⚠️ Prompt Scope (Critical Before You Analyze)
All charts in this dashboard:
Include all prompts
Do not auto-apply a Non-Branded filter
If analyzing competitive share of voice or discovery performance, apply the filter:
Branded = False
Why This Matters
Branded prompts can:
Inflate Presence %
Inflate Citation %
Mask discovery weaknesses
Skew competitor comparisons
Always confirm prompt scope before interpreting trends.
How to Read Scrunch Data
Scrunch monitors how AI platforms describe your brand across:
Prompts
Platforms
Time
Citations
Competitor entities
Key Principles
AI responses are non-deterministic
Scrunch collects in clean, non-personalized environments
Data is directional, not deterministic
Prompt × Platform = Prompt Variant (unit of analysis)
All metrics derive from:
Seed Prompts
Collected Responses
Extracted Citations
Configured Competitors
1. Brand Presence
How visible is my brand across AI platforms?
KPI Summary
KPI | Definition | Why It Matters |
Prompts | Unique Seed Prompts in range | Defines coverage |
Responses | Total AI responses collected | Statistical stability |
Presence % | % of responses mentioning brand | AI visibility rate |
Citation % | % citing your domain | Authority influence |
Prompts
Metric: Distinct Seed Prompts
Dimension: Seed Prompt ID
Low prompt count reduces reliability.
Responses
Metric: Count(Response ID)
Watch for:
Sudden drops
Platform outages
Sampling shifts
Presence %
Formula
Brand Presence % = SUM(Brand Mention Flag) / COUNT(Response ID)
Measures:
Mentions
Entity references
Brand domain mentions
Does NOT measure:
Traffic
Clicks
Conversions
Look For
Platform divergence
Branded vs non-branded gaps
Sudden weekly drops
Strong presence but weak citation
Citation %
Formula
Citation % = SUM(Brand Citation Flag) / COUNT(Response ID)
Presence ≠ Citation
You may:
Be mentioned but not cited
Be cited but not top-positioned
Signals
Pattern | Likely Meaning |
High presence + low citation | Content likely unstructured |
Citation spike | Newly indexed content |
Brand Presence Over Time (Line Chart)
Dimensions: Week, AI Platform
Metric: Brand Presence %
Aggregation: Weekly Avg
Identifies
Platform volatility
Algorithm shifts
Competitive movement
Optimization impact
2. Competitor Presence
How visible are competitors relative to me?
Competitive KPIs
KPI | Formula | Insight |
Avg Competitor Presence | AVG(Competitor Presence %) | Baseline share |
Competitor Citation % | SUM(Flag) / COUNT(Response) | Authority advantage |
If competitor average rises while brand declines → share shift.
High competitor citation often signals:
Better structure
Strong third-party presence
Crawlable resources
Competitor Trend Chart
Dimensions: Week, Competitor
Metric: Competitor Presence %
Use to detect:
Content launches
Promotions
Share shifts
Competitor Table
| Competitor | Presence % | Citation % |
This is your AI Share of Voice view.
3. Citations
Where is AI going for answers?
Citation Domain Table
Dimension | Description |
Source Domain | Extracted domain |
Domain Category | Brand / Competitor / Other |
Response Count | Citation frequency |
% of Total | Share of citation mix |
Custom domain extraction:
REGEXP_EXTRACT(Source URL, r'^(?:https?:\/\/)?(?:www\.)?([^\/:]+)')
Interpretation
Pattern | Meaning |
High "Other" | Third-party influence |
High competitor | Competitive gap |
There is no universal “good” citation %. Focus on trends.
Source URL Table
Use to:
Reverse engineer competitor structure
Identify formatting differences
Discover repeat-cited content
4. Prompts Table
Which questions drive or limit my visibility?
Key Metrics
Metric | Meaning |
Presence % | Visibility rate |
Position Score (1–100) | Placement order |
Sentiment Score | Tone classification |
Position Score
Metric: Brand Position Index
Low presence + high position = niche dominance
High presence + low position = buried mention
Look For
High-value prompts with low presence
Competitor dominance
Branded prompts skewing results
5. Sentiment
How positively is my brand described?
Sentiment Score (Looker)
Classification | Numeric Value |
Positive | 100 |
Neutral | 50 |
Negative | 0 |
Formula:
AVG(Sentiment Score)
⚠️ Neutral responses count as positive in classification.
Do not overinterpret small changes.
Difference From Scrunch UI
UI Dashboard | Looker Dashboard |
% positive / neutral / negative | Weighted average score (100/50/0) |
A score of 85 ≠ 85% positive.
UI = distribution
Looker = intensity
Both are useful.
Sentiment Trend Chart
Dimensions: Week
Metrics: Avg Sentiment, Avg Position, Avg Presence %
Signals
Pattern | Meaning |
Rising sentiment + falling presence | Niche authority |
Falling sentiment + rising presence | Visibility risk |
Domain Sentiment Influence
Formula:
(100 - Brand Sentiment) * Responses / 100
Reveals which domains shape AI framing.
Look for:
Review sites influencing tone
Third-party dominance
High-impact citation sources
6. Topic × Platform Heatmap
Where am I strong or weak by topic?
Dimensions
AI Platform
Prompt Tag
Tags come from prompt configuration. Remember, single prompts can have multiple tags at a time.
Interpretation Patterns
Pattern | Meaning |
Strong on one platform | Model-specific indexing difference |
Weak across all | Structural content gap |
Strong branded, weak discovery | Brand reliance |
Data Scope & Constraints
Weekly aggregation recommended
Non-personalized collection
Directional representation
Strategic Workflow
Confirm prompt scope
Check overall Presence %
Compare platform divergence
Review Topic heatmap
Analyze competitor citation dominance
Investigate top-cited URLs
Validate accessibility in Site Audit
Measure impact in 2–3 week intervals