Compare runs

Overview

Compare runs (/prompts/compare) lets you put two prompt runs side by side and see what changed. For each monitored question, you see whether your brand was mentioned and your domain cited in Run A (older) versus Run B (newer), with highlights when either flag flipped.

This is the fastest way to validate whether a content publish, site restructure, or positioning change actually improved how AI answers your category questions.

Who it's for

• Content and SEO teams measuring before/after impact of site changes
• Founders checking if a launch or rebrand moved the needle in AI answers
• Anyone with at least two completed prompt runs who wants a quick delta view

Prerequisites

1. At least two prompt runs with Ask ChatGPT results.
2. Ideally, runs separated by a meaningful change (content update, new pages, repositioning).
3. On Free, remember each company prompt runs once—use Pro to re-run the same questions for true apples-to-apples comparison.

Step-by-step

Open compare runs

1. From /prompts, click Compare runs, or go directly to /prompts/compare.
2. Obsurfable loads your recent runs (up to 30).

Select two runs

1. Choose Run A (older)—typically your baseline before changes.
2. Choose Run B (newer)—your follow-up run after changes.
3. The comparison updates automatically for each shared prompt.

Read the comparison

1. Each card shows one buyer question.
2. Run A and Run B columns display: Mention (Yes/No) and Cite (Yes/No).
3. Amber callouts appear when mention or citation status changed between runs.

Investigate changes

1. Click through to individual runs from /prompts if you need the full AI response text.
2. Check your AI Visibility Score for the aggregate impact.
3. Generate Insights on the newer run for recommended next steps (Pro).

Tips

• Compare runs taken after the same site change—mixing unrelated edits makes it hard to attribute what worked.
• On Pro, re-run the same company prompts with web search enabled for both runs to compare live retrieval behavior.
• Look for citation changes even when mentions stay the same—citations often matter more for long-term retrieval.
• A single prompt flipping from No to Yes on mention is a win; prioritize patterns across multiple discovery prompts.
• Pair compare runs with Site analysis to connect content gaps to AI answer changes.

FAQ

Do both runs need the same prompts?

Compare works across the union of prompts in both runs. Shared prompts show the clearest comparison; prompts only in one run show partial data.

Can I compare more than two runs?

The compare page supports two runs at a time. For broader history, review individual runs on /prompts or check the visibility trend chart.

Why don't I see any changes?

AI answers can be stable between runs, especially without web search. Try Pro web search, ensure meaningful site changes occurred between runs, or wait longer between measurements.

Does compare show the full AI response?

No—compare focuses on mention and citation flags. Open each run's detail page for full response text.

I'm on Free and can't re-run prompts. Can I still compare?

You can compare any two runs that exist, but Free limits each company prompt to one lifetime run. For repeated before/after testing on the same questions, upgrade to Pro.

How is this different from the visibility score trend?

The visibility score trend shows your composite score over time. Compare runs shows per-question mention and citation deltas between two specific runs.