How to Use Perplexity Threads for Follow Up Research

Perplexity threads keep conversational context so each follow up can refine scope without repeating the whole brief. That continuity is powerful for research that unfolds in stages: map the landscape, narrow candidates, extract numbers, then verify citations. For a product overview and related workflows, start at /explore/perplexity.

This guide shows how to run threads as a deliberate research workspace rather than a loose chat dump. Pair fast first passes with /blog/how-to-use-perplexity-for-fast-web-search-and-better-answers. For deeper multi source synthesis, use /blog/how-to-use-perplexity-research-for-deep-reports.

When to use threads for follow up research

Use a single thread when the question has phases that share the same audience, constraints, and source priorities. Examples include vendor shortlists, policy scans, product comparisons, and literature maps where later questions depend on earlier names and definitions. Threads shine when you need iterative narrowing rather than one shot answers.

Start a new thread when the topic changes, when confidential scope appears, or when an old thread is polluted with wrong assumptions. Threads hold context. They also hold mistakes. A clean restart often beats correcting a messy history line by line.

How threads work with live web answers

You ask a question. Perplexity retrieves live web pages, drafts an answer, and attaches numbered citations. The thread remembers prior messages so follow ups can refer to vendors already named, tables already drafted, or exclusions already set. Modes such as Search, Research, and Pro Search sit in the mode selector on web, mobile, and the Mac app. Help also mentions Best and Reasoning Search. Treat mode labels as product names that can evolve, and confirm the live selector in your account.

Free plans include Search with citations plus limited Research and Pro Search. Pro is listed at $20 per month on the official pricing page and expands deeper modes plus model selection. Max is listed at $200 per month at a high level. Model lists change. Pro help has listed models such as GPT 5.2, Claude Sonnet 4.6, and Gemini 3.1 Pro among others. Always check the live model list in product. Strong prompts still use instruction, context, input, keywords, and output format.

Step by step thread workflow

1. Open with a durable brief

The first message should state the decision, audience, constraints, source priorities, and desired deliverable. Later follow ups can be short because the brief stays in thread memory.

Thread brief
Decision: Shortlist 5 analytics tools for a 40 person SaaS team
Audience: analytics lead and security reviewer
Constraints: SOC 2, SSO, export CSV, under $40 per seat per month if public pricing exists
Source priority: official docs and pricing pages
Exclude: affiliate roundups
Output: ranked shortlist with citation numbers for every hard claim

2. Run a map pass before deep dives

Ask for names and official URLs first. Mapping prevents the thread from arguing about products that are out of scope.

Follow up 1 (map):
Using the brief, list 10 candidate tools with official homepage URLs only.
No feature essays yet.
Flag any candidate that likely fails SSO or SOC 2 based on public pages.

3. Narrow with explicit keep and drop rules

Tell the thread what to keep and what to drop. Explicit rules reduce drift when context grows long.

Follow up 2 (narrow):
Keep only tools with public pricing pages or clear contact sales notes.
Drop tools without English docs.
Return Markdown table: Tool | Why kept | Main risk | Citation numbers.

4. Extract claims that need verification

Ask for a claim list before you trust numbers. Threads are excellent for producing a verification queue you can work through quickly.

Follow up 3 (claims):
From the current shortlist, extract every price, limit, and security claim.
Format: Claim | Citation number | Confidence (high if official page, low if secondary)
Do not invent missing prices.

5. Close with a handoff note

End the research stage with a note a teammate can reuse. Include open questions and which citations still need a human open.

Follow up 4 (handoff):
Produce a handoff note:
- Decision and constraints
- Top 5 tools with one line each
- Citation numbers that must be opened
- Open questions for sales calls
- What not to trust yet

Copyable thread prompts for common jobs

Use these templates inside an existing thread. Swap bracketed parts. Keep verification lines.

Competitor thread continuation

Continue this thread for competitor research on [product A] vs [product B].
Add only: positioning quotes from official pages, pricing signals, and integration lists.
If sources conflict, list both claims with citations.
Do not reset prior exclusions.

Policy thread narrowing

Narrow the prior policy scan to [region] only.
Prefer government or regulator primary pages.
Output: dated timeline bullets with citation numbers.
Remove commentary that lacks a primary page.

Academic map then deep slice

From the survey map already in this thread, deep dive only on [subtopic].
Prefer peer reviewed reviews from the last 5 years.
For each paper page used: year, one line method, one line finding, citation number.
Do not invent DOIs.

Pricing refresh in an old thread

Refresh pricing claims for the vendors already listed.
Time window: pages updated in 2025 or 2026 when visible.
Replace stale numbers.
Mark contact sales only plans clearly.
Citations required.

Table reshape without new research

Do not retrieve new vendors.
Reshape the existing answer into:
Tool | Plan signal | SSO | Export | Risk | Citations
Keep prior exclusions.
If a cell is unknown, write Unknown rather than guessing.

Conflict resolution pass

Scan prior answers in this thread for conflicting claims.
Create a Conflicts section:
Claim A + citation
Claim B + citation
Suggested resolution rule (prefer official primary page)
Do not hide disagreements.

Mode upgrade inside a thread

Switch intent for the next message:
Use Research style depth for a multi source landscape on the shortlist only.
Keep the same brief and exclusions.
Return sections: summary, vendor notes, risks, open questions.
Cite every factual claim.

Weak follow up versus strong

Weak: Tell me more.
Strong: Expand only security and SSO rows for the top 5 tools.
Prefer official security pages.
Add citation numbers.
Leave pricing rows unchanged unless a factual fix is required.

Thread hygiene reset note

Summarize this thread into a clean restart brief for a NEW thread:
Include: decision, constraints, keep list, drop list, source priorities, open questions.
Exclude: discarded vendors and speculative asides.
Keep under 250 words.

Verification inside long threads

Long threads can feel coherent while resting on one weak secondary page. Open the citations behind any claim you will reuse. For structured output patterns see /blog/how-to-get-structured-and-formatted-answers-from-perplexity. For citation focused fact checking see /blog/how-to-fact-check-information-with-perplexity-citations.

• Re state exclusions every few turns if the thread grows long
• Prefer official docs over roundup blogs when numbers matter
• Separate confirmed facts from still open questions in handoff notes
• Start a new thread after major wrong turns instead of endless corrections
• Keep private data out of prompts even inside a familiar thread
• Re run Search or Research after major product releases instead of trusting a week old chat

Budget time for verification as part of the thread, not as an optional extra. Opening three key citations often takes less time than fixing a wrong shortlist later. When a topic is moving quickly, schedule a fresh run rather than relying on an old conversation.

Keep a running note of which citations you opened and which claims still need a primary page. That habit turns Perplexity from a curiosity tool into a research workflow you can trust for weekly decisions. Prefer official docs, regulator pages, and peer reviewed surveys when those exist for your topic.

If a thread grows noisy, summarize a clean restart brief and open a new conversation with the keep list, drop list, and open questions only. Continuity helps, but clarity matters more than preserving every prior message. Use plain language in prompts and in the notes you keep for yourself.

Common mistakes

• Starting with a vague curiosity prompt and hoping follow ups repair the brief
• Letting wrong early assumptions poison later turns without a restart
• Asking tell me more instead of naming the slice to expand
• Copying prices from summaries without opening citation pages
• Mixing unrelated projects into one mega thread
• Pasting confidential internal strategy into a web grounded tool
• Assuming Free plan depth matches Pro without checking help and pricing pages

Related Perplexity articles: /blog/how-to-use-perplexity-for-fast-web-search-and-better-answers, /blog/how-to-use-perplexity-research-for-deep-reports, and /blog/how-to-get-structured-and-formatted-answers-from-perplexity. For the Explore overview, return to /explore/perplexity.