“Data-driven” has become a buzzword at startups, on par with “disruption” or “lean.” But truly having a data-driven culture means more than tracking your monthly active users or running the occasional experiment. It means building out a scalable data infrastructure, focusing on experimental design and using data to make decisions.
In my experience as an entrepreneur, I’ve seen firsthand how a healthy use of data can help take a business to the next level. To get ahead in today’s competitive startup landscape, here are seven tips for using data to improve your own business.
1. Start thinking about data from day one
Building a “data-driven culture” is a massive undertaking if you only start tracking metrics a year into your company. One of the most important things we did at DoorDash was setting up a robust data architecture from day one. In those early days, we only had a handful of deliveries per day, and could have easily tracked them in a simple document like Google Spreadsheets. Instead, we set up a data tracking infrastructure before our first delivery and built it to scale.Today, we can look back at all of our data, starting with the very first customer, and even know the delivery time of that very first order.
2. Understand your customers and your product
Oftentimes, startups will build a product for a certain use or a specific type of customer, only to find that users are doing something very different than expected. Data can help you understand who your customers are, how they interact with the core features and what your product team may be missing. You can use this feedback loop to help determine if you truly have product and market fit.
For example, you may find that a certain feature is being used more than anything else. That’s what happened to the Slack team who found that the chat function of their game was more popular than the game itself. Or you may find that a certain type of customer is using the product the most, the way that Justin.TV’s team found that gamers were posting the most videos, leading them to pivot and launch Twitch. You’ll never know where your product could go without understanding how it’s being used.
Related: Disruptive Technology in 2017
3. Be naturally skeptical when it comes to data
You can make data say almost anything depending on how you cut it, how you graph it and the story you tell alongside it. But you need to make sure the data actually makes sense. Before you change your entire operations based on new results, first do some mental calculations. Does the data seem right based on what you know about the business? Is it trending the way you think it should? If not, you need to dig in further. Always approach data as a skeptic, and ask, “What don’t I know?” before jumping to conclusions.
4. Run meaningful experiments
Experimentation can be key to optimizing and tweaking your product (Google famously tested 41 shades of blue to determine the final color for links on its search results page). To get accurate results, you have to run thoughtful and well-designed experiments, not simply make a change and see what happens.
A solid A/B testing framework can help you determine if correlations uncovered by analysis of historical data truly represent causal relationships.
5. Don’t ignore the plumbing
One of the biggest mistakes startups make is not having a strong data infrastructure. This underlying plumbing is not glamorous, is rarely discussed, isn’t going to generate revenue or give you any new business metrics to share with your CFO. If you don’t focus on the underlying infrastructure and put time, energy and thought into collecting clean data, you’ll never get good results.
6. Make data accessible
Data adds little value if it’s sitting in a database, queryable by only a few people. In contrast, opening up data to employees and teaching them how to use it can make the entire company more empowered. By knowing how to pull data, people across your business can truly understand how their efforts are impacting the business, make better decisions and better prioritize their work.
Of course, this is easier said than done. Many people don’t know how to pull data, let alone best leverage it. You may consider offering tutorials to all employees so that they can start analyzing data on their own, or hosting weekly office hours for one-on-one tutoring.
7. Ignore your data
Finally, you have to be comfortable ignoring your data. You may have the best in class data infrastructure, testing tools and data analysts, but data can only tell you so much. Particularly, when you’re in your company’s earliest days, you need to have a vision and trust your gut.
That’s not to say that you shouldn’t use data at all. You can use data to inform yourself, check that you’re on the right track, and prove a hypothesis. Data can quickly show you when you’re wrong and help you adjust. At the end of the day, sometimes you have to make a decision with no data, and you need to be comfortable trusting your instincts and taking a stand. Ultimately, data informs decisions but it doesn’t make decisions: you do.