Reverse Engineering To Disrupt
By Karen Geier
Reverse engineering is a phrase that often evokes a cloak-and-dagger operation of a person taking another person’s invention apart piece by piece and figuring out the magic behind it. The reality of reverse engineering is that it’s often done by companies to their own products with the objective of finding a better way to achieve their business goals.
The success of the US reboot of House of Cards was based on taking millions of data points that Netflix was collecting and using them as a lens to create a show that reflected the broadest tastes of an attractive demographic. The British version of House of Cards was surprisingly popular, as are movies starring Kevin Spacey. The result was a smash hit that generates thousands of press mentions and word-of-mouth advertising. It had been falsely assumed up to that point that audiences are fickle, and that attempts at predicting their tastes would fail out of the gate.
How to Make Reverse Engineering Work for You
The key to reverse engineering is the same whether you’re breaking down a piece of equipment or redesigning your website’s user experience: it’s learning everything that can be learned about the way the thing works in the first place. Research needs to be your first step.
Start With a Business Goal
You need a lens by which to interpret your data. What does your business want to achieve? Is it a higher conversion rate on your website? Is it more page views per session? Decide on the outcome you want to achieve and plan your data interpretation around it.
Devise a Plan to Capture Data
If your company or product exists online in any form, whether it’s in an e-commerce environment or you simply have a Facebook page or Twitter account, you should grab all the data you can. Take this time to review all the ways your company touches the outside world and think about how you might be able to gather more data.
If your product is on a shelf, can you work with retailers to find out more about when your product is purchased most often or if it tends to be purchased alongside another product? Think about high- and low-tech solutions to capture representative data.
Decide What Metrics Will Help Inform and Define Success
Once you have relevant data, it’s time to interpret it and devise a method of testing the success of your modified product. Don’t forget about predictive metrics – those based on patterns, trends and correlations. They might not indicate 100% success (like a completed sales funnel) but might reveal that if someone puts your product in his or her cart after X and Y, then he or she is more likely to purchase your product. Don’t forget about looking for predictors of failure, too. These will help you focus and more effectively move in the direction of success.
Challenge Your Data-driven Hypotheses with A/B Testing
You can always test the success of your hypotheses. Think about a small, cost-effective method of testing updates to your product, whether it’s through multivariate website testing (determining the better of two or more content variations) or by sending an updated version of your product to high-value customers for feedback. You need a control (no change) and a variable. Resist the temptation to “go big” or test multiple variants. This will lead to inconclusive information.
Test for a Short Period of Time and Optimize
Whatever you’ve decided to test, do it for a short period of time, and optimize your results. If you get a strong indication something worked, try to increase that thing and test again. You are looking for a threshold that the changes, once implemented, are optimal.
Analyze the Results
Once you’ve tested and optimized, you should have a roadmap for where to take your product that is informed by customer feedback and behavior, and is reproducible. You should still do a full, deep analysis for any other indicators your results raise that can help you plan further iterations or help you steer clear of pitfalls.
What If You Get a Negative Result?
Your theory might be proven wrong. There is always a possibility you may have misjudged the market, but if you’ve followed the protocol, you haven’t created the next “Coke 2”; you’ve just created a “seasonal variation” that won’t return.
Negative lessons can also be positive data, because your competitors might not know what you know, and you’ll be steering clear of pitfalls along the way.
Reverse Engineering is a fancy term for a simple concept: customers are verbally and non-verbally voting on your product all the time. All you need to do is to focus on the data, devise low-cost, informative tests, optimize, launch, and repeat. Once you’ve got a seamless method for testing and optimizing, you can engineer better results.