Lyft has made a series of great moves over the years in its advertising and branding campaigns. The popular ride-sharing company, which originally capitalized on flubs made by competitor Uber, made headlines more recently for steadily working to move beyond the context of Uber and toward being recognized as important in its own right. It started from shifting promotions from Uber-related reactions to building brand awareness by addressing the values of riders. They also looked into the data of their drivers, to gauge their happiness factor to make necessary adjustments.
These were brilliant moves on their end, but there were a few others that didn’t get the limelight they deserved, and they each revolved around their approach to user acquisition. Interestingly enough, there are many fundamentals to their approach that can easily carry over to how DTC brands can best take on user acquisition campaigns. Let’s dive right into it!
From a traditional standpoint, user acquisition is typically led by a data-driven cross-functional team that focuses on scale, measurability, and predictability. Lyft works with different partners, technologies, and strategies to make sure that Lyft is the top choice for consumers. Over time, it became clear that there were too many cooks in the kitchen, and the best way for them to scale efficiently was by creating a data-driven learning system. The company went on to do so, and named their new internal solution Symphony, “an orchestration system that takes a business objective, predicts future user value, allocates budget, and publishes that budget to drive new users to Lyft.”
Some of the problems Symphony addresses include:
Symphony is dynamic and continuously learning— which is exactly as predictive modeling should be. Of course, their solution is not one size fits all. Lyft adjusts the models, and what is driven from them (budget and bids) according to different channels and different campaign strategies.
One of the main components of Symphony is the LTV forecaster. This forecaster, which improves as the user progresses through the user journey, is based on feature importance of early user funnel data points, and products the likelihood to activate, as well as the lifetime value at different time intervals.
If you know us, our product, and our service. you probably understand why this piqued our interest.
As we dug deeper into Lyft’s thought process into creating their LTV forecaster, we quickly realized that there were some elements that DTC brands can learn from, and apply.
Understanding the potential value of a user is critical for any business, regardless of the industry. When it came to forecasting LTV, Lyft’s goal was to measure the efficiency of various acquisition channels based on the value of the users coming from those channels. Based on the findings, the budget can then be allocated in relation to the expected value for users coming from a given channel, and the price they are willing to pay in a particular region for those types of users.
DTC brands can also similarly allocate their budget towards customers that demonstrate higher LTV, for greater ROI in the long-term.
But how did Lyft calculate a user’s expected LTV? This is another point DTC brands can easily replicate: by looking into historical data.
The more a user interacts with the service, even across different channels and touchpoints, the greater the LTV they demonstrate.
Using AI technology to zero in on customers with greater LTV serves to be one of the greatest forms of churn mitigation for DTC brands. As we mentioned in a previous blog post, churn rates have a way of throwing a wrench into UA efforts, and reducing it not only increases LTV, but ultimately leads to greater return on CAC.
Fortunately, we are living in times in which such advanced capabilities do not solely have to be enjoyed by big brands like Lyft that develop products, such as Symphony, internally. In another one of our previous posts, we talked about how the Facebook conversions API and Google’s Server-Side Tagging allow media buyers the integration that is essential to fire back server side signals, such as LTV and offline conversions, in order to optimize campaigns based on them. Of course, it goes without saying that richer raw data yields more accurate models, which is why UA campaigns backed by AI for user-level LTV predictions are akin to uncovering buried treasure.
For most brands, software development for internal usage costs far too much in terms of resources, time, and of course finances. When factoring the preliminary project stage, application development stage, and post-implementation operating stage—the creation of internal use software isn’t always feasible. They can also be a tough sell to the higher ups. This is where SaaS solutions come in, to help speed up the process for growth marketers and user acquisition managers that are looking to get on the path to exponentially greater ROI.
And let us not forget, when artificial intelligence steps in to assist with loyalty marketing, it enables human counterparts to focus on more meaningful tasks, such as updating creatives, experimenting with new ad formats, and new goals. It’s a win across the board!