Inside an AI Visibility Audit for a SaaS Brand
A step-by-step walkthrough of what the audit actually contains and what the client receives: how we define the query set, sample answers across assistants with repeated runs, scan the site for technical AI-visibility issues, trace the sources behind the answers, and deliver a findings report with a prioritized playbook.
Methodology walkthrough
This page shows, step by step, how we run this type of engagement. Where figures appear, they illustrate the mechanics - client results are published only with written permission and supporting data.
The Typical Challenge
A SaaS brand with real customers keeps hearing that prospects "checked ChatGPT first." Leadership asks how the brand shows up in AI answers, and nobody can respond with evidence - one teammate has a screenshot where the brand appears, another has a run where it does not, and both are single samples with no dates or model versions attached. Meanwhile the team is about to commit budget to content and community work without knowing which gaps actually matter. The audit exists to replace those anecdotes with a defensible baseline and a ranked list of what to fix.
Our Approach
1. Define the Query Set
Everything downstream depends on testing the prompts buyers actually ask, not the prompts the brand wishes they asked.
- Pull candidate prompts from sales calls, support tickets, Google Search Console queries, and community threads in the category
- Cover the full buying arc: discovery ("best tools for X"), comparison ("A vs B"), and validation ("is A legit / worth it") prompts
- Freeze the panel as a fixed list. A fixed panel is what makes the baseline resampleable later - an ever-changing prompt list cannot show movement
2. Sample Across Assistants, Repeatedly
AI answers are non-deterministic, so a single run proves nothing. We run structured sampling instead.
- Each prompt in the panel is run across ChatGPT, Perplexity, and Google AI Overviews - repeated runs spread over multiple days, not one sitting
- Every run is logged with date, platform, and model version, so any finding can be traced back to its evidence
- Variance is disclosed in the report: how often an answer mentioned the brand across runs, not whether one lucky run did
- This is also how to judge any audit vendor: if a report shows single-run screenshots with no dates, model versions, or variance notes, treat it as an anecdote
3. Technical AI-Visibility Scan
In parallel, we check whether the site itself is readable to the systems that feed AI answers.
- Crawler access: robots.txt rules for AI crawlers, rendering issues that hide content, and pages that answer buyer questions but are unreachable
- Structured data: schema markup on the pages most likely to be retrieved, checked for validity and coverage
- Entity signals: whether the brand name, description, and category are consistent across the public surfaces AI systems reconcile - site, directories, profiles, and knowledge sources
4. Source Analysis
For every sampled answer, we trace where it came from - because the sources are what you can actually influence.
- Catalog which pages, comparison articles, and community threads the answers cite, and open each cited source to verify what it says
- Map where competitors appear and the client does not - the competitor dominance zones that explain most omissions
- Assess which of those sources are realistic for the brand to earn honest presence on, and which are not worth pursuing
5. The Deliverable: Findings Report + Prioritized Playbook
The client receives two documents, not a data dump.
- The findings report: a representation snapshot across the prompt panel, a risk map (omission, misrepresentation, competitor dominance, and low-risk zones), and the entity and technical findings - each tied to logged evidence
- The playbook: a short ranked action list ordered by likelihood of impact - what matters now, what can wait, and what to deliberately ignore. A ranked short list, not a hundred-item backlog
- The team can execute the playbook internally, with another partner, or continue with us - the audit stands on its own either way
What We Work Toward
An audit is a diagnostic, so the outcomes are clarity outcomes - no growth promises, no guarantees. What a client should walk away with:
A defensible baseline
Citation share of voice across the fixed prompt panel - how often the brand appears versus competitors, sampled repeatedly and reported with variance notes, so the number can be defended and re-measured later.
A ranked gap list
The specific prompts and sources where competitors appear and the brand does not, ordered by likelihood of impact rather than listed alphabetically or padded for length.
A decision-ready investment view
Enough evidence to decide where to invest next - fix gaps internally, continue with a partner, or do nothing because the risk is genuinely low. All three are legitimate outcomes of a good audit.
A methodology you can rerun
The prompt panel, sampling procedure, and logging format are documented in the report, so any team can repeat the measurement later - with us or without us - and compare like with like.
Key Principles
- Fixed prompt panels - a baseline only means something if the same prompts can be resampled later under the same rules.
- Repeated sampling with variance disclosure - single screenshots are anecdotes; frequencies across logged runs are evidence.
- Evidence over opinion - every finding traces to an observed answer, a cited source, or a technical check the client can verify.
- Prioritization by likelihood of impact - the playbook is a ranked short list with explicit "ignore this" calls, because a backlog nobody executes helps nobody.
How We Measure
- Data sources: Repeated ChatGPT / Perplexity / Google AI Overviews prompt sampling, Google Search Console, Google Analytics 4, Site crawl and structured-data checks
- Timeframe: Sampling window during the audit; the fixed panel supports resampling at 4-8 week checkpoints if work continues
- Metric definition: Citation share of voice = the share of sampled assistant answers to the fixed prompt panel that mention or cite the brand, versus named competitors, with dates and model versions logged for every run. Technical findings = observed crawler-access, structured-data, and entity-consistency issues on the live site. See our methodology page for detailed definitions.
Validation & Evidence Standards
How Results Get Validated on a Real Engagement
On a live engagement, every reported metric is cross-checked across multiple data sources. We combine platform analytics, third-party tools, and observational methods to confirm directional trends.
Validation tools we use:
- Repeated prompt sampling across ChatGPT, Perplexity, and Google AI Overviews
- Google Search Console (branded and category query context)
- Site crawl tooling for robots.txt, rendering, and schema checks
- Third-party SEO tools (Ahrefs, SEMrush) for source-strength cross-checks
Cross-validation methods:
- Findings reported as frequencies across repeated runs, never from single outputs
- Cited sources opened and verified rather than trusted from the answer text
- Technical scan results verified against the live site before entering the report
About This Walkthrough
This walkthrough shows exactly how we run this type of engagement. Where figures appear, they illustrate the mechanics. We publish client numbers only with written permission and supporting data exports - transparency about method over dressed-up numbers.
Measurement Limitations
AI outputs are non-deterministic and vary by prompt wording, model version, and time. Our measurements are proxy-based and observational, not precise counts. Results should be interpreted as directional indicators rather than absolute guarantees. See our methodology page for detailed measurement definitions.
Replication Prompts
These are the kinds of prompts we track on an engagement like this. Try them (or your own category prompts) yourself - AI responses vary by model, wording, and time, so treat any single run as directional:
What are the best analytics tools for SaaS startups?Compare the top customer support platforms for B2B SaaSWhich subscription billing software do you recommend for a small SaaS team?
Want This Level of Clarity for Your Brand?
Start with an AI Visibility Risk Audit - a fixed-scope diagnostic of how AI systems currently represent your brand, with a findings report and a prioritized playbook.