How to scrappily hire for, measure, and unlock growth | Crystal Widjaja, Gojek and Kumu
Crystal Wajaya, CPO at Kumu, shares insights from her experience at Gojek and Kumu, covering growth strategies, analytics best practices, and building effective growth teams. She also discusses her nonprofit, Generation Girl, empowering women in STEM.
Deep Dive Analysis
16 Topic Outline
Crystal's Non-Standard Career Path and Journey
Scale and Impact of Gojek and Kumu in Southeast Asia
Why Super Apps Haven't Emerged in the U.S.
Lessons for U.S. Startups from Asian Growth Strategies
Being Intentionally Scrappy and Using Small Data Sets
Strategies for Increasing Retention
Effective and Ineffective Growth Investments
Framework for Founders Approaching Growth Strategy
Optimizing Funnels and Leveraging Existing Resources for Growth
Biggest Growth Unlocks from Crystal's Experience
Why Most Analytics Efforts Fail and How to Avoid It
Signs an Organization is Getting Analytics Wrong
Recommended Metric Tracking Stacks for Founders
Setting Up and Structuring Growth Teams
Hiring the First Growth Person and Interview Process
Generation Girl: Nonprofit Helping Women in STEM
5 Key Concepts
Wizard of Oz Experience
This refers to simulating a feature or service manually to validate its value proposition and conversion rates without actually building the underlying technology. It allows teams to test user interest and behavior with minimal engineering effort.
User Psych Model
A mental model that emphasizes the need to build momentum in a user's journey to help them overcome friction points, such as typing in credit card details. It focuses on sequencing the right steps effectively to lower friction and facilitate decision-making.
Physics of the Market/Product/Model/Channels
These are the fundamental constraints and elements of a company's operating environment, including the market, product, business model, and distribution channels. Understanding these 'physics' helps identify existing levers and constraints for growth, rather than trying to dramatically change the entire system.
Measurements vs. Insights
Measurements are observed facts or data points (e.g., 'power users do four times more bookings'), while insights provide context and answer 'why,' leading to actionable changes in behavior or strategy. Insights offer understanding that allows for different actions in the real world, unlike mere observations or 'entertainment' data.
Instrumentation
The process of setting up data tracking to capture not just events, but also rich properties and context around those events. Effective instrumentation allows for segmentation of user behavior and the formation of hypotheses, which can then be tested to gain causal understanding and actionable insights.
8 Questions Answered
In Southeast Asia, there's a different cultural sentiment towards conglomerates and a greater need for single apps due to limited phone storage and a leapfrogged computer era, making a single app for many services more practical than multiple specialized apps.
Startups should run experiments even with small sample sizes (e.g., 30 people) because the underlying trends and information gained won't be very different at larger scales, though precision will increase with more data. The goal is to get *some* data rather than none.
For a free product, aim for at least 60% week-over-week cohort retention that flattens after 2-3 weeks; for a paid product, 20-30%. For friends and family using an early product, retention should be near 80%.
Founders should first understand the 'physics' of their market, product, model, and channels, then identify constraints or underutilized levers, and finally optimize specific parts of the funnel or loop by making small, measured changes rather than trying to overhaul everything at once.
Analytics efforts often fail because teams focus on tracking OKRs or gathering 'entertainment' data rather than identifying actionable insights, and they fail to instrument events with rich properties that provide context and answer 'why' behind user behavior.
A key sign is an instrumentation specification with many events but each having few or no properties tracked, indicating a lack of context to understand user behavior and make informed decisions, leading to an inability to answer 'why' questions.
The first growth hire should be strong in statistics, understand how to run numbers, interpret ratios correctly, and be able to design experiments appropriately, focusing on workable opportunities rather than just flashy new features.
Organizations like Generation Girl offer free classes and programs for girls aged 12-17, partner with schools to train teachers on new STEM concepts, and focus on early intervention to counteract cultural biases and lack of early exposure.
19 Actionable Insights
1. Validate Ideas Scrappily
To quickly gather real user data and validate new features or ideas, implement “Wizard of Oz” tests, like manually selling subscriptions via WhatsApp or overlaying design mockups on screenshots, avoiding extensive engineering investment.
2. Experiment with Small Data
Run experiments even with a small sample size (e.g., 30 users) in early stages, as the underlying trends will likely remain consistent, though precision increases with more data. This provides valuable directional information, which is better than no data at all.
3. Optimize Pre-Conversion Steps
Instead of solely focusing on retention, identify and solve for the specific “setup moments” or preceding events that users must complete before reaching their “aha moment” or converting (e.g., building trust). This approach facilitates the ultimate conversion by addressing earlier barriers.
4. Drive Action with Insights
Treat analytics as a tool for generating actionable insights that explain why user behavior occurs, rather than just observing metrics. An insight provides context and directly informs changes to your strategy or product.
5. Contextualize Events with Properties
When tracking data, ensure every event includes rich properties that capture the context and characteristics of the user’s experience (e.g., number of drivers on screen, city, surge pricing). This detailed context is essential for understanding the “why” behind user actions and making informed decisions.
6. Define Specific Retention Goals
Avoid vague retention goals; instead, define concrete, specific actions or behaviors that indicate user engagement (e.g., “everything orange in a construction site” analogy). This specificity makes growth efforts more actionable and measurable.
7. Reduce Conversion Friction
Identify the most painfully long or high-friction steps in your user’s conversion path (e.g., typing in credit card details) and focus on shortening or simplifying these processes. Optimizing these points can significantly improve conversion rates.
8. Set Retention Benchmarks
Aim for at least 60% week-over-week retention for free products and 20-30% for paid products at scale, expecting retention to flatten around these percentages after 2-3 weeks. For friends and family, target closer to 80% retention to validate product-market fit.
9. Map Growth Physics & Levers
Systematically analyze the “physics” of your market, product, model, and channels to understand your current growth environment and available resources. Then, identify and utilize underutilized levers (e.g., drivers as salespeople) to drive growth without needing to drastically alter fundamental dynamics.
10. Implement Pause/Snooze Feature
For subscription products, if cancellation reasons reveal temporary needs (e.g., “I still have too much beer”), introduce a “pause” or “snooze” option. This provides a temporary solution that addresses the user’s immediate problem and prevents permanent churn, which is harder to reacquire.
11. Refine Copy for Onboarding
If users are landing on your app or webpage but not taking a first action, focus on improving your copy to resonate with their pain points rather than just describing your solution. This helps users understand how the product fits into their lives and increases initial conversion.
12. Anchor New Concepts to Familiarity
When introducing unfamiliar product concepts (e.g., a digital wallet), tie them to existing, familiar mental models (e.g., using a credit card image for a virtual account number). This helps users quickly grasp the concept and drives adoption.
13. Use Friction for Deeper Engagement
For products involving complex human emotions or community building, strategically introduce artificial friction (e.g., detailed questionnaires for community access). This helps differentiate deeply committed users from casual ones, allowing for more tailored experiences and better understanding of true intent.
14. Beware Early International Uptake
When expanding internationally, be cautious if you see a rapid initial uptake, as this might be primarily early adopters who are already equipped to use your service. Recognize that acquiring the broader market after this initial surge will likely require significantly more effort.
15. Hire Stats-Savvy Growth Talent
When making your first growth hire, prioritize candidates with strong statistical intuition who understand concepts like sampling and selection bias. This ensures they can correctly interpret data, measure impact, and focus on the most impactful growth opportunities.
16. Assess Growth via Case Studies
In the hiring process for growth roles, use case studies that require candidates to design experiments and demonstrate a deliberate, measured approach. Provide ample time (e.g., 4 hours over 5 days) and encourage them to research and articulate their problem-solving process.
17. Prioritize Quick Growth Hacks
For early-stage growth teams, focus on implementing quick, “hacky” solutions (e.g., a simple Python script for SMS campaigns) that can deliver immediate impact. This approach maximizes the opportunity cost of time and helps unlock future potential rapidly.
18. Start with Free Data Tools
For early-stage companies, begin with free or open-source analytics tools like Google Data Studio (for single data warehouses) or Metabase (for SQL users and multiple databases). This allows for effective data analysis without significant initial investment.
19. Support Women in STEM
Support Generation Girl, a nonprofit co-founded by Crystal Wajaya, which provides free STEM classes to girls aged 12-17 in Indonesia and develops resources for teachers. You can help by donating via their website (generationgirl.org) or offering licensed software.
7 Key Quotes
What's better than having 30 data points? Certainly having 100, but what's better than having zero is definitely 30.
Crystal Widjaja
Real news is information that changes what you do in the real world. And if you don't change what you're doing, what you are doing is just getting entertainment.
Crystal Widjaja
Don't make the same mistake that Netflix and Spotify have made. Which I guess is when they've launched, they've started international expansion and they see this very small percentage of users start to sign up for Spotify or Netflix... when in reality, it's like you just pulled forward everyone who could have possibly subscribed to you. Now you're going to have to work a lot harder to get everyone else.
Crystal Widjaja
The worst possible thing is to have a growth person who thinks they are doing the right thing and is measuring things wrong and then focusing on the wrong areas.
Crystal Widjaja
Everything in growth is an opportunity cost of time trade-off with what you could have been doing to the product in that time.
Crystal Widjaja
Yes, I absolutely love makeup, but I also am badass at writing Swift code. So step aside.
Crystal Widjaja
I am 0% surprised to hear that. I've consistently seen that improving onboarding is one of the highest leverage opportunities for both sign-up conversion and increasing long-term retention.
Lenny Rachitsky
3 Protocols
Gojek Subscription Feature Validation (Wizard of Oz)
Crystal Widjaja- Add 100 randomly selected drivers to a WhatsApp group.
- Instruct drivers to pitch a subscription package to customers during rides, stating they are the only ones who can sell it.
- Have customers give $10 to the driver if they agree to the subscription.
- Drivers text a designated phone number when a customer agrees.
- A team member looks up the customer in the backend and applies the corresponding vouchers.
- Deduct $10 from the driver's balance.
Gojek GoPay Adoption through Driver Incentives
Crystal Widjaja- Build a small service to check if a customer allocated to a driver has previously used the GoPay product or has a digital balance.
- If the customer hasn't used GoPay, message the driver immediately with an incentive.
- Offer the driver extra money if they get the customer to give them cash to deposit into their virtual GoPay wallet.
- The driver, acting as a salesperson, explains the benefits directly to the captive audience customer, facilitating the top-up.
Growth Experiment Design (Hiring Process)
Crystal Widjaja- Provide candidates with case studies of observed data (e.g., 'here's what we see').
- Ask candidates to design an experiment to understand 'why' the observed data is true, focusing on random sampling and a measured, deliberate approach.
- Allow candidates 5 days to complete the task, with an expectation of approximately 4 hours of work.
- Assess their ability to research solutions (e.g., Google white papers) and explain their approach during a follow-up discussion.