Raman Malik: Inside Perplexity’s Growth Machine: What Worked, What Did Not Work | E1226

Intro (0s)

• Effective marketing involves having others promote a product or bundling it into an existing product, as people typically need to hear about a product three to seven times before trying it (9s).
• Getting a user to perform three queries in their first session on Perplexity is a strong indicator of potential success (18s).
• In growth, there is no middle ground, and being the first or significantly better than others is crucial (27s).
• Raman Malik is the head of growth at Perplexity and is being interviewed about his experiences (45s).
• Raman Malik grew up in Alabama and credits his family for encouraging him to pursue opportunities outside of his hometown (1m5s).
• Before joining Perplexity, Raman Malik was part of an NBA program, which is an unusual background for someone in his current role (1m20s).

From MBA to Startup Founder (1m22s)

• Founding a startup after being in an MBA program can be a viable option, as it allows individuals to tinker with ideas and potentially secure funding from the school, essentially serving as a pre-seed experience (1m23s).
• The MBA experience can be beneficial for those who want to break into a specific industry, such as private equity, becoming a product manager at Amazon, or working at McKinsey, as it provides a preset path and a community to work with (2m14s).
• However, for those interested in startups, the MBA program can be seen as an expensive bridge, and it may be more valuable to dive straight into the startup world (2m35s).
• The decision to leave an MBA program early to pursue a startup idea can be a good one, as it allows individuals to learn more and have a better experience by going full steam ahead on their idea (2m3s).
• The MBA program can have a mixed reputation in the tech industry, with some people having a great experience and others not (1m35s).
• Securing funding from the school and starting conversations with investors can be a natural progression for those in an MBA program who are working on startup ideas (1m50s).
• The MBA experience can be valuable for those who want to break into a specific field, but for others, it may not be the best fit (2m41s).

From Founder to Team Member at Perplexity (2m43s)

• The transition from being a founder to joining a team, specifically Perplexity, is a different kind of difficult journey, as it involves adjusting from running one's own business to being part of a team (2m45s).
• After shutting down a previous startup, conversations with various companies, including Perplexity, led to the realization that Perplexity was a company moving fast and having fun, making it an attractive opportunity to join (3m31s).
• The decision to join Perplexity was influenced by conversations with Arvin, Dennis, Johnny, and Demetri, which highlighted the company's rapid growth and enjoyable work environment (3m32s).
• As a founder, one is accustomed to variety and steering the ship, but joining a team requires a total adjustment in terms of framing and descaling oneself to focus on a specific role (3m53s).
• The transition from founder to team member involves acknowledging the difficulties of startups and the bruising of one's ego, requiring a shift in mindset and focus (3m51s).
• Upon joining Perplexity, the role was focused on driving growth, which required hyper focus and descaling oneself to excel in that specific area (4m11s).

What is a Head of Growth? (4m17s)

• The term "Head of Growth" is ambiguous and can mean different things, but it can be specifically defined as an independent function that bridges product and marketing (4m19s).
• Growth can be divided into two main areas: growth product and growth marketing, with both having the same goal of driving the user funnel from acquisition to monetization (4m34s).
• Growth product is focused on the user funnel, specifically acquisition, activation, retention, and monetization, and involves a product team consisting of engineering, design, and data science working together (4m38s).
• The core product of growth product is the user funnel, and the team obsesses over the user life cycle to facilitate retention and monetization (4m46s).
• Growth marketing, on the other hand, uses different tools such as marketing channels, life cycle communications, community, and marketing campaigns to drive the user funnel (5m16s).
• Growth marketing and growth product teams need to work closely together, with the marketing team thinking about copy, email drip campaigns, and other aspects that complement the product team's work on the onboarding experience (5m40s).
• The ultimate goal is to bring all these elements together as a cohesive unit to drive growth and facilitate the user life cycle (5m52s).

When to Build Out a Growth Independent Function (5m56s)

• The right time to build out a growth independent function is dependent on the type of company, stage, founder makeup, and other factors, but a common answer is post product-market fit, which can be misleading as it costs millions for growth product and growth marketing teams, and requires significant revenue, such as $25 million in annual recurring revenue (AR) (6m28s).
• A more practical approach is to look for early signs of retention, such as 30% retention in month three or four, which indicates that it's worth investing in growth, and this investment doesn't have to be about turning on marketing channels and spending money, but rather about improving retention and conversion rates (6m55s).
• When building out a growth team, it's essential to start by understanding the current state of the company, including acquisition, engagement, and monetization, which often involves turning over stones and asking questions that lead to more questions (7m53s).
• The first few weeks in a growth role typically involve building a map of the world, understanding what's happening in different areas, and identifying gaps in infrastructure and measurement (7m57s).
• Building the right infrastructure is crucial for measuring and understanding the world, and this involves engineering, logging, attribution, and other technical aspects, such as first-party cookies (8m14s).

Is Growth About Micro-Optimizations or Big Swings (8m25s)

• Growth can be achieved through either micro-optimizations, such as increasing retention from 30% to 35% on D90, or big swings, like introducing a new product that could change the game (8m26s).
• Micro-optimizations are often underrated, and building a growth model for a company involves understanding the impact of small changes on new users, retained users, resurrections, and churned users (8m39s).
• A growth model typically includes metrics such as new users, retained users, resurrections, and churned users, which can be used to calculate weekly or monthly actives (8m50s).
• Small changes, such as a 10% increase in retention, can have a significant impact on the overall active user base, making micro-optimizations a valuable strategy (9m10s).
• However, micro-optimizations have diminishing returns, and at some point, it becomes difficult to achieve incremental gains without wasting experimentation time (9m39s).
• To balance this, it's essential to take big swings, such as launching a new feature or marketing campaign, at least once a quarter, which should be high-risk and high-reward (9m54s).
• The success rate for big swings depends on the time invested, but a 25% success rate would be considered great, with the goal of having at least one highly successful campaign or feature per year (10m15s).

The Role of A/B Testing Today (10m35s)

• A/B testing is a valuable tool for understanding the impact of changes on various metrics, determining what is incremental and what is not, and adjusting direction and cadence accordingly (11m0s).
• There are three types of A/B tests: good, bad, and ugly. A good A/B test is well-scoped and provides valuable insights, while a bad A/B test is poorly scoped and yields no useful information. An ugly A/B test is run solely to report a number or impress a manager (10m56s).
• A/B tests are essential for product managers (PMs) to optimize direction, determine if they are working on the right thing, and measure impact. However, PMs should enter an A/B test with a strong confidence in the expected outcome and be aware of potential downsides (12m12s).
• Intuition plays a crucial role in A/B testing, as it helps PMs anticipate potential outcomes and identify areas to watch out for. Without intuition, A/B testing can be like flying blind (12m27s).
• A/B tests with mixed results can be challenging to navigate, requiring PMs to weigh the importance of different metrics and make decisions quickly (12m46s).
• To value different metrics, PMs must consider the long-term implications of their decisions. For example, an increase in activations might come at the cost of query volume, but retained metrics over time might make up for the initial loss (13m19s).
• "Milestone metrics" can be used to identify key events or behaviors that correlate with retention. By analyzing the first few sessions of retained users, PMs can establish a working metric to guide their decisions (14m14s).
• In the case of Perplexity, a milestone metric might be getting a user to perform three queries in their first session, which indicates a high likelihood of retention (14m38s).

Balancing Hitting Growth Metrics with Ensuring Real Value (14m55s)

• The biggest problem of growth is that every goal or metric can be gamed, and to address this, it's essential to monitor whether the metric still correlates with the desired outcome once the initial gaming or tinkering phase is over (15m3s).
• To set the right goals, a rounded approach is necessary, focusing on multiple metrics, such as the number of queries, time in the first session, and whether that first session meets a specific time period, like 10 minutes (15m40s).
• A longer time in the first session can be beneficial, as it may indicate that users are finding value and exploring the platform, leading to a better experience and increased retention (15m45s).
• Perplexity's growth has been primarily organic, driven by word of mouth, with about 80% of acquisition coming from this channel, and the company is focused on nurturing this organic growth (17m23s).
• The remaining 20% of Perplexity's acquisition makeup comes from big partnerships to drive distribution, which may vary depending on priority countries or international markets (17m30s).
• Alex Schulz, one of the original growth experts from Meta, emphasized the importance of ensuring that growth metrics are not being gamed and that they still correlate with the desired outcome (14m59s).
• To achieve effective acquisition, it's crucial to create a magical demo or show a little bit of magic, as this can drive curiosity traffic and quick usage, especially for AI companies (16m52s).

Lessons from Partner Programs (17m35s)

• A WhatsApp group called "Growth" has been created, consisting of the biggest growth leaders in the world, and it would be great to have new members join after the conversation (17m35s).
• The most popular question from the group was about aggressive partner deal strategy, specifically regarding LinkedIn, Exfinity, and Lenny's podcast, where a year of paid membership is given away for free (17m44s).
• Partner programs have been incredibly effective at driving distribution, as it is more effective when someone else is promoting the product or bundling it into an existing product (17m58s).
• The Partnerships team is very skilled and has been successful in creating deals that change the unit economics, but this is seen as an acquisition cost that is worth trading off for a month of Pro membership (18m16s).
• The goal of these partnerships is to capture users and convert them into paying users, as the product's value becomes more apparent with increased usage (18m39s).
• The desired partnerships that are currently missing are those that can help reach new audiences, such as the student audience, to drive distribution at scale (19m1s).
• Specific partnerships in the student world are being targeted to help break into this market and drive distribution among students (19m9s).

Mistakes in Acquisition (19m17s)

• Influencer marketing was initially considered as a key acquisition strategy, but it was underestimated how much effort would be required to work individually with creators on specific content and messaging, leading to a significant time investment and ultimately falling flat (19m30s).
• The strategy has shifted towards working with fewer creators and developing strong relationships with them, rather than trying to manage a large number of creators (20m10s).
• Upon joining Perplexity, it was quickly diagnosed that the main issue was not an acquisition problem, but rather a retention problem, with a focus on improving early activation rates and user retention (20m24s).
• An analysis of the end-to-end funnel revealed that the early activation rate and week one to week two retention were key areas for improvement, with a focus on increasing the number of activated users and weekly actives (20m52s).
• A good activation rate is highly dependent on the company and its specific product, with Perplexity's ability to allow users to start using the product without signing up being a unique advantage (21m15s).
• The logged out visitor to logged in activation rate is a challenging metric to navigate, with a goal of pushing 30% of users to hit milestone metrics and converting them to signed-in users (21m39s).

What Good Retention Is? (21m50s)

• Good retention is when the retention curve stabilizes, with a target of 45% retention by month six, especially for non-social products that don't have a naturally high retention curve (22m9s).
• Achieving 45% retention by month six is considered very good, and it indicates that the product is providing value to its users (22m40s).
• Duolingo is an example of a product with world-class retention, with 50% retention at 12 months, and it has built an incredible engine for retaining users (22m57s).
• When retention is identified as a problem, there are finite ways to improve it, including setting up milestone metrics that correlate with deeper retention earlier in the funnel (23m22s).
• Optimizing channel distribution is one way to improve retention, as organic and referral traffic tends to have higher retention than paid acquisition and partnerships (23m39s).
• Targeting different audiences can also improve retention, as some audience segments may have naturally higher retention rates than others (23m58s).
• Perplexity's horizontal nature means that it can be used by a wide range of people, but some segments, such as students or professionals using the product for work, may have higher retention rates than others, such as trial users or those using the product out of curiosity (24m6s).
• The length of user queries can also influence retention, as longer queries provide more information about what the user is looking for, allowing for better responses and more targeted acquisition and retention efforts (25m16s).

The Biggest Needle-Mover (25m44s)

• The single biggest needle mover in driving numbers up by 10 to 15% has been a combination of factors, including mixing and shifting the audience and platform to achieve cross-device conversion and retention (25m45s).
• Mixing and shifting the audience involves targeting specific high-quality cohorts, while mixing and shifting the platform involves getting users across multiple devices, which is strong for retention (25m56s).
• Driving product improvements is also crucial for retention, and there is always room for improvement (26m23s).
• One retention strategy that has been experimented with but has not had a big difference is tinkering with the messaging as soon as users land on Perplexity (26m31s).
• The messaging aims to convey that Perplexity has up-to-date data, shows sources, and provides inherent value, and experimentation is ongoing to drive this messaging effectively (26m40s).
• A game was played where the Perplexity website was torn apart, and it was suggested that the messaging could be improved to make it clear what Perplexity is and what value it can drive for users (27m2s).
• The current messaging on the Perplexity website is considered wordy, with a score of 6 out of 10, and there is a balance to be struck between providing information, functionality, and brand orientation (27m55s).
• The ideal messaging is being experimented with, but the goal is to use simple language to convey that Perplexity allows users to ask anything and search like never before, with live search and up-to-date answers (28m46s).
• The focus is on search, and the messaging aims to highlight the key differentiation points of Perplexity, such as its live search and real-time answers (28m55s).

What CAC-to-LTV Metrics Were Crucial (29m27s)

• Initially, the focus was on ensuring good customer acquisition cost (CAC) to lifetime value (LTV) ratios before investing in user acquisition through marketing budgets on platforms like TikTok and Instagram (29m31s).
• The primary goal was to achieve top-of-funnel new user growth and sticky retention, rather than immediate monetization through paid acquisition (30m6s).
• Activation rate and retention rate were the key metrics to focus on, with the idea that figuring out monetization would come later (30m27s).
• The decision to prioritize CAC-to-LTV as a more important metric came later, as the company is now looking to optimize its platform for driving subscriptions (30m35s).
• To optimize for subscriptions, the company is considering adding more gates and usage limits, such as limiting the number of searches or file uploads for free users (30m46s).
• CACs and LTVs are variable and can change significantly over time, requiring a lot of data to understand true LTV (31m15s).
• When advising founders, the importance of CAC-to-LTV in the metric stack depends on the company's stage, with a focus on retention and activation in the early stages and a shift to CAC-to-LTV optimization when scaling (31m33s).
• Founders should focus on maximizing the efficiency of channels and CAC-to-LTV ratios when they reach a point of diminishing returns on retention and activation efforts (32m3s).

Why Perplexity Does Adverts? (32m13s)

• Perplexity's growth is driven by a strong acquisition flywheel, making paid advertising seem unnecessary, but the company still uses paid acquisition for specific purposes (32m15s).
• Paid acquisition can be a "drug" that makes everyone feel good due to increased top-of-funnel numbers, but it often underperforms on retention and may not be incremental (32m30s).
• The head of growth at Lyft once turned off all paid acquisition channels for new passengers, and installs and signups barely moved, indicating that the marketing dollars were cannibalizing organic traffic (33m2s).
• Paid acquisition can be useful for marketplaces to balance supply and demand, and in cases where unit economics support it, to target under-indexed audiences (34m31s).
• Perplexity has run lightweight paid acquisition tests on TikTok, targeting 18-24-year-olds with specific value propositions, but finds the platform to be a "crazy beast" due to its algorithm being built for virality (34m51s).
• The company is learning that AI companies need to position themselves differently with new audiences who don't care about AI, and instead focus on the value proposition of getting answers immediately backed by sources (35m34s).
• AI has become almost a negative in product marketing, and having the "AI sticker" can lose value at some points, although it worked well early on (36m2s).
• Perplexity's tests on TikTok have shown solid top-of-funnel campaigns, but the company is still waiting to see how retention will perform and is spending a minimal amount to test and learn (36m30s).

About Brand Marketing (37m9s)

• The concept of needing someone to hear about a product three to seven times before giving it an honest trial is believed to be true, and it's incredibly hard to measure the value of brand marketing, but that doesn't mean it's useless (37m17s).
• Perplexity doesn't have a traditional marketing department, instead, team members with passionate ideas that have viral potential are encouraged to execute and make them happen (37m51s).
• The metric that matters in brand marketing is the incremental impressions gained from a campaign, beyond the expected number of impressions, especially if the moment can go viral and get reposted on social media (38m18s).
• A successful example of brand marketing is a recruitment platform that took out a billboard outside Goldman Sachs' headquarters in London, which got written about in every newspaper and had enormous incremental impressions, especially when hitting net new audiences (39m1s).
• Perplexity ran a TV ad with Jim Harbaugh, which got retweeted and commented on by influencers and reporters in sports, introducing the brand to a new audience of sports fans who may not have been reached otherwise (39m35s).
• The goal of brand marketing is to create moments that can go viral and get reposted, especially when targeting net new audiences, which can be very effective (39m49s).

The Biggest Challenge with Paid Acquisition at Perplexity (39m54s)

• There are many channels to test for paid acquisition, but it's essential to focus on the ones that provide the most learnings, rather than spreading efforts too thin across multiple channels (39m55s).
• Newsletter sponsorships can be challenging unless a product has a clear and simple value proposition that can be communicated effectively in a short ad placement (40m10s).
• Paid acquisition is a game of intuition, and the goal is to keep learning and figuring out the right direction, then working on cadence to drive growth in that direction (40m43s).
• In growth marketing, there's no middle ground - either be the first to a growth channel or be significantly better than existing players (41m7s).
• Many successful companies, such as Dropbox, Groupon, and Pinterest, were first to a particular growth channel, which contributed to their success (41m16s).
• Currently, many channels, including Facebook and Instagram, are crowded and expensive, making it harder to achieve growth (41m42s).
• Community-driven growth, where power users share the value of a product with their own networks, can be a powerful strategy (42m11s).
• One potential approach is to empower power users with resources, such as marketing budgets and swag, to drive growth on their own campuses or networks (42m40s).
• Building density in a specific community or campus can lead to word-of-mouth growth and increased adoption (43m9s).
• The biggest competitor in the search market is likely to be Google, but competing with a big player can also present opportunities for growth (43m21s).

What Channel Could be Better (43m31s)

• The goal is to find ways to improve and be better, particularly in terms of channel growth, with a focus on organic growth and partnerships. (43m31s)
• To achieve this, it's essential to build relationships with creators and tastemakers and work with them in more creative ways, rather than just throwing sponsorship placements on their newsletters or podcasts. (43m44s)
• A key principle is that content must either educate or entertain, and if it doesn't do either, it has no place in the strategy. (44m11s)
• Many social posts, such as simply announcing a new product, are ineffective and don't educate or entertain, so it's better to try and entertain, even if it doesn't work. (44m18s)
• One growth tactic that didn't work well was newsletter sponsorships, which are still regretted. (44m44s)
• An area for improvement is making sharing more inherent to the Perplexity user journey, as there is currently no inherent viral loop, but users are still sharing content, such as answers, on platforms like Twitter. (44m54s)
• The goal is to drive sharing and find creative ways to promote the sharing of knowledge curated by users, which could lead to more traffic and growth. (45m24s)

Hiring Growth Team (45m34s)

• The right first hire in a growth team depends on the company's needs, such as whether they require a marketer to focus on the top of the funnel or a product person to run marketing campaigns, with the ideal candidate being someone who can balance both roles (45m39s).
• The makeup of the current team and the founders also plays a crucial role in determining the right hire, with product people being a good fit for companies with complex, cross-device funnels (46m6s).
• Founders can make good hires for growth teams due to their comfort with taking big swings and being willing to fail, which is essential for driving growth and innovation (46m34s).
• However, as the company scales, founders may struggle with the need for more coordination and structure, and some may even leave to join another startup where they can maintain their "jazz band" mentality (47m18s).
• To manage founders effectively, it's essential to give them the freedom to experiment and try new things, without imposing too many guardrails or restrictions (47m39s).
• Hiring for "taste" involves looking for individuals who can understand what good looks like and reverse-engineer it, with the ability to ask the right questions and think deeply about challenges being key indicators of this skill (48m24s).
• To assess a candidate's taste, it's helpful to present them with different products or challenges and ask them to analyze and provide feedback, with the goal of understanding their thought process and ability to identify areas for improvement (48m47s).
• Mike Maples' analogy of a jazz band is relevant when it comes to early-stage teams, where the ability to improvise and adapt is crucial, and founders are often well-suited to this type of environment (47m1s).

Interview Questions (49m20s)

• In interviews, some common questions asked include "trade war stories" to learn about a candidate's past experiences, especially during difficult times, and to understand how they handled failures (49m24s).
• Another question is about the candidate's first entrepreneurial endeavor, as research shows a high correlation between early entrepreneurial activity and later success, indicating a personality trait that can be evaluated (49m44s).
• Take-home assignments are given to candidates in growth roles, with the goal of assessing their thinking and approach to open-ended scenarios, while being mindful of the scope and not making it too demanding (50m1s).
• An example of a take-home assignment is providing two or three hypothetical scenarios related to the company, such as launching a new product, and asking the candidate to walk through their thinking and approach (50m12s).
• The assignments are designed to simulate brainstorming with the team and to gauge how quickly the candidate can ramp up and fit in, as well as their ability to think critically and come up with creative solutions (51m4s).
• The ideal candidate is someone who can come in and quickly build their own onboarding plan, turn over stones, and start working on their area of ownership with minimal guidance (51m35s).
• A high level of trust is given to new hires, with the expectation that they will be able to figure out what needs to be worked on and have the motivation to drive results (51m47s).
• The best employees often start before their official first day, coming in prepared and already having done research on the company and its dashboards (52m6s).

Hiring Mistakes (52m30s)

• The biggest hiring mistakes often involve hiring too quickly and not waiting for the perfect person, which can lead to poor team dynamics and ultimately, employees quitting the team rather than the company (52m30s).
• The goal is to create a team environment where everyone has mutual respect, loves working with each other, and has too much fun to quit, making it essential to find someone who fits in and will be a great teammate (52m42s).
• Overthinking the hiring process can be a problem, and it's crucial to find a balance between being slow to hire and taking risks with new hires to avoid missing out on great candidates (53m21s).
• The trade-offs of hiring too quickly versus waiting for the perfect person are a common dilemma, and it's essential to weigh the pros and cons of each approach (53m41s).
• The importance of team dynamics and finding the right person for the team cannot be overstated, as it can make all the difference in creating a positive and productive work environment (52m51s).

Quick-Fire Round (53m49s)

• The most common and expensive irreversible mistake founders make when it comes to growth is assuming users' pain points without truly understanding their needs, which can lead to pursuing the wrong growth strategy (53m57s).
• The most polluted growth channel today is likely Instagram or Facebook ads, especially with the latest Apple changes, making it challenging to efficiently work at scale as a core growth channel (54m23s).
• For someone starting a new growth role, the advice is to obsess over every single number in the dashboard, get hands-on with as much data as possible, and learn how the growth machine works by understanding how numbers correlate with each other (54m41s).
• Knowing the numbers better than everyone else can help build intuition about the growth machine, identify friction points, and make better decisions about what to work on (55m15s).
• The single biggest lesson from Lyft is that retention and iterating on the product are crucial, rather than relying on unpaid ads (55m30s).
• Lyft overestimated how much better they could make the ride-sharing product, hitting diminishing returns sooner than expected, and the takeaway is to understand where the moment of diminishing returns is (56m16s).
• In AI, it's critical to be honest about the true Total Addressable Market (TAM) and whether growth is still possible through product iteration (56m33s).
• The most desired change in the world of growth is clarity on what growth entails, as different people have different definitions, and a better understanding of growth product and growth marketing would be beneficial (56m43s).
• The change in perspective over the last 12 months is that the application layer can capture and monetize value, and being at this layer can be advantageous (57m16s).
• The application layer is considered an amazing place to be, where a lot of value is going to emerge, and the whole ecosystem is reentering and reshifting their mindset around where value accrues (57m21s).
• Eight or nine months ago, it wasn't clear where value would be created on top of models, but now people are gaining insight into those who are successful at the application layer and making a lot of money (57m45s).
• Raman Malik's biggest strength is his ability to go from a high-level company strategy to incredibly specific details, such as user funnels, in a short amount of time (58m5s).
• Raman Malik's biggest weakness is that Perplexity is his first big swing at a company, and he is learning on the fly, picking up intuition quickly (58m29s).
• A successful growth strategy outside of Perplexity is creating products or features that drive viral sharing with users, such as Spotify Wrapped, which can be adopted by AI startups (59m1s).
• Creating something sharable or unique allows users to share it, making it the fastest way to growth, and Perplexity could potentially create a "Perplexity Wrapped" feature (59m31s).