Deep Research
Discovers a brand's real competitors via web-grounded search, scrapes their social posts, computes a quantitative benchmark, analyses the field's positioning (multimodal), and synthesises a strategy report. Runs during onboarding and can be triggered from the chatbot.
6-stage pipeline
Competitor discovery
Gemini with Google Search grounding finds 3-5 real competitors (direct + indirect). Each gets a one-line rationale.
Handle resolution
Scrape competitor websites for social links (Instagram, TikTok, X, Threads, Facebook, YouTube, LinkedIn). Grounded fallback for missing handles.
Post scraping
Pull recent posts from each competitor's social accounts via ScrapeCreators. Cap: 30 posts per competitor.
Quantitative benchmark
Pure code — engagement stats, format mix, posting cadence per competitor and market aggregate.
Qualitative analysis
Multimodal Gemini call: top competitor captions + thumbnails → how the field positions itself (angles, visual language, tone).
Strategy report
Structured call with 3 parallel variants → summary, competitive landscape, white space, differentiators, threats, recommended angles, platform guidance.
Benchmark metrics
| Metric | How it's computed |
|---|---|
| count | Total posts scraped |
| medianEngagement | Median of (likes + comments + shares) |
| postsPerWeek | Posts / span in weeks |
| formatMix | % image / video / text |
| byPlatform | Per-platform count + median engagement |
| topPosts | Top 3 by engagement |
Strategy report
The report includes:
- Summary — where the brand sits vs the market
- Competitive landscape — how competitors position
- White space — openings nobody owns
- Differentiators — the brand's real edges
- Threats — crowded/risky moves
- Recommended angles — concrete content angles for posts
- Platform guidance — invest / maintain / test / skip per platform
How findings persist
After the report, writeResearchToMemory() saves key findings to brand_memory:
| Key | Category | Confidence |
|---|---|---|
| brand_differentiators | insight | 0.8 |
| market_white_space | insight | 0.7 |
| competitive_threats | constraint | 0.7 |
| recommended_content_angles | preference | 0.7 |
Every future editorial plan, content batch, and chatbot conversation sees the competitive positioning without re-running research.
anomalia studio <slug> research — run competitor discovery + benchmark.