The first viral query was the user’s own name because the result doubled as identity content. People were not sharing an AI search product. They were sharing what the product said about them. The screenshot format reduced explanation cost to nearly zero.
COMPANY PLAYBOOK 07 / 15
THE PERPLEXITY SYSTEM
Perplexity
From irrelevant to an exponential usage curve
Give users a self-referential reason to share, then remove the cognitive work that stalls the next query.
THE QUICK READ
Perplexity’s first growth loop was not “better search.” It was a playful use case that produced screenshot-worthy answers about the user. Its deeper inflection came when the product helped people formulate what to ask next.
ProductAudience
01 / WHAT THE RECEIPTS SAY
02 / BACKLOG-READY PLAYS
2 plays worth stealing.
Open a play for the mechanism, the exact receipt, the failure mode, and a deployment brief Elena can put into your backlog.
20 Make the first viral query about the user
THE MOVE
Early users entered their own social handles, received AI-written profile summaries, and posted screenshots in Discord and on social platforms.
WHY IT COMPOUNDS
Identity content is intrinsically shareable. The output explains the product while flattering, surprising, or provoking the subject.
THE RECEIPT
Srinivas identifies the self-lookup screenshot behavior as the first release’s viral loop and the move that took Perplexity from irrelevance to initial relevance.
STEAL THIS
Find the query that makes the product reflect the user, their company, or their work back to them. Design the output to survive as a screenshot.
DO NOT COPY BLINDLY
A novelty loop earns attention, not retention. It must lead into the recurring core job.
RECEIPT STATUS
Operator-reported. Aravind Srinivas recounts the self-query, screenshot, and early social response.
Open the exact source ↗ ELENA DEPLOYMENT BRIEF
Turn this observation into a real experiment.
- Run it when
- The product can produce a useful, flattering, or surprising result about the user.
- Owner
- Founder and product growth
- First sprint
- Create the self-query, make the result screenshot-legible, seed it with 25 relevant people, and track derivative queries.
- Leading signal
- Result shares, profile-tagged replies, self-query starts, and referred activation.
- Stop rule
- Stop if results are inaccurate, invasive, or generate attention without repeat product use.
21 Design the next question
THE MOVE
Perplexity shipped related questions immediately after New Year 2023, guiding users into a chain of inquiry instead of returning a dead-end answer.
WHY IT COMPOUNDS
The feature removes the user’s weakest step: knowing what to ask. Every answer becomes the start of another session depth event.
THE RECEIPT
Srinivas describes usage as going exponential after the release and calls it the product’s true growth inflection.
STEAL THIS
After every AI output, recommend the two or three highest-value continuations based on the user’s goal instead of generic prompt suggestions.
DO NOT COPY BLINDLY
Suggestions that maximize clicks but not progress create shallow engagement. Optimize for completed inquiry, not query count alone.
RECEIPT STATUS
Operator-reported. Srinivas explains why related questions changed a single answer into an extended session.
Open the exact source ↗ ELENA DEPLOYMENT BRIEF
Turn this observation into a real experiment.
- Run it when
- Users reach a satisfactory answer and leave before discovering adjacent value.
- Owner
- Product growth and search or recommendation lead
- First sprint
- Generate three next questions from live context, measure selection quality, and compare session depth with a holdout.
- Leading signal
- Related-question click rate, useful-answer rate, session depth, and seven-day return.
- Stop rule
- Remove suggestions that increase clicks while reducing trust, answer quality, or task completion.
03 / MAKE IT YOURS
Do not copy Perplexity. Adapt the system to your constraint.
Elena learns your product, customer, funnel, and current bets. Then she chooses the relevant pattern, scopes the first sprint, and watches the leading signal.