53% of observed answers
Fictional sample · not a customer result
AI Visibility Audit Sample Report
See how a Signal OS report connects dated observations to transparent math, honest limitations, and a ranked action plan. Every brand detail and observation on this page is demonstration data.
30manually recorded observations
six frozen prompts
four named answer engines
six frozen prompts
four named answer engines
01 · Controlled scope
What this sample measured
A reader should be able to see the frame before interpreting a percentage. The same six frozen prompts were used throughout this fictional exercise.
Audit dateJuly 31, 2026
A single dated snapshot.
MarketUnited States
English-language answers.
AudienceMidsize SaaS
Operations leaders evaluating support tools.
OfferNorthstar CX
Fictional AI-assisted customer support platform.
Method: 30 manually recorded observations across ChatGPT, Claude, Perplexity, and Gemini. The sample records whether the invented brand appeared, whether its domain was cited, and whether each brand claim was factually consistent with the fictional source sheet.
02 · Executive readout
Counts before conclusions
Every percentage keeps its denominator visible. The confidence label reflects this sample’s own evidence rules—not a universal rating.
37% of observed answers
81% of brand mentions
23% observed noise band
03 · Prompt evidence
Where visibility broke down
Prompt-level counts show the commercial question behind each aggregate. The weakest result was the named-competitor comparison.
| # | Frozen buyer prompt | Mentions | Citations |
|---|---|---|---|
| 1 | Which customer support platforms should a midsize SaaS team consider? | 4 / 5 | 3 / 5 |
| 2 | What are the best alternatives to RelayDesk? | 1 / 5 | 0 / 5 |
| 3 | AI customer support software for a growing SaaS company | 3 / 5 | 2 / 5 |
| 4 | What does Northstar CX integrate with? | 5 / 5 | 4 / 5 |
| 5 | How can a SaaS company reduce its support backlog with AI? | 2 / 5 | 1 / 5 |
| 6 | What security controls matter in customer support automation? | 1 / 5 | 1 / 5 |
What “Stronger signal” means
The fictional run cleared the product’s demonstration threshold for repeated observations and consistency. It does not mean market leadership or predict what a future answer will contain.
Why the noise band matters
A 23% observed noise band is too wide for a confident movement claim. Repeat the highest-value prompts under the same controls before describing a trend.
Limit: this is a simulated convenience sample, not a census of AI answers. Signal OS does not guarantee citations, rankings, traffic, leads, sales, or revenue.
04 · Ranked actions
Do the most defensible work first
The action board converts observed gaps into scoped changes. Scores are transparent so a reviewer can challenge the order.
Correct and substantiate integration claims on the product page
Resolve the three inconsistent integration statements, link each supported claim to first-party documentation, then repeat the integration prompt.
Publish a RelayDesk alternatives page with first-party comparison evidence
Address the lowest-visibility commercial prompt with verifiable criteria, clear fit boundaries, and source links instead of unsupported superiority claims.
Add a security-controls explainer linked from technical documentation
Make the fictional product’s security facts easier to retrieve, while keeping legal and technical review attached to every claim.
Priority formula: 2 × (evidence strength + commercial intent + expected impact) ÷ effort. The inputs remain visible in the purchased workstation; the score supports a decision rather than replacing judgment.
Your next step
Start free—or build the client-ready system.
Use the free scorecard for a nine-observation baseline. Choose Agency when you need the complete offline workstation, commercial client-use license, pricing tools, QA controls, and delivery assets.