How to Make Better Decisions with Data
Companies often use A/B testing to optimize their websites, but they rarely use it for anything else. This is a wasted opportunity. It turns out that if you capture enough data, any repeated, measurable activity can be framed as an A/B test. For example, how does breakfast affect your morning work productivity? If you spend a month or two tracking your productivity along with the breakfast meals that you eat, you’ll quickly learn if skipping breakfast turns you into a zombie or if having eggs instead of pancakes will help you get a promotion.
Once you internalize the idea that anything you repeat can be an A/B test, you starting seeing optimization opportunities everywhere.
Using Data to Optimize Sales
For example, if you’re at a B2B company and have at least a few dozen customers, you can start optimizing your sales process in many directions. A lot of companies just pursue “more” sales, but not all sales are created equal. By collecting data on each customer – data like the customer’s industry or employee count or annual revenue – you gain the ability to decide what kind of prospects to go after:
Want bigger accounts? Figure out what your biggest customers have in common, then look for similar prospects.
Want good case studies? Figure out what your most engaged customers have in common, then look for similar prospects.
Want quicker sales? Figure out what your quickest sales have in common, then look for similar prospects.
Instead of selling blindly, you might discover that your most engaged customers are from a specific industry, or that your quickest sales happen in mid-sized companies that don’t yet have a procurement team. These insights give you powerful levers for growing your business, and the icing on the cake is that you can often fill out data retroactively instead of only collecting it for future sales.
Using Data in Other Areas
A few other areas where startups can use data for better decision-making:
Recruiting. Keep track of where hires came from, who interviewed them, and how they perform in their first 3 or 6 or 12 months. Over time, you’ll have insights on which channels produce the best hires, which interviewers are too strict or too alienating, and so on.
Marketing. Along with measuring the cost per prospect for each marketing channel, also measure how much those prospects end up being worth, how quickly they sign up, whether they are prone to churn, and so on. The results will help you create targeting that’s more aligned with your goals.
Pitching investors. Track how you meet investors (demo day vs. intro vs. cold email), what kinds of investors they are (angel, corporate VC, traditional VC, etc.), and how well your meetings go. After a few weeks of pitches, you’ll have a good sense of which investors to pursue and how to pursue them.
Using Data Everywhere
Here’s a generalized algorithm for using data to improve decision-making:
Pick an activity you do regularly. The results of the activity should be measurable.
Make a list of attributes that you think might contribute to doing the activity well (or poorly). No attribute is too crazy or too trivial.
Every time you do the activity, write down the value of each attribute along with how things turned out.
After you have a reasonable number of data points, go back and correlate different attributes with outcomes. Some attributes will turn out to predict success, others will predict failure, and the rest won’t have any predictive value.
Incorporate what you learned into how you do the activity. Go back to step #1, if desired.
Activity: selling b2b software.
Details that might contribute to success or failure: experience level of salesperson, how long salesperson has worked for you, prospect’s title, prospect’s industry, length of initial meeting, meeting venue (e.g. their office vs. your office vs. video call vs. phone call), etc.
(In this case, ??? = “Track the details in step #2 for 30 or 50 or 100 sales meetings, then figure out which details lead to more sales.”)
Two things to keep in mind as you optimize your decisions with data:
When deciding which attributes to track, the more independent attributes you have, the better your insights will be. If you’re tracking the age of a sales prospect and whether they have gray hair and their level of experience and whether they have grown-up children, you’re really just tracking the same thing in four different ways. Instead, monitor independent attributes, like experience level and industry and how the prospect heard about your company.
Be careful not to overgeneralize from small samples. If you track 50 sales and the biggest one was introduced to you by your friend Alex, that doesn’t mean you should try to get all of your future sales leads through Alex. One example is an anecdote, not a pattern.
If you’re doing X repeatedly, whether X is sales or hiring or programming or fundraising, chances are you can be doing it even better. Write down pertinent data and facts each time you do X, then revisit that data once in a while to understand how you can improve your process. Tracking and analyzing data over time will help you do more of what works and less of what doesn’t.
If you liked this article, put it to use! Pick one area of your personal or professional life that you want to improve, list out things that might contribute to that area, and start tracking data until you start seeing interesting insights. I think you’ll be pleasantly surprised.