Co-occurrence analysis
Which brands the AI consistently mentions alongside yours.
Last updated: 2026-05-05
Co-occurrence analysis
Co-occurrence shows which brands the AI mentions alongside yours — and, more importantly, which pairings happen more often than chance. If you and a specific competitor co-appear in 60% of queries when each of you only appears in 30% individually, that's a strong signal you're being framed as direct alternatives.
How it's calculated
For every pair of brands in your scans, Pranas computes a lift statistic:
lift(A, B) = P(A and B together) / (P(A) × P(B))
- Lift = 1 → they appear together exactly as often as random chance predicts.
- Lift > 1 → they appear together more than chance — they're being framed as related (often as alternatives, sometimes as complements).
- Lift < 1 → they appear together less than chance — the AI sees them as belonging to different sub-categories.
The widget reads from the same Unaided cache as the Home dashboard so co-occurrence is always 1:1 consistent with your SOV and Top Brands numbers.
Lift reference
| Lift value | What it means | Typical interpretation | Action |
|---|---|---|---|
| Greater than 1 | Co-appear more often than chance | Direct alternatives, complements, or a category-defining pair | Register the brand and plan comparison content |
| Exactly 1 | Co-appear as often as chance predicts | No meaningful relationship | No action needed |
| Less than 1 | Co-appear less often than chance | The AI sees you in different sub-categories | Confirm your category positioning is intended |
Widget controls
| Control | Default | Purpose |
|---|---|---|
| Minimum query support | 3 co-appearing queries | Filters out one-off noise |
| Core-theme flag | Starred rows | Highlights highest-priority competitive rows |
| Rubric denylist | Enabled | Strips generic terms like "alternatives to" and "vs" |
What "high lift" usually means
- Direct alternatives — most common. Both brands solve the same problem and the AI lists them in the same breath. This is your shortlist competitor.
- Complementary tools — the AI recommends them as a stack (e.g., a CMS + a CDN). Co-marketing potential.
- Category-defining pair — when two brands define a category, the AI struggles to mention one without the other.
Filters and threshold
- Minimum query support — by default, pairs are only shown if they co-appear in at least 3 queries. This prevents noise from one-off mentions.
- Core-theme flag — pairs that are flagged as belonging to your core theme are starred. These are the highest-priority competitive intelligence rows.
- Rubric denylist — generic comparison terms ("alternatives to", "vs") are filtered to prevent fake pairings.
How to use it
- Refine your competitor list. A high-lift brand you haven't registered should usually be added. Click the brand name to register and trigger retro-tagging.
- Plan comparison content. If lift with Brand X is high but your Comparison-prompt SOV against X is low, write a head-to-head landing page.
- Identify positioning drift. If a brand from an adjacent category starts co-appearing with you frequently, the AI is blurring category lines — reinforce your positioning.
Related articles
- Top Brands widget
- Share of Voice (SOV)
- Manage competitors
- Narrative Drivers