SpyderBot Docs
Search Sign in

SpyderBot Documentation

Reading Results

A Prompt Explorer investigation produces more than an AI-generated response.

Overview

Each investigation combines AI responses with supporting observations, evidence, comparative analysis, and investigation findings to help explain why AI behaved the way it did.

Rather than reading the results from top to bottom, Prompt Explorer is designed to support an investigation workflow.

Its purpose is not simply to show what happened, but to help you understand why it happened.

Investigation Mindset

Prompt Explorer should be approached as an investigation rather than a report.

Instead of asking:

What response did AI generate?

Ask:

  • Why did AI generate this response?
  • What evidence supports this observation?
  • Is this behavior consistent across AI models?
  • Is this an isolated observation or a recurring pattern?
  • What should we investigate next?

This mindset helps transform AI responses into actionable investigation findings.

Investigation Workflow

We recommend reviewing Prompt Explorer investigations in the following order.

Business Question
↓
AI Responses
↓
Supporting Evidence
↓
Cross-Model Comparison
↓
Investigation Findings
↓
Next Investigation

Each stage builds additional context for understanding AI behavior.

Step 1 — Start with the Business Question

Every investigation begins with a business question.

Before reviewing the results, remind yourself:

  • What were we trying to understand?
  • Which decision will this investigation support?
  • Why was this prompt selected?

Keeping the original objective in mind helps prevent overinterpreting individual observations.

Step 2 — Review AI Responses

Next, review the responses generated by the selected AI models.

Focus on:

  • Overall response themes.
  • Recommendations.
  • Competitor mentions.
  • Brand mentions.
  • Response structure.

Avoid drawing conclusions from wording differences alone.

The objective is to identify meaningful behavioral patterns rather than compare individual sentences.

Step 3 — Review Supporting Evidence

After identifying initial observations, review the supporting evidence.

Evidence may include:

  • Observation coverage.
  • Confidence indicators.
  • Entity relationships.
  • Citation patterns.
  • Cross-model consistency.

Evidence helps determine whether an observation represents a recurring pattern or an isolated AI response.

For additional information:

Step 4 — Compare AI Behavior

Prompt Explorer is designed to investigate behavior across AI models rather than evaluate a single response.

Compare:

  • Recommendation behavior.
  • Citation behavior.
  • Entity recognition.
  • Response structure.
  • Competitive positioning.

Look for observations that remain consistent across multiple AI systems.

Consistent behaviors generally provide stronger evidence than isolated differences.

Step 5 — Form Investigation Findings

After reviewing responses and supporting evidence, identify the findings that best explain AI behavior.

Examples include:

  • Competitor recommendations are driven by stronger entity recognition.
  • AI consistently associates the brand with a specific category.
  • Commercial prompts produce different recommendation patterns.
  • Multiple AI models reference similar authoritative sources.

Investigation findings should explain observations rather than simply repeat them.

Common Investigation Patterns

Organizations frequently identify patterns such as:

Consistent AI Behavior

Multiple AI models generate similar observations.

These findings often represent stable AI behavior.

Model-Specific Behavior

One AI model behaves differently from the others.

Further investigation may be required before drawing strategic conclusions.

Competitive Differences

Competitors consistently receive stronger recommendations or citations.

These findings often become candidates for additional Prompt Explorer investigations.

Emerging Changes

New behaviors appear that were absent in previous investigations.

Organizations should determine whether these represent temporary variation or meaningful AI ecosystem changes.

Best Practices

Investigate Before Concluding

Do not rely on a single response or metric.

Review all supporting observations before drawing conclusions.

Focus on Patterns

Recurring behaviors generally provide more valuable insights than isolated AI responses.

Validate Important Findings

When findings influence strategic business decisions, repeat the investigation or expand the observation depth before taking action.

Continue Investigating

Most investigations generate new questions.

Prompt Explorer is designed to support iterative investigation rather than one-time analysis.

Relationship to Prompt Observatory

Prompt Explorer explains current AI behavior.

Prompt Observatory explains how that behavior changes over time.

If an investigation identifies a strategically important prompt, consider creating a Prompt Observatory to monitor future changes automatically.

Prompt Observatory

Related Concepts

To better understand Prompt Explorer investigations:

Related Pages

Reading Results works together with:

These pages provide additional guidance for interpreting investigation findings and evaluating their reliability.

Next Steps

Once you understand the investigation findings: