Deep Research Mode
Deep Research is where users go for high-stakes queries. The output reads like a report, citations are extensive, and being absent here matters more than being absent in default chat.
Interesting and early. Worth a spike or exploration session.
AI overview·Citation
What It Is
Deep Research is a long-form synthesis mode that takes minutes (rather than seconds) to produce a multi-page output with extensive citations. Available in ChatGPT (Plus and Pro tiers), Claude (extended thinking with web search), Gemini (Deep Think and Deep Research), and Perplexity (Deep Research mode). Rate limits have been tightening across providers (Perplexity reportedly cut from 600/day to 20/month, ChatGPT Plus to around 10 runs/month in April 2026).
Why It Matters
Deep Research is the surface for high-stakes queries. B2B vendor selection. Considered consumer purchases. Investment research. Medical or legal exploration. The citation profile is different from default chat: more sources, more specific attribution, longer-lived outputs that users save and share. For AEO buyers, being cited in Deep Research has a longer half-life than being cited in default chat, because the report itself often gets shared or referenced later.
Rate limiting is the practical constraint. Users get few runs per month, so each Deep Research query is a high-intent signal. The AEO buyer should care about being present in the source set Deep Research surfaces, not just in default chat answers.
Key Developments
- April 2026: ChatGPT Plus Deep Research limits tightened to ~10 runs/month. Perplexity Deep Research limits cut from 600/day to 20/month.
- 2025: Deep Research mode rolls out as a standard offering across ChatGPT, Gemini, and Perplexity. Claude adds extended thinking with web search.
- Feb 2025: OpenAI launched Deep Research in ChatGPT.
What to Watch
Watch the rate-limit trajectory. As limits tighten, each Deep Research run becomes higher-intent, which raises the value of being in the cited source set. Track whether engines start exposing Deep Research-specific analytics or publisher tooling. Currently there's none, which leaves AEO measurement to manual prompt testing. Watch for Deep Research outputs that include comparison tables across vendors. Those are AEO-critical for B2B and considered-purchase categories.
Strengths
- High-intent audience: Users invoke Deep Research for considered queries. Citation here has a higher conversion signal.
- Long-lived output: Reports get saved, shared, and re-read. Citations in them have a longer half-life than chat citations.
- Rich source attribution: Multi-page outputs cite many sources, opening more AEO surface area than single-answer chat.
- Cross-engine convergence: All major engines now offer a Deep Research equivalent, so AEO playbook generalises.
Considerations
- Low query volume: Rate limits constrain raw volume. Deep Research is a quality surface, not a scale surface.
- Hard to test: Each run takes minutes and costs the user a quota slot. AEO measurement is slow and expensive.
- Hidden retrieval logic: Engines don't always show the full source set they considered, only the ones they cite.
- No publisher tooling: No analytics on Deep Research citations from any major engine.
Deep Research Mode· Query Fan-Out· Google AI Mode· Google AI Overviews· Zero-Click Search