Competitor AI Gap Analysis
A 7-day playbook to find your 'real competitors' in AI answers and reclaim source share
When you hear 'our name doesn't appear in AI search,' the more accurate diagnosis is this — it's not that your name isn't appearing, but that someone else is taking your spot. That 'someone else' might not be a direct competitor. News outlets, comparison platforms, and public institutions are often the bigger issue.
What you'll get
Success Metrics
Step-by-step guide
- Check top 10 domains in GEO Insights → Visibility Leaderboard (last 28 days)
- List 3-5 direct competitors
- List 5-10 source competitors (news, institutions, comparison platforms, communities)
- Write 1-page domain share summary
When your share is low and 'Others' is high, many teams think 'we need more content,' but the real reason is usually a source structure problem. Separate the causes first.
- Classify keywords by gap type — only us strong / both strong / both absent / only competitor strong
- Select 5-10 Quick Win candidates
Strategy by Gap Type
| Type | Meaning | Action Now |
|---|---|---|
| Only us strong | Maintain existing strength | Keep freshness and structure |
| Both strong | Intense competition | Differentiate with depth/trust/freshness |
| Both absent | Preemption opportunity | Build reference hub first |
| Only competitor strong | Recapture priority target | Top priority for quick improvement |
Brand head keywords are hard to overturn with content alone since news/portals are strong. But procedure/condition keywords like 'claim documents' can be quickly reversed with reference hubs.
- Compose Top 20 questions (10 recommendation/comparison + 10 procedure/caution)
- Create mapping table of 'who's winning now' per question
How to Interpret Question Gaps
- •We're mentioned but not cited → document structure issue (no answers, FAQs, tables)
- •Competitors are cited but we're not → trust structure issue (weak source evidence)
- •Only news/institutions are cited → source graph issue (our domain isn't linked as reference)
- Derive 'our weakest questions Top 5' per platform
- Create improvement checklist per platform
Platform-Specific Attack Points
- •ChatGPT — Prefers well-organized documents with clear definitions, summaries, FAQs. One-sentence definition in first paragraph is essential
- •Claude — Values balance, accuracy, source clarity. Strengthen exceptions, caveats, evidence links
- •Gemini — Heavily influenced by SEO signals. Maintain structured data + rankings + freshness
- •Perplexity — Update logs (dates, versions) and reference links are key
- Write 1-page 'why they win' summary per competitor
- Finalize reference hub template we need to create
6 Things to Check
- •Content format — guide, comparison, checklist, FAQ, or tutorial
- •Document structure — answer-first in first paragraph, includes tables/lists/FAQs
- •Evidence structure — public institution citations, numbers, definitions, exception notices
- •Freshness — presence of update dates and change history
- •Internal links — hub-cluster structure
- •Structured data — FAQPage/Article schema applied
News outlets capture AI sources not because they're 'news,' but because information is accumulated in the most organized form. If we build that form, we can reverse the situation.
- Create source graph target list
- Prioritize '10 channels to connect per reference hub'
- Fix 1 reference URL in [References] for URL-possible channels
- Standardize and repeat exact page titles for URL-difficult channels
- 40%+ — Defense mode (maintain freshness and trust structure)
- 20-40% — Pursuit mode (overtake with depth and accuracy)
- 10-20% — Focus mode (win only in specific question clusters)
- Below 10% — Preemption mode (build reference hubs on empty topics first)
30/90-Day Execution Backlog Structure
- •30 days: 10 Quick Wins (SEO-strong/AI-weak areas + citation-inducing procedure questions)
- •90 days: Expand 5 reference hubs + systematize source graph
Common questions
Data collection takes one day; including strategy building, it takes 5-7 days.
Check top domains in GEO Insights → Visibility Leaderboard (last 28 days) to immediately spot unexpected source competitors.
Quick Wins (adding FAQs, applying schema) show changes within 2-4 weeks. Reference hub construction takes 1-3 months.
