Growth Hacking Through Leading Indicators
Find the early metric that predicts long-term success and optimize it
Growth Hacking Through Leading Indicators is a systematic approach to driving startup growth by identifying an early metric in a user's lifecycle that predicts long-term engagement or revenue, then optimizing that metric through rapid experimentation. The process involves finding a correlation between an early user behavior and a critical business goal, building predictions based on that correlation, and then modifying the user experience to improve the early metric.
The most famous examples come from the largest tech companies. Facebook discovered that users who reached seven friends within 10 days of creating an account became long-term engaged users. Twitter found that users who follow a minimum number of people, and have some follow back, are likely to become core users. Dropbox found that putting at least one file in one folder on one device significantly predicted engagement. LinkedIn tracks connections established within a certain number of days.
The key insight is that viral coefficient alone is often too high-level to optimize. Instead, you need to instrument the specific actions within your product that drive virality and engagement. Viral cycle time may matter more than viral coefficient: halving the time it takes users to invite others can produce exponential differences in growth. The three types of virality, inherent, artificial, and word-of-mouth, should be measured and optimized separately because they bring in users with different engagement profiles.
- The leading indicator should come early in the user's lifecycle so you have data points for every user, not just those who stuck around
- Good leading indicators relate to social engagement, content creation, or return frequency
- The indicator must be clearly tied to a part of the business model like user retention, daily traffic, or revenue
- Viral cycle time is often more important than viral coefficient; halving cycle time can mean the difference between thousands and millions of users
- Treat inherent, artificial, and word-of-mouth virality as distinct channels with different conversion rates and engagement levels
- Segment engaged users from disengaged onesUsing cohort analysis, identify the group of users who became long-term engaged or paying customers and compare them to those who churned. Look for behavioral patterns in the first days or weeks of their experience that differentiate the two groups.Pro tipLook at first-day or first-week behaviors specifically. The earlier in the lifecycle you can find the indicator, the more data points you will have and the faster you can act.
- Identify the leading indicatorFind the specific early behavior that correlates most strongly with long-term success. This might be number of connections made, content created, features used, or actions taken within a specific time window. The correlation does not need to be causal yet; correlation is enough to start predicting.Pro tipFacebook found seven friends in 10 days. Twitter found following a minimum number of people with some following back. Zynga found first-day return. Dropbox found one file in one folder on one device. Your indicator will be product-specific but follow similar patterns.
- Build predictions based on the indicatorUse the leading indicator to predict future outcomes. If you know that users who add five connections in their first week have a 70 percent chance of still being active in 90 days, you can forecast your future active user base based on today's onboarding metrics.
- Modify the user experience to improve the indicatorRedesign your onboarding flow, feature prompts, and user experience to naturally guide new users toward hitting the leading indicator threshold. If adding friends predicts engagement, surface friend suggestions prominently during onboarding. Then measure whether improving the early metric actually improves the downstream goal.Pro tipBe careful to verify causation. It is possible that the leading indicator and the outcome are both caused by a third factor. Run controlled experiments to confirm that improving the early metric actually improves the downstream goal.WarningDo not mistake correlation for causation without testing. Users who add seven friends might be inherently more social, and forcing less social users to add friends may not produce the same long-term engagement.
- Instrument and optimize the viral loopMap out every step in your viral loop: invitation sent, invitation received, invitation opened, signup started, signup completed, first use, invitation sent by new user. Measure conversion at each step and identify where the loop is collapsing. Optimize the weakest link first.Pro tipFocus on increasing acceptance rate, extending customer lifetime for more invitations, shortening cycle time, and convincing users to invite more people. Even small improvements compound dramatically.
Timehop had proven stickiness with 40 to 50 percent email open rates sustained over nearly two years. When it was time to grow, the founders discovered that mobile users shared 20 times more than email-only users. They shifted their OMTM to percent of daily active users who share something, with a target of 20 to 30 percent.
Former Facebook growth team leader Chamath Palihapitiya revealed that a user who reached seven friends within 10 days of account creation would become a long-term engaged user. This leading indicator became the optimization target for the entire growth team.
This framework was synthesized from stories shared at growth hacking conferences and from interviews with veterans of Facebook, Twitter, Zynga, Dropbox, and LinkedIn. Richard Price of Academia.edu documented several of these leading indicators at a Growth Hacking conference where former growth team leaders from these companies revealed their key metrics.
The concept of the viral coefficient itself dates back to 1997 when Draper Fisher Jurvetson coined the term viral marketing after observing Hotmail's growth. But the sophisticated approach of finding early leading indicators and optimizing them came from the growth teams at companies like Facebook and Twitter who had enough data to discover the correlations between early user behavior and long-term engagement.