The Twenty Minute VCRaman Malik: Inside Perplexity’s Growth Machine: What Worked, What Did Not Work | E1226
CHAPTERS
- 0:00 – 1:05
Perplexity growth principles: social proof, repeated exposure, and the ‘3 queries’ aha moment
Raman opens with a set of core beliefs about how Perplexity grows: third-party endorsement beats self-promotion, users need repeated touches before trying, and early in-session engagement predicts retention. He also frames growth as a search for "alpha"—being first to a channel or materially better than everyone else.
- •Partner/bundled distribution outperforms shouting about your own product
- •Users often need 3–7 exposures before giving a product a real try
- •Activation heuristic: getting a user to ~3 queries in the first session is a strong signal
- •Growth teams hunt for ‘alpha’; otherwise you must win by being much better
- 1:05 – 2:43
From Alabama to MBA to founding a startup: using school as a launchpad
Raman recounts his path from Alabama into an MBA program and then into building a startup. He argues MBAs are most valuable as an expensive bridge into a specific track—or as a safe sandbox to tinker, access early funding, and accelerate investor conversations.
- •MBAs are polarizing in tech, but can be powerful if used intentionally
- •Schools can function like ‘pre-pre-seed’ funding sources for experimentation
- •Skipping interviews to tinker full-time can accelerate learning and conviction
- •MBA value is highest when you need a structured on-ramp into an industry
- 2:43 – 4:17
Founder to Perplexity team member: ego reset, focus, and ‘de-scoping’ yourself
After shutting down his startup, Raman joins Perplexity and describes the emotional and practical adjustment of leaving the founder seat. He emphasizes narrowing scope—going from owning everything to obsessing over one mandate: growth.
- •Startup failure is demoralizing and bruises ego—requires reframing
- •Joining a fast-moving team can be energizing after a founder journey
- •Key shift: de-scope from ‘100 things’ to a single growth charter
- •Role clarity enables excellence, but takes adjustment
- 4:17 – 6:05
What a Head of Growth actually is: the bridge between product and marketing
Raman defines growth as an independent function spanning growth product and growth marketing. The ‘product’ being built is the funnel itself—acquisition through monetization—requiring tight coordination across engineering, design, data, and marketing tooling.
- •Growth bridges product + marketing into one funnel-obsessed unit
- •Growth product: engineers/design/data optimize acquisition→activation→retention→monetization
- •Growth marketing: channels, lifecycle comms, community, and campaigns drive the same funnel
- •Onboarding and lifecycle work require cross-functional cohesion
- 6:05 – 8:25
When to build a growth function: retention kindling, not a vague ‘PMF’ moment
They discuss when it makes sense to invest in a dedicated growth team. Raman suggests looking for early retention signals (e.g., meaningful month 3–4 retention) as evidence there’s ‘kindling’ worth fueling—often before expensive scaling attempts.
- •‘Post-PMF’ is too fuzzy; look for concrete retention signals instead
- •Example threshold: seeing meaningful month 3–4 retention indicates a real base
- •Growth investment can mean improving funnel efficiency, not just spending on ads
- •Early work often starts with mapping reality: turning over stones, building measurement
- 8:25 – 10:35
Micro-optimizations vs big swings: raising the whole waterline (and when to swing hard)
Raman argues micro-optimizations are underrated because small retention/activation gains compound across the entire active base. But he also warns of diminishing returns and advocates for at least a couple of high-risk, high-reward bets each quarter.
- •Simple growth models show small retention lifts can compound dramatically
- •Micro-optimizations ‘raise the water level’ of actives across cohorts
- •Diminishing returns eventually make incremental tests inefficient
- •Rule of thumb: take a few big swings per quarter; expect ~25% hit rate on big bets
- 10:35 – 13:55
A/B testing today: the good, the bad, the ugly—and making decisions with mixed metrics
Raman explains how A/B tests can clarify incrementality and guide direction, but also how they can be mis-scoped or run for performative reasons. He highlights the difficulty of mixed outcomes (e.g., improving sign-ups while harming engagement) and the need for intuition and trade-off frameworks.
- •Good A/B tests reveal incrementality and unintended metric harm
- •Bad tests are poorly scoped with low detectable effects—waste time
- •Ugly tests are run to ‘move a number’ for reviews, not to learn
- •Mixed results force trade-off decisions without waiting 60 days for perfect data
- 13:55 – 16:34
Milestone metrics and gaming-proofing: correlates of retention and a balanced scorecard
They dive into identifying early behaviors that correlate with long-term retention and how to avoid gaming a single metric. Raman describes Perplexity’s ‘milestone metrics’ approach and why multiple measures (queries, time-in-session, etc.) must be monitored together.
- •Work backwards from retained users to discover early-session predictors
- •Perplexity milestone: ~3 queries in the first session correlates with retention
- •Any single metric can be gamed; monitor correlation drift over time
- •Use a rounded set of metrics (queries + time-in-session + downstream retention)
- 16:34 – 17:39
Acquisition at Perplexity: organic word-of-mouth dominance and partnership-led distribution
Raman outlines Perplexity’s acquisition mix, emphasizing AI’s current ‘curiosity traffic’ tailwinds and Perplexity’s strength in organic word-of-mouth. Partnerships play a major role in distribution, complementing organic growth with bundled exposure and credible endorsement.
- •AI products benefit from ‘magic demo’ curiosity traffic, but quality varies
- •Perplexity’s acquisition is ~80% word-of-mouth/organic
- •Partnerships are a major distribution lever, especially internationally
- •Key question: how to keep nurturing and sustaining organic ‘magic’
- 17:39 – 19:17
Partner programs: why bundling works, unit economics trade-offs, and student distribution ambitions
They unpack why partner deals can be so effective: third-party endorsement and bundling drive trial more efficiently than direct marketing. Raman explains why offering free Pro can make economic sense, and notes he’s especially interested in partnerships that unlock the student segment at scale.
- •Bundled/partner endorsement drives more trial than direct promotion
- •Trade-offs resemble acquisition cost: free Pro is variable cost unless heavily used
- •Goal shifts to post-trial conversion into paying users
- •Future focus: partnerships that reach students and build awareness/density
- 19:17 – 21:50
Acquisition mistakes and the retention diagnosis: influencers, activation friction, and logged-out complexity
Raman shares lessons from acquisition missteps—especially the operational burden of influencer marketing and the challenge of explaining Perplexity simply. He also describes his early funnel audit at Perplexity and why the biggest leverage was improving activation and early retention, not driving more traffic.
- •Influencer marketing was underestimated in effort and messaging complexity
- •‘AI search engine’ is not a clear value prop; creator output can fall flat
- •Initial assessment: top-of-funnel was strong; biggest opportunity was activation/early retention
- •Logged-out usage is great for trial but complicates logged-out→logged-in activation targets
- 21:50 – 25:45
What ‘good retention’ looks like and how to move it: mix shift by channel, audience, and use case
Raman gives a concrete retention benchmark for consumer products and then outlines the limited levers available to improve it. He emphasizes mix shifting toward higher-retention channels and cohorts, and targeting use cases like work and studying over pure curiosity traffic.
- •Consumer retention benchmark: stabilizing curve; ~45% by month six is ‘really good’
- •Retention levers are finite: milestone-driven activation, channel mix, and audience targeting
- •Organic/referral cohorts generally retain better than paid/partner cohorts
- •High-retention segments: work and students solving knowledge gaps vs ‘weekender’ curiosity users
- 25:45 – 29:23
Biggest needle-movers: audience targeting, cross-device adoption, and product improvements
Raman details the changes that drove meaningful retention/activation gains, highlighting mix shifts to higher-quality cohorts and getting users active across multiple devices. He adds that ongoing product improvements consistently underpin retention gains.
- •Mix shift to higher-quality cohorts via audience targeting
- •Cross-device adoption (web + mobile) is a strong retention amplifier
- •Multi-context usage (weekend + workday) integrates product into life
- •Product improvements remain a persistent retention driver
- 29:23 – 37:09
Monetization and paid: when CAC:LTV matters, why paid is a ‘drug,’ and positioning beyond ‘AI’
They discuss when to prioritize CAC:LTV and how Perplexity is intentionally not optimized for subscriptions yet. Raman explains why paid acquisition often isn’t incremental, shares a Lyft story about cannibalization, and describes small TikTok tests aimed at learning which value props resonate with younger users—without leaning on the ‘AI’ label.
- •Early focus was activation/retention over immediate CAC:LTV payback
- •Subscription optimization could involve gates/limits and contextual upsells
- •Paid acquisition often cannibalizes organic; incrementality is the key question
- •TikTok tests: small spend (~$10k/week) to learn messaging/use cases; lead with value, not ‘AI’
- 37:09 – 39:54
Brand marketing and virality: measuring the unmeasurable via incremental impressions
Raman argues brand works because people need repeated exposure, even if it’s hard to measure directly. He proposes focusing on incremental impressions driven by virality on top of paid reach, and cites Perplexity’s Jim Harbaugh TV spot as an example of reaching new audiences through culturally resonant moments.
- •Brand requires 3–7 exposures; lack of measurement doesn’t mean lack of value
- •Key KPI: incremental impressions beyond the paid-for baseline via virality
- •Perplexity’s marketing is passion-led without a traditional marketing department
- •TV ad example: sports audience discovery and organic amplification via retweets/comments
- 39:54 – 45:37
Channel strategy: focus, avoiding low-learning spend, community-led growth, and creator unlocks
Raman explains the biggest challenge in paid acquisition is staying focused across too many possible channels and avoiding spend that yields few learnings (e.g., newsletter sponsorships). He reiterates ‘be first or be better,’ advocates community-led growth via power users (e.g., student campus reps), and describes how Perplexity wants to work with fewer creators more creatively.
- •Hardest part of paid: too many channels; avoid tests with low learning value
- •Newsletter sponsorships underperformed and didn’t yield clear insights
- •‘Be first or be better’ framing: crowded channels demand differentiation
- •Community strategy: arm power users (swag/budget) to build campus density; creators via deeper relationships, not shallow placements
- 45:37 – 51:15
Hiring and leading growth teams: first hires, founder-operators, taste, and interview process
Raman outlines how to choose the first growth hire based on the company’s needs: top-of-funnel marketing strength vs complex product funnel work. He shares why founders can be great hires (comfort with big swings), how to manage them (‘let them cook’), what ‘taste’ means, and how he evaluates candidates via questions, war stories, and scoped take-homes.
- •First growth hire depends on needs: marketer-experimenter vs product-funnel builder
- •Founder hires excel at big swings and ambiguity; risk is scaling into ‘sheet music’ orgs
- •Managing strong operators: give trust and autonomy; avoid excessive guardrails
- •Hiring for taste: assess depth via questions; use war stories + scoped, Perplexity-related take-homes
- 51:15 – 1:00:23
Hiring mistakes and operating cadence: onboarding signal, slow-to-hire trade-offs, and quick-fire takeaways
Raman shares his rule that needing a detailed onboarding checklist is an early warning sign in a fast-moving environment—he wants people who create their own map by turning over stones. He discusses being slow to hire to protect team quality, then closes with quick-fire views on common growth mistakes, polluted channels, lessons from Lyft, clarity on what growth is, and the power of shareable product moments.
- •Bad-hire signal: needing a rigid onboarding doc vs self-directed discovery
- •He optimizes for team chemistry and mutual respect; trade-off is slower hiring
- •Quick-fire: biggest founder mistake is misunderstanding early user needs; Meta/IG ads are polluted post-Apple changes
- •Growth advice: know the dashboard numbers cold; focus retention over paid; build shareable viral moments (e.g., Spotify Wrapped-style mechanics)
