authority, technical optimization and competitive positioning. No signup is required. Its paid tracking plans start at $49 a month for 15 monitoring prompts, or $41 a month billed annually, and its audit page links free tools such as a GEO rank tracker.

Where it falls short: the audit mostly scans your own content, schema, technical signals and bot access, with competitive positioning as one metric of six. To see which brands the engines name for a specific prompt, you need its separate trackers or a paid plan.

Tracking tools: Otterly.AI, Ahrefs Brand Radar, Profound and Peec AI

Tracking tools run a fixed set of prompts on a schedule and report how your visibility moves over time. They suit teams already doing AEO work in-house, since the same prompts week after week show whether the work moves them.

Otterly.AI

Best for: in-house teams that want to track prompts themselves on a small budget.

Otterly's pricing page lists Lite at $29 a month for 15 search prompts across ChatGPT, Google AI Overviews, Perplexity and Copilot, Standard at $189 for 100 prompts, and Premium at $489 for 400. Claude, Google AI Mode and Gemini are paid add-ons.

Where it falls short: Lite's 15 prompts cover a narrow slice of buyer questions, and extra prompt packs are sold only from Standard up. Lite does include 1,000 GEO URL audits a month and 3 recommendations a week, and Standard makes the recommendations unlimited.

Ahrefs Brand Radar

Best for: teams already paying for Ahrefs.

Brand Radar runs search-backed prompts through AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini and Copilot. Custom prompts come free with every paid Ahrefs plan, from 5 prompts checked daily on Lite to 20 on Advanced, and larger custom-prompt packages start at $50 a month. The AI Visibility Index, which maps visibility across 454M prompts with no setup, starts at $199 a month, or $699 a month for all models.

Where it falls short: the included prompts need a paid Ahrefs plan (Lite or above) underneath, and 5 daily prompts on Lite is a small sample. Price the base plan in.

Profound

Best for: enterprise teams that want many engines in one platform.

Profound's pricing page shows a free trial of 50 prompts run daily for 7 days across ChatGPT, Gemini and Google AI Overviews, and an Enterprise plan with custom pricing that tracks up to 9 answer engines.

Where it falls short: for brands, there is no self-serve paid tier between the trial and Enterprise. Agencies get a separate self-serve Agency Growth plan.

Peec AI

Best for: teams that want per-engine visibility, position and sentiment with clear metric definitions.

Peec AI tracks visibility, position and sentiment per engine, and publishes plain definitions of each. Its pricing page lists Starter at $95 a month for 50 prompts, Pro at $245 for 150 prompts and Advanced at $495 for 350 prompts, each tracking 3 models daily. Enterprise is a sales call. The plan comparison table ticks gap analysis and recommended actions on every plan, Starter included, and Peec's docs say its Actions list is ranked by impact and refreshed once a week.

Where it falls short: each self-serve plan covers 3 models, and tracking more is a paid add-on.

Free audit, paid audit or tracking tool: which do you need?

Five findings an audit should look for

Each of these needs a different fix, so a good audit names which ones apply to you.

You are cited but not named. An engine reads your page and recommends the brands listed on it. It can happen on any page that lists other brands, such as a listicle or a comparison page, and the fix is being named on pages you do not own.

You are absent from the pages the engines retrieve. For generic category questions, engines draw on third-party lists and reviews. If you are not on those pages, no amount of work on your own site gets you into that answer.

A search crawler is blocked by accident. A blanket rule written to stop training crawlers can also catch OAI-SearchBot or PerplexityBot.

No page answers the buyer's exact question. Your site answers the question you want asked. The buyer asks a different one, and the engine picks the page that answers theirs.

The blended number hides the losing engine. A single visibility score can look healthy while one engine names you in almost no answers.

Only one of the five is a technical fault. Three of them sit in the answers and the sources the engines cite, which an audit of your own site never reads. That is why getting named in third-party sources often matters more than another round of on-page edits.

Run the first pass yourself

If you want to see the shape of the problem before anyone else looks, you can run a basic audit in an afternoon. Pick 20 questions your buyers ask, run them in each engine, and record who is named, where you appear, and which pages are cited. Our step-by-step method is in how to run a share-of-answer audit in 90 minutes, and the tools that automate it are compared in the best AEO tools for B2B SaaS.

The DIY version will show you the gap. It will not tell you why the engines pick the sources they pick, which is the part that decides what to fix first.

Get your free AI visibility audit

The free LoudFace AI visibility audit shows where you stand against your top competitors in AI answers, with one fix you can implement within a week. If the fix is bigger than a week, that is what our SEO and AEO program is for.