What an AIO Diagnosis Can Tell You, and Its Limits
Tomohiro Iida ·
An AIO diagnosis is only as interpretable as the conditions it actually observed. Before comparing scores or recommendations, check when the observation ran, which AI service and mode were used, what questions were asked, how many times they ran, whether a market or region was specified, and how failures or unobserved runs were handled.
Key takeaways
- A diagnosis result applies to the time, service, search mode, questions, repetitions, and other conditions that were actually observed.
- A failed, unobserved, or indeterminate run is different from a successful observation that found zero citations.
- For SIGNAL, the public offer and the runtime/report that a user actually sees are separate facts. Report layout and citation-URL visibility vary by active report mode, so this page does not promise one fixed free-report screen.
1. Check the observation conditions before the result
AI-search answers are not fixed values. Results can change with the service, model or mode, search availability, execution time, and wording of the question. A useful diagnosis therefore makes the observation conditions inspectable before asking the reader to compare results.
- Observation date and time
- AI service and model or mode
- Whether search was enabled
- Question language, and whether a market or region was specified
- The actual questions used
- Number of runs per question
- How many runs could not be observed, and why
For market and region, keep the requested value separate from what the provider actually confirmed. If the observation geography cannot be confirmed, leaving it unknown is more accurate than copying the requested region into the confirmed field.
2. What an AIO diagnosis can examine
A general AIO diagnosis can inspect public information and sampled AI answers to see how a company is understood in search and AI surfaces. Public-page checks and permissioned first-party data analysis should be treated as separate scopes.
- How the company, services, and business are described on public pages
- How AI answers describe, recommend, or cite the company and competitors under the observed questions
- Whether the official site contains evidence and explanations that search and AI systems can retrieve
- Public-page information architecture such as title, description, headings, internal links, and structured data
- Technical retrieval problems that occur before a page can become a candidate for search or AI answers
Actual Search Console indexing data, GA4 traffic and behavior, and downstream inquiry or pipeline impact cannot be established from public pages alone. Those require permissioned data analysis.
3. What the result cannot establish
- Treating an unobserved question as evidence that there is no problem
- Treating a failed or indeterminate observation as zero citations
- Identifying a cause from a single observation
- Attributing a change in metrics solely to the work that was performed
- Guaranteeing future inclusion, citation, recommendation, ranking, inquiries, or revenue
The distinction between “unobserved” and “zero” matters. Provider failures, missing search results, or an observation that could not be judged should not be converted into a successful zero-citation observation. Doing so also corrupts before/after comparisons.
The free diagnosis uses public information and available AI observations to identify an entry point for improvement. GA4, Search Console, and card registration are not required.
4. What can currently be said about SIGNAL’s free diagnosis
The current free diagnosis uses public information and, on applicable paths, AI observations made at diagnosis time. A diagnosis with zero retrieved pages or zero publishable content is treated as a failure rather than a successful zero-result report. If AI observation fails partly or entirely while other publishable diagnosis content remains, a report can still be published and the processing state can be partial success.
Language, market, and region need careful wording
The free diagnosis uses Japanese questions and its intake and result interfaces are Japanese. Stored or displayed locale and region information does not, by itself, prove that the AI provider confirmed a particular observation market.
The free report is not one fixed layout
SIGNAL has multiple free-report rendering paths depending on the stored report generation and active feature flags. The presentation of findings, observation states, own-domain citation counts, observation conditions, and reference URLs differs by path. This page therefore does not promise that every free diagnosis shows one particular screen or field set.
One path suppresses the normalized citation-URL list in the free output, while another report mode can display per-question reference URLs. We therefore do not state that citation URLs are always free or always paid at this evidence point.
5. Offer copy and the paid scope
SIGNAL’s public offer describes the free diagnosis as an entry point toward a high-priority hypothesis and one first correction instruction. That is offer scope, not proof that every active report mode renders the same bundle on screen. The actual rendered composition depends on the active report path.
- Free diagnosis — offer scope
- Public-information diagnosis; the number and conditions of search / generative-AI checks; observed facts, interpretation, business hypothesis and how to verify it; target URL and observed facts; one first correction instruction.
- Paid scope added in the offer
- GA4 and Search Console first-party analysis, continued observation, improvement content for multiple locations, approval and implementation management, and post-implementation measurement.
6. Choose the next page by the problem stage
Frequently asked questions
- What can an AIO diagnosis tell me?
- It can inspect public information and AI answers under specified conditions. To interpret the result, check the observation time, service, search mode, questions, repetitions, language and market or region specification, and the count of unobserved runs.
- Can an unobserved question be treated as zero citations?
- No. Failed, unobserved, or indeterminate runs are different from a successful observation that found zero citations.
- Does SIGNAL’s free diagnosis observe the Japanese market?
- The free diagnosis uses Japanese questions and its intake and result interfaces are Japanese. Stored or displayed locale and region information does not, by itself, prove that the AI provider confirmed a particular observation market.
- Are citation URLs visible in the free diagnosis?
- This page does not make a universal claim. SIGNAL has multiple stored/report paths. One path suppresses the normalized citation-URL list in free output, while another can show per-question reference URLs.
- What is added in paid analysis?
- The public offer places GA4 and Search Console first-party analysis, continued observation, multi-location improvement work, approval/implementation management, and post-implementation measurement in the paid scope.