Benchmarks and baselines for evaluating AI search performance
To evaluate and improve AI search performance, start with a clear baseline. Track a focused set of metrics over time — brand presence, citation share, referral traffic, and share of voice against competitors. Scrunch recommends monitoring a representative prompt set across multiple AI platforms, including ChatGPT, Perplexity, and Google Gemini. Compare your brand's exposure and citations to direct competitors for the same prompts.
Baseline metrics to monitor
| Metric |
What it measures |
| Brand presence |
How often your brand is mentioned or included in AI responses, segmented by persona, country, topic, platform, funnel stage, or custom tag. |
| Citations |
Frequency of your brand being cited in AI answers to business-relevant prompts, plus which third-party sources are most cited for your topics. |
| Referral traffic |
Website traffic that comes from AI responses to your target prompts. |
| AI agent traffic |
How often AI agents visit your site for training, indexing, and retrieval. |
| Share of voice vs. competitors |
Your brand's presence and citation share versus key competitors for the same prompts. |
For a deeper dive on setting baselines, see the Scrunch blog on establishing initial benchmarks and baselines.
Measuring your brand’s performance over time
- Track trends in 2–3 week windows to separate real movement from one-off changes.
- Compare like-for-like periods (e.g., Q2 vs. Q1) by adjusting date ranges for brand presence, prompt placement, sentiment, and citations.
- Analyze results side-by-side with competitors using line charts to spot relative gains or losses over time.
- Review which sources AI models cite most for your priority topics. Identify gaps in coverage that your content can fill.
If visibility declines, follow these steps:
- Check Insights to spot issues with content relevance or metadata.
- Identify citation gaps to see which sources are winning and why.
- Prioritize fixes on one or two high-impact topics.
- Verify AI agents can crawl your site (robots.txt and page rendering).
Learn more in the related FAQ on tracking visibility over time.
Comparing your visibility to competitors
- Measure AI share of voice by running the same prompt set across multiple AI platforms, then view results in aggregate and by platform to see where performance diverges.
- Track both presence and citation share to capture exposure and authority signals.
- Use filters (persona, country, topic, funnel stage) to understand where you win or lag.
Get step-by-step directions in the guide on how to measure AI share of voice.
How Scrunch measures AI visibility and trends
Scrunch tracks the exposure and authority signals that determine how AI systems present your brand. It measures brand mentions, citations, prompt placement, and sentiment across a structured prompt panel covering multiple AI platforms.
- Measures brand mentions, citations, prompt placement, and sentiment across a representative prompt panel and multiple AI platforms.
- Captures referral traffic from AI answers and AI agent traffic to your site.
- Surfaces issues and opportunities via Insights, and lets you drill down by persona, country, topic, platform, funnel stage, or custom tags.
How Scrunch differs from traditional keyword SEO tools:
- Traditional SEO tools estimate keyword search volume in web search engines such as Google and Bing.
- AI platforms do not publish comparable search volume data. Scrunch tracks observable outcomes instead: how often and how prominently your brand appears and is cited for target prompts.
- Treat it as a trend index of AI visibility. Establish your current exposure as a baseline, then monitor directional change over consistent 2–3 week intervals.
How Scrunch compares to other AEO platforms
When evaluating answer engine optimization (AEO) tools, assess each option against these six criteria:
| Criteria |
Why it matters |
| Model and platform coverage |
More platforms means a broader view of AI visibility across different user bases. |
| Data filtering and segmentation |
Granular filters (persona, region, funnel stage) surface where you win or lag. |
| Ease of use and reporting |
Clear dashboards reduce time-to-insight for teams without data science resources. |
| Competitive benchmarking |
Side-by-side comparisons reveal relative share of voice gains or losses. |
| Citation and source analysis |
Identifies which third-party sources AI models favor for your topics. |
| Vendor experience, reviews, and pricing |
Validates reliability, support quality, and cost fit for your team. |
With Scrunch, teams can:
- Monitor the same prompts across AI platforms. Segment results by persona, region, topic, funnel stage, and more.
- Benchmark brand presence, citation share, sentiment, and prompt placement against competitors.
- Use Insights to automatically flag content and metadata issues that may limit AI visibility.
- Pair measurement with action via Agent Pages to deliver AI-optimized content to AI agents visiting your website.
Quick example
Example: An HR software company wants to determine whether Q2 performance beat Q1. Follow these steps to run the comparison in Scrunch:
- Set the date range to Q2. View brand presence, competitive presence, prompt placement, sentiment, and citations.
- Switch the date range to Q1. Compare the numbers side by side.
- Plot brand presence vs. competitors as a line chart to see relative movement.
- If performance dropped, review Insights. Check which sources were cited instead, address gaps, and confirm AI agents can crawl your pages.
Related resources