Primary KPIs (Highest Value)
These are the North Star metrics that are consistently tracked and valued most. Each metric below is named the way it appears in the Explorer picker, with its underlying API field noted for reference.
Looking for the exact Explorer name or API field for a metric? The Scrunch Metrics Reference lists every metric, breakdown and filter with the API field it maps to.
Explorer supports up to 5 metrics, 3 breakdowns, and 8 filters in a single view, so you can combine several of these KPIs (for example Mentions Rate, Citations Rate, and Sentiment Average) into one table or chart.
1. Mentions Rate
What it measures: Percentage of relevant AI-generated answers where your brand is mentioned by name (API field: brand_presence_percentage). This metric was previously called "Brand Presence"; in the Explorer picker it now appears as "Mentions Rate."
Why it's valuable:
Most widely tracked metric across all use cases
Easy to understand and explain to stakeholders
Establishes a clear trend line over time
Core indicator of overall AI visibility
Best practices:
Pull regularly to establish baseline and track progress
Use as the foundation for quarterly reviews
Compare against previous periods to show momentum
Check the sample size shown in the chart tooltip (for example, "43% (37 of 86 responses)") before drawing conclusions from a small number of responses
2. Citations
What it measures:
Citations Rate (API field: brand_citation_rate): share of responses citing at least one source URL owned by your brand, independent of whether the brand is mentioned by name
Citations Total (API field: brand_citation_count): total number of brand citations across AI-generated answers
Citations Share of Voice (API field: brand_citation_share_of_voice): of responses that cite anything, the share citing a source owned by your brand versus competitors. In the legacy Citations dashboard, this was called "Citation Consistency."
Citations Total broken down by Citation Source Owner (API field: citation_unique_response_domains): your brand's share of total citation volume. In the legacy Citations dashboard donut chart, this was called "Citation Share."
For the full breakdown of every citation metric, see Understanding Citation Metrics in Scrunch.
Why it's valuable:
Tangible proof of brand mentions
Direct indicator of Citations Share of Voice (distinct from the mention-based Share of Voice metric) in AI search
Competitive intelligence (e.g., "We're cited in 49% of responses vs. Competitor at 83%")
Often reported as the #1 tracking metric
Best practices:
Break down by topic or customer journey stage
Identify third-party sites mentioning competitors but not you = outreach opportunities
Track citation sources to identify high-value domains
Remember that Grok no longer contributes citations, so it won't appear in citation-based breakdowns, even though its presence, ranking, and sentiment data are unaffected
Insight: "Citations are the most important KPI to track when brand presence might be skewed by prompt selection or geography."
3. AI Referral Traffic
What it measures: Website sessions driven by specific AI models (ChatGPT, Perplexity, Gemini, etc.) via a Google Analytics integration
Why it's valuable:
Direct business impact metric
Connects AI visibility to actual website traffic
Easier for executives to understand than abstract visibility metrics
Demonstrates ROI of AI search optimization efforts
Best practices:
Track by individual AI platform to identify which are driving most value
Monitor conversion rates from AI traffic vs. other channels
Use as proof point for expanding AI search investment
Secondary KPIs (Still Important)
These metrics provide additional context and competitive intelligence:
4. Competitive Presence
What it measures: Brand performance compared to key competitors across the same prompts. See Understanding Competitive Presence in Scrunch for a full walkthrough.
Why it's valuable:
Identifies gaps and opportunities
Shows relative market position in AI search
Helps prioritize optimization efforts
Best practices:
Focus on 2-4 key competitors rather than tracking too many
Compare your Mentions Rate (brand_presence_percentage) or Share of Voice (brand_share_of_voice) against competitor_presence_percentage and competitor_share_of_voice to quantify gaps
Filter prompts by competitor to find specific gaps
5. Position Average
What it measures: A 0-100 positioning score for the brand (API field: brand_position_score), averaged over responses that mention it: top=100, middle=50, bottom=0. In the Explorer picker this appears under Position as "Position Average."
Why it's valuable:
Similar to traditional SEO rankings
Indicates prominence within responses
Useful for tracking improvement over time
Important context:
Monitor but don't over-index on this metric
Similar to SEO: drop-off is high after "page 2"
Question whether users read entire lengthy LLM responses
Watch for unusual trends rather than focusing heavily on the absolute score
6. Sentiment Average
What it measures: Breakdown of positive, mixed, negative, and unclear brand mentions. Responses with no detectable sentiment fall into an "Unclear" band. See Understanding Sentiment in Scrunch for how each band is calculated.
Why it's valuable:
Quality of mentions, not just quantity
Early warning system for negative narratives
Helps identify messaging opportunities
Important context:
Mixed sentiment is typically fine (pro/con lists, comparisons)
Only investigate if negative sentiment becomes significant
Break sentiment down in Explorer, or check the Signals tab to flag concerning trends
7. Content Delivery Status
What it measures: Whether AXP is currently serving optimized content for a cited page (API field: content_delivery_status), across four buckets: "optimized" (AXP-deployed), "not_optimized" (your own page, not yet optimized), "control_group" (deliberately excluded as a baseline), and "na_third_party" (the cited page isn't your site at all). See Understanding the Content Delivery (AXP) tab for more detail.
Why it's valuable:
Shows how much of your cited content is actually being served through AXP optimization
Highlights pages that are eligible for optimization but haven't been enabled yet
Lets you separate a genuine control-group baseline from pages that simply haven't been optimized
Best practices:
Break citations down by Content Delivery Status to see what share of your cited pages are optimized
Watch for pages sitting in "not_optimized" that are still earning citations; they're good next candidates for AXP
Use the control group bucket to validate the impact of optimization over time