Taking a scientific approach to your business can set you free
Taking a scientific approach to your business can set you free
Test your assumptions instead of guessing
It’s been a minute. I have this love/hate thing with the responsibility of creating stuff from my head and heart. I crave providing people with some sense of consistency but I can’t even deliver that to myself.
I learned something recently. If you want to do a thing, don’t overcommit. If you want to do pushups every day, just start with 10, not 100. I’m going to try to write one of these things a week. Instead of the pressure to pack it full of substantial goodness, I’m going to just share a nugget of insight with as few words as possible. From there, I hope it grows and clarifies itself to me. It’s also a great reminder that it’s got to start as a thing for me. I’ve lived most of my life in service to others — this can be a relief for my brain with the byproduct being something of use to you.
Daily Snack vol. 1
Testing is freedom. I work in a startup. One of the earliest lessons I’ve learned is the difference between a startup and a big, established company is a culture of rapid testing. For the uninitiated, it’s not that different from science. You want to know what will get a customer to open an email — A/B test the subject line. You want to know what messaging will generate more sales — A/B test the landing page.
Why is testing freedom? The closer I get to certainty the more at ease I feel. We all take some educated guesses but without learnings that you can replicate, eventually you end up tired and still clueless. No fun. Here’s what it looks like in the simplest terms:
Goal. Identify the purpose for the experiment.
Hypothesis. What do I expect to happen if I take a specific action.
Process. What/where/how can I set up the test to measure an outcome.
Measurements. What signals and results will prove/disprove the hypothesis.
Test. Run the test.
Analyze. Interpret the data for insights.
Decide. Take action based on learnings.
Rinse and repeat if needed. Here’s one real life example to drive it home:
Goal: I want to understand whether customers will buy a red shirt or a blue shirt before I place an order with the factory.
Hypothesis: Based on color trends and past selling, I expect the blue shirt to sell better than the red shirt.
Process: We have samples of one color of the shirt. We’ll photograph it and add a pre-order to the website. We’ll color correct the imagery so we showcase both colors. We’ll promote the pre-order and then run the sale for 1 week to gauge demand.
Measurements: The KPI (key performance indicator) is total sales of each color. This is fairly straightforward, but secondarily, we could look at the click rates on email promotions or survey responses to customers who didn’t place a pre-order; etc.
Test: Promote and open up pre-orders for a week.
Analyze: Compare sales of the blue and red shirt to see which color won. We also want to see if there’s enough demand to potentially manufacture both colors for customers.
Decide: Based on overall sales volume and the differential between colors, we’ll decide on which colors to produce. We may decide to repeat this exercise for more product styles and colors. If the time period was too short, we may also decide to run future experiments longer. You can learn a lot from each experiment, even in areas you weren’t initially focused on.
Basic, yet effective. You can test everything — even things that are more complex, you’ll just need to break it down into different pieces. Think like a scientist. Isolate the variable you are wanting to learn more about. You can’t test everything at once. I hope this is a helpful. Let me know if concrete examples would be useful.
See you on the next one.
-Justin (@freelancekills)