How to Use Perplexity's Source Modes to Get Better, More Targeted Answers
By Ashley Gross
·
·
Perplexity

Default search pulls from everywhere, including content marketing blogs you would never trust for an executive decision. Perplexity lets you choose where it looks. Most people never open the menu.
Pick the source before you ask, and a generic AI summary becomes research you can audit.
What You'll Need
Plan level: Free or Perplexity Pro
Prerequisites: A research question that needs more than one kind of evidence, such as published research, real-world opinion, or video
Time required: 10 minutes
Note on feature changes: Perplexity rebuilt this part of the product. The old Focus button and its six modes are gone. Sources now live in the Connectors menu. This walkthrough reflects what is on screen as of October 2026.
Step-by-Step Walkthrough
Step 1: Find Your Source Options
Click the plus button at the bottom left of the search box, then choose Connectors. You'll see two built-in sources, each with a checkbox:
Web (on by default): Searches the entire internet. Best for news, current events, and general questions.
Academic: Searches academic papers. Best for anything that needs published research behind it.
The same menu holds Add files from cloud and Manage Connectors, where you connect your own apps as sources.
Don't confuse sources with search modes. The Search dropdown next to the plus button controls how deep Perplexity goes: Search, Deep research, or Learn step by step. Connectors control where it looks.
Why it matters: You can't control your sources until you know where the controls are. They moved, and most guides still point to the old spot.

Step 2: Pick Your Source Before You Type
Check the source that matches your core question before you write the query. For a research baseline, check Academic and uncheck Web. An Academic tag appears in the search box, so you know the filter is on.
Perplexity may switch a research-heavy question to Deep research and ask a few scoping questions first. Answer them or click Skip.
Why it matters: Choosing the source first keeps low-quality web content out of your baseline.

Step 3: Run One Thread per Source
Decide your sources before you hit enter. Once a thread is running, the Connectors menu in the follow-up box no longer shows the Web and Academic checkboxes. To research across sources, run a separate thread for each one.
Then name each thread so you can find it again. Open the three-dot menu at the top of the thread, choose Rename Session, and give it a clear label like "Academic Baseline" or "Market Sentiment." Your threads live under Sessions in the left sidebar.
Why it matters: Knowing this limit saves you from hunting for a mid-thread toggle that isn't there.

Step 4: Turn Web Off for Pure Synthesis
Uncheck Web when you want Perplexity to work only with what you give it: pasted text, an uploaded file, or a draft that needs formatting.
Why it matters: With Web off, Perplexity doesn't pull outside pages into work that should stay inside your document.

Step 5: Analyze YouTube Videos With Timestamps
There is no YouTube mode to select anymore. You don't need one. Paste a video link straight into the search box and ask for the key claims as a bulleted list, with a timestamp for each claim. Perplexity returns each point with a clickable timestamp.
Why it matters: Timestamps let you check every claim against the video in seconds. That's how you catch a misquote before it reaches your deck.

Step 6: Compare Sources to Spot the Gaps
Ask the same core question two ways, in two threads. Run the first with Academic checked. Run the second on Web and name the community you want to hear from, for example "What do employees say on Reddit about this policy?" Then read the two answers side by side.
The research tells you what should happen. The community tells you what does.
Why it matters: The distance between published findings and lived experience is where implementation problems hide.

Real-World Example: Remote Work Retention Study
A product manager researches how remote work policies affect employee retention.
1. Baseline Phase (Academic Source)
Prompt: "Synthesize peer-reviewed research on hybrid work policies and employee turnover rates in knowledge workers."
Output: A sourced research report. In our run, the strongest causal study showed that two remote days per week cut six-month attrition from 7.2% to 4.8%, a 33% relative reduction.
2. Practical Sentiment Phase (Web Source, Reddit Named)
Prompt: "What are the primary complaints employees share on Reddit regarding mandated return-to-office policies?"
Output: A synthesis of Reddit threads covering commute time and cost, lost flexibility, and distracting offices. Perplexity also flagged that these are recurring themes, not a representative survey.
3. Executive Insight Phase (Video Link)
Prompt: "Summarize the key claims from this video as a bulleted list, with a timestamp for each claim," followed by the link to a conference keynote.
Output: Timestamped claims the product manager can verify and quote.
4. Final Synthesis
The product manager brings all three into one presentation. Academic data shows the retention benefit. Community threads name the friction points. Video timestamps supply executive-ready quotes.
Common Mistakes to Avoid
Leaving Web on for everything: Default search makes you sift through promotional blogs and SEO content to find the evidence.
Using Academic for fast-moving news: Peer-reviewed publishing takes months or years. For anything recent, Academic returns thin or outdated results.
Trusting video summaries without checking timestamps: Skip the check and you risk citing a misheard quote or context the speaker never gave.
Looking for the old Focus modes: Guides that mention Focus, Writing mode, or a Reddit mode describe an older Perplexity.
Take It Further
Add your source-specific threads to a Project so the baseline, the sentiment, and the video notes sit in one place. Open the three-dot menu on any thread and choose Add to project. Then make two habits standard for every project: a timestamp check on every video claim, and a peer-reviewed citation for every statistic.
The people getting sharper answers from Perplexity aren't writing better prompts. They're choosing better sources. Open the Connectors menu before your next research question.
