Probe Issues

Diagnose and fix IndexMind AI probe failures, engine rate limits, empty results, and incomplete probe data.

The probe stage queries major AI language models (GPT-4, Claude, Gemini, and others) with questions relevant to your website's industry and content. Probe failures can result in incomplete AI visibility scores. This guide covers the most common probe problems and their solutions.

Probe timeout

Symptom. The probe stage shows "Timed out" after running for several minutes. The AI Visibility pillar may show incomplete data.

Common causes:

  • High AI engine latency. External AI providers occasionally experience slower response times during peak usage. This is outside IndexMind's control.
  • Large probe set. Sites with many pages and broad topical coverage generate more probe queries. The probe stage may need more time to complete.

Typical probe times by query count:

  • 5 queries: 1 to 2 minutes
  • 20 queries: 5 to 10 minutes
  • 50 queries: 15 to 25 minutes

Solutions:

  1. Retry the probe stage. Click Retry next to the error on your project dashboard. Transient latency issues typically resolve on retry.
  2. Run the analysis during off-peak hours. AI provider response times tend to be faster during weekday mornings (US Eastern time).
  3. Reduce crawl scope. Fewer crawled pages produce fewer probe queries, which reduces total probe time.

AI engine rate limits

Symptom. The probe report shows "Rate limited by AI provider" or similar errors. Some probe queries completed while others did not.

Why this happens. IndexMind distributes probe queries across multiple AI providers (ChatGPT, Perplexity, Gemini, Claude). If one provider hits its rate limit, queries to that provider fail. IndexMind automatically retries with exponential backoff, but persistent rate limiting can cause partial results.

Solutions:

  1. Wait and retry. Rate limits are temporary. Wait 15 to 30 minutes and retry the probe stage. Avoid running multiple probes simultaneously on the same project.
  2. Contact support if the issue recurs. Repeated rate limiting for the same project may indicate the probe configuration needs adjustment. Email support@indexmind.com with your project name.

Rate limit errors from AI providers are temporary and do not affect your monthly analysis quota. A failed probe stage does not count as a completed analysis.

Empty probe results

Symptom. The probe stage completes successfully, but the AI Visibility pillar shows a score of 0 or "No data available."

Empty results are a valid outcome. They mean AI language models did not mention your site in their responses for the tested queries. The recommendations tab will suggest specific improvements.

Common causes:

  • New or very niche website. AI language models may not have encountered your content during their training. This is expected for newly launched sites or highly specialized industries.
  • Insufficient crawl data. If the crawl stage collected very little content (fewer than 5 pages with minimal text), IndexMind may not generate enough meaningful probe queries.
  • Content mismatch. If your site's content spans unrelated topics, the generated queries may not align well with any single topic.

Solutions:

  1. Expand your content. Sites with comprehensive, well-structured content generate better probe queries and are more likely to appear in AI responses.
  2. Verify crawl completeness. Check the crawl report to ensure the crawler accessed your key pages. A thin crawl produces thin probes.
  3. Add structured data. Schema markup helps IndexMind understand your site's topics and generate more targeted probe queries.

Partial probe data

Symptom. The probe stage shows a mix of successful and failed queries. Your AI Visibility score is based on incomplete data.

IndexMind calculates scores using the data it has. Partial probe data produces a directionally accurate score, but the confidence is lower. The report indicates when a score is based on partial data.

What to do:

  1. Review the partial results. Even incomplete data reveals which AI models mention your site and which do not. This is actionable information.
  2. Retry the failed queries. Click Retry on the probe stage to attempt the failed queries again without re-running successful ones.
  3. Run a full analysis later. If retry does not resolve the issue, start a new full analysis at a different time.

What if a specific engine always fails

Individual engine failures are logged in the probe job details. Open the probe jobs list from the AI Visibility tab to see per-engine results. If one engine consistently fails, the other engines still provide citation data. Engine availability depends on the external provider and is outside IndexMind's control.

Probe queries seem irrelevant

Symptom. The probe report lists queries that do not match your business or industry.

Why this happens. Probe queries are auto-generated based on the content IndexMind collected during the crawl stage. If the crawl captured non-representative pages (e.g., only blog posts, footer pages, or error pages), the queries may miss your core topics.

Solutions:

  1. Review your crawl scope. Ensure your key product, service, or topic pages are included in the crawl. Adjust path inclusions in project settings if needed.
  2. Add or improve structured data. JSON-LD schema markup gives IndexMind stronger signals about your site's primary topics and industry.
  3. Increase crawl depth. If key pages are more than 3 levels deep, increase the crawl depth in project settings.

Next steps

If probe issues persist after trying these solutions, contact support with your project name and a screenshot of the probe error. See also the technical troubleshooting documentation for advanced probe diagnostics.