Perplexity: Show your sources, earn the next question.
Perplexity answers questions by searching the web and attaches clickable citations to every answer. Explore how showing where an answer came from changes what people trust, check and ask next.
See the idea in one picture ↓- web search
- Real time
- comes with citations
- Every answer
- narrows the sources
- Focus
Figures as reported by the company.
Independent educational analysis. Scenarios, people, numbers and illustrations on this page are original PMcademy practice material, not company research results or an endorsement.
Some people check, and everyone wants the option.
“I need an answer I can actually cite.”
Read the analysis
Our fictional learner, Priya, is a student writing an essay due tomorrow. She wants a quick answer, and she needs something she can cite. A bare answer leaves her to search again to back it up. An answer with numbered sources lets her read the original, judge it and quote it. Ask what your users need to do with an answer after they get it, not only how they feel when they read it.
A confident answer is not the same as a true one.
Trust the answer, or search again
Check any claim in one click
Read the analysis
Imagine your product gives people answers instead of a list of links. That saves them time, and it also asks them to trust you. If an answer is wrong, there is no page to compare it with and no author to judge. The product question is not only how good the answers are. It is how a person can tell when to believe one, and what they can do when they are unsure.
How an answer shows its work
- 1Search first
The question is looked up across several sources in real time.
- 2Cite every claim
Each part of the answer carries a number that opens its source.
- 3Follow up and focus
The next question builds on this one, and Focus narrows the sources.
Animated illustration. Names, numbers and people are fictional teaching material.
Tie each claim to where it came from.
Illustrative data and simplified behaviour. This model does not represent a live company system or measured company results.
Read the analysis
Use the model below to compare three set-ups. In the first, the product gives an answer only. In the second, each claim carries a numbered citation that opens the source. In the third, citations sit alongside follow-up questions that build on the conversation. Notice that citations change the job from trusting the product to checking the source. The numbers in the model are illustrative.
Measure checking, not only clicking.
Perplexity describes real-time answers with clickable citations, follow-ups and Focus.
How often people actually open the sources.
- Citations opened per answer
- Return after checking
- Answers reported or re-asked
Read the analysis
Perplexity describes clickable citations, follow-up questions and a Focus setting that narrows sources. That is the company’s own account of what the product offers, not data on how people use it. In your project, look for behaviour that shows trust being earned: how often people open a source, whether they come back after checking, and which answers they correct, report or ask about again.
Citations show your work, including the weak parts.
- Checkable answers
- Easy to cite
- Trust earned, not assumed
- Weak sources exposed
- False confidence from a number
- Clutter on small screens
Read the analysis
Showing sources makes an answer checkable, and it also exposes when a source is thin, old or only loosely related to the claim. A citation can lend false confidence if people see the number and never click. More sources can also mean more clutter on a small screen. Decide how many sources to show, how to signal quality and what happens when the sources disagree.
How should your answers show their sources?
Your fictional team is building an assistant that answers employees’ questions about company policy. You can show answers alone, attach a link to the policy document, or quote the exact passage next to each claim. Choose a default, decide what happens when no source supports an answer, and name the behaviour that would tell you people trust it for the right reasons.
Choose a next move
Leave with a trust trail.
Your artifact
Write a trust trail for one answer your product gives: its source, the one-step check, what the product says when unsure, and the likely follow-up.
Critique my project artifact using only the evidence I provide. Write a trust trail for one answer your product gives: its source, the one-step check, what the product says when unsure, and the likely follow-up. Separate observed facts, assumptions, proposed tests and missing evidence. Identify a failure mode and a cheap check. Do not invent company results or user research. Ask for my artifact before starting.
Read the analysis
Pick one answer, recommendation or number your product gives people. Write where it comes from, how a user could check it in one step, and what the product says when it is unsure. Add the follow-up a user is most likely to ask next, and one signal that shows people checking and still trusting the result.