Snapshot Comparison

Overview

AI models do not produce static, fixed answers. The same prompt can yield different responses over time as models are updated, as new training data is incorporated, and as the competitive landscape shifts. Snapshot Comparison lets you place two snapshots side by side to see exactly what changed between them — down to the individual word.

This feature answers questions that raw metrics alone cannot:

• What specifically changed in the AI response, not just that a score went up or down?
• Did the AI model drop your brand from its answer, or did it change how your brand is described?
• Which citations were added or removed, and does the AI now reference a competitor's content instead of yours?
• How did sentiment shift — and what language drove that change?

Snapshot Comparison gives you the evidence behind the numbers.

How to Access Comparison

1. Navigate to Snapshots from the sidebar.
2. Select exactly two snapshots by clicking the checkbox on each row.
3. Click the Compare button that appears in the toolbar.

The comparison panel opens as a full-screen overlay with your two snapshots labeled Snapshot A (the first selected) and Snapshot B (the second selected).

Tip: For the most meaningful comparisons, select two snapshots of the same prompt captured at different times or from different AI models. Comparing snapshots from the same prompt reveals how an AI engine's answer has evolved. Comparing the same prompt across different models reveals how each engine frames your brand differently.

View Modes

The comparison panel offers two viewing modes. Toggle between them using the buttons in the header.

Side by Side

The default view displays both snapshots in parallel columns. Each column shows the full metadata and response for one snapshot:

• Prompt — The question that was sent to the AI model.
• Source — Which AI model produced the response (e.g., GPT-4o, Claude, Gemini).
• Date — When the snapshot was captured.
• Brand Mentioned — Whether your brand appears in the response.
• Sentiment — The detected tone of the response (Positive, Neutral, or Negative).
• Rank Position — Where your brand appears in any ordered list within the response.
• Full Response — The complete AI-generated text.
• Citations — All URLs referenced in the response.
• Competitors Mentioned — Other brands detected in the response.

Side by Side is best for getting a quick overview of both snapshots and manually scanning for differences.

Diff View

Diff View merges the two responses into a single document with word-level highlighting that makes every change immediately visible:

• Green highlighted text indicates words that were added in Snapshot B (not present in Snapshot A).
• Red highlighted text with strikethrough indicates words that were removed from Snapshot A (not present in Snapshot B).
• Unhighlighted text is unchanged between the two snapshots.

Above the diff, compact summary cards show the metadata for each snapshot so you can reference dates, models, and metrics without switching views.

A legend at the top of the diff section labels the color coding for quick reference.

Note: For very long responses (over 800 words), the diff algorithm automatically switches from word-level to sentence-level comparison. This keeps the comparison fast and readable without sacrificing accuracy. The visual presentation remains the same — you will still see green and red highlighting — but the granularity shifts from individual words to complete sentences.

Understanding the Change Summary

Immediately below the header, a Change Summary Strip displays badges that highlight the most important metadata differences between the two snapshots at a glance. This strip appears in both view modes.

The following changes are detected and surfaced as badges:

If no metadata changes are detected, the strip displays a message indicating that the responses may still differ in content even though the high-level metrics are the same.

Citation Comparison

In Diff View, a dedicated Citation Comparison section categorizes every cited URL into three groups:

• Shared — Citations that appear in both snapshots. These are stable references the AI model consistently relies on. Displayed with a link icon.
• Only in A — Citations present in Snapshot A but absent from Snapshot B. These are sources the AI model stopped referencing. Displayed in red with a minus icon.
• Only in B — Citations present in Snapshot B but absent from Snapshot A. These are newly referenced sources. Displayed in green with a plus icon.

Each citation shows the source domain with its favicon for easy identification. Click any citation to open the original URL in a new tab.

This breakdown is especially valuable for understanding which content sources are gaining or losing influence in AI-generated answers. If a competitor's page appears in the "Only in B" column, it means the AI model has started citing their content more recently.

Competitor Comparison

The Diff View also includes a Competitors Mentioned section that visualizes changes in the competitive landscape between the two snapshots:

• Competitors mentioned in both snapshots appear as standard badges.
• Competitors mentioned only in Snapshot A appear with a red strikethrough, indicating the AI model dropped them.
• Competitors mentioned only in Snapshot B appear in green with a "+" prefix, indicating the AI model introduced them.

Copying and Sharing Results

Click the Copy Summary button in the comparison header to copy a formatted plain-text summary of the comparison to your clipboard. The copied text includes:

• Prompt text, source, date, and key metrics for both snapshots
• A bullet list of all detected changes
• The full response text from both snapshots

This makes it easy to paste comparison results into reports, emails, Slack messages, or internal documents without taking screenshots.

Use Cases

Tracking AI Response Evolution

Capture the same prompt weekly or daily, then compare snapshots across time periods. This reveals how AI models gradually change their answers — whether your brand is being mentioned more or less frequently, whether your rank in lists is improving, and whether the overall narrative is shifting.

Monitoring Brand Narrative Changes

When you detect a sentiment change on your Dashboard, use Snapshot Comparison to find the exact language that drove the shift. A sentiment score moving from Positive to Neutral is actionable only when you can see that the AI model replaced "industry-leading" with "one of several options."

Evaluating Cross-Model Differences

Compare snapshots from different AI models for the same prompt to understand which engines favor your brand. If GPT-4o ranks you first but Gemini omits you entirely, Snapshot Comparison shows exactly what each model says so you can tailor your optimization strategy by platform.

Validating Optimization Impact

After publishing new content, updating structured data, or changing your SEO strategy, compare a "before" snapshot with an "after" snapshot for the same prompt. The diff view will show you precisely what changed in the AI response, confirming whether your optimization efforts had the intended effect.

Investigating Citation Shifts

Use the Citation Comparison to monitor which sources AI models reference. If a competitor's blog post appears in the "Only in B" column, that competitor may have published content that is now influencing AI answers — and you may need to respond with your own authoritative content.

Related Features