DataMuse | Monetise Your Data
Get practical insights that impact the bottom line.
growth_hacking_datamuse_sydney_4.jpg

The Muse

We blog here about growth and data.

How To Think Like A Scientist And Grow Your Business

Just because you’re not a scientist, doesn’t mean you can’t think like one.

When I hear “scientist” I see someone sitting in a dimly lit laboratory in a white coat and goggles bent over test tubes or complex equations solving really important problems. I don’t know exactly what it is they're doing, but I know it’s important.

Scientists are a big deal. You don’t pick a fight with a scientist. You can’t win an argument with a scientist. Most of us don’t know any scientists, and those of us that do are somewhat in awe of them.

But apart from an air of mystery scientists have one other thing in common: rigorous scientific training. This means that they are trained to approach any problem in a particular, structured way.

We can adopt this approach to grow revenue and optimise the sales funnel. The scientific approach can be incredibly effective, because it replaces “gut feel” and “intuition” and relies on structure, data, and analysis to help us make decisions.

Now that we can track everything that happens on our digital storefront, we can approach challenges in marketing and business like a scientist would.

The scientific approach can transform how successful we are in optimising the sales funnel at every step. Here’s how to do it:

1. Be really specific when defining the problem you're trying to solve.

Often we don’t know how to approach a problem because we can’t whittle it down to its essence. Scientists are really good at breaking large problems into smaller parts and methodically solving each micro-problem. Use the data you have to get really specific. It’s not enough to try and solve for “more profit”, you have to break it down into bits and approach each one individually because there could be a different solution to each one.

Example: “My problem is that I’m not able to convert visitors to subscribers on my homepage.”

2. Come up with a hypothesis.

A hypothesis is an idea. And It’s here that you need all of your gut feel and intuition. A hypothesis is the dart you throw at a dart board, a plausible reason for why you think you have the problem that you do. You can have multiple hypotheses as to why a particular problem is occurring. If that’s the case, write them all down and make your way down the list, one by one.

Example: “I hypothesise that the reason I’m not able to effectively convert visitors to subscribers on my homepage is because my Call to Action (CTA) is too difficult to find. My second hypothesis is that I’m not converting visitors because the colour of my button blends in with the background. My third hypothesis is that the channels that are driving traffic my way are the wrong ones.”

3. Set up an experiment to test your most plausible hypothesis first.

Once you have a hypothesis, you need to test it. Set up a series of tests to see whether or not your hypothesis is true. Make sure you run the experiment for a long enough period of time with a large enough dataset and also, make sure your experiment is sound. That is, that you have a control you are testing against and that you are changing (i.e. testing) only one thing at a time. This allows you to isolate the reason and to be confident in the veracity of the results when you get them.

Example: “I’m going to move my CTA to the top of the page and run this variation for 3 weeks to see if there is a lift in conversions. I'm going to set up an A/B test to do this and run the traffic 50/50 through the control and my test to make sure there are no seasonal differences that accidentally skew the results."

4. Analyse your results.

Since you’re scientific, you’re running the right kind of analytics platform so you’re collecting the right data to be able to determine whether or not your experiment was a success. Also, because you've run a control alongside your variation, you know that it is the change you made that has resulted in the improvement you measured. 

Example: “In the last three weeks, conversions have improved by 30% in the page variation with a more prominent CTA as compared to the control.”

5. Conclude and Implement.

This is a spin on the experimental “Conclusion” you would have written for your high school science class, but the concept is the same. You have results, now let’s implement them.

The only catch is that you may need to use your judgment at this step too. Mostly, data doesn't lie, and you must be prepared to accept the results even if you were sceptical that the experiment would be a success.

But equally, if you get a huge outlier of a result, it may be that there's something wrong with the way your experiment was set up. Use common sense in to distinguish and review if necessary.

Example: “I’m going to make that outperforming variation a permanent feature of my homepage.”

6. Repeat.

You had a second hypothesis, remember? Let’s do this again, this time, changing the colour of your button to something more prominent to see if it affects your conversion rate. Remember to run a control page in parallel, and to run the experiment for a long enough period of time!

But you knew that already.

7. Iterate.

This is a step we’ve added but it’s super important. Running experiments has to be iterative to be successful and that means making small improvements and changing hypotheses on the basis of what you’ve learned.

Example: “Since making my CTA more prominent on my homepage increased my conversions by 30%, I’m going to run a conversion analysis by traffic source and try and serve personalised landing pages before changing my channel spend.”


So there you have it. A scientific approach to growth. At the heart of it, this method gets you closer and closer to your sales goal because every conversion goes straight to your revenue line.

Wait, you’re not converting leads to paying customers as well as you could be?

But that’s the subject of another blog for another time.

If you’d like to know how we can help you do that (we don’t wear lab coats or goggles, but we’re really good at data and growth) let’s start that conversation.