It's not enough to just have data. You have to know what to do with it.Click To Tweet
- The data-driven approach helps us react to things that have already happened. Tools like Google Analytics give us an historical perspective that allow us to make smart decisions, but it is often a difficult place to test new theories about how to make our marketing better. In other words, we know what to fix, but not how to do it.
- The scientific approach is different in that it begins with a question, rather than the data itself. By starting with a question and developing a hypothesis, we become more proactive with our marketing techniques. Rather than just reacting to negative trends, we are actively creating positive ones–or at least providing a framework for testing them.
The Scientific Method Explained
"The scientific method is a body of techniques for investigating phenomena, acquiring new knowledge, or correcting and integrating previous knowledge." (Wikipedia)In simpler terms, the scientific method is a way to ask and answer specific questions by making observations and conducting experiments. Did you catch the key phrase in there? Experiments. The scientific method is framework for conducting experiments. It goes something like this:
1) Ask A Question The first step requires the formulation of a question based on an observation that you've made. This is where your analytics data comes into play. Observing past behavior is a great way to formulate questions that can receive further testing. What do you want to understand more about?
Example: You observe that only 5% of the people that visit your site sign up for your email newsletter. You ask the question: why aren't more users signing up?
2) Construct A Hypothesis Once you know the question that you are trying to answer, it is important to develop a hypothesis. Based on the knowledge obtained while formulating the question, what is your theory, guess, or speculation of the cause?
Example: Based on the data, you determine that people aren't seeing the email signup box clearly, resulting in a low conversion score. This results in the hypothesis that "a more noticeable signup box will lead to additional conversions."
3) Make A Prediction Many people combine their hypothesis and prediction, but in reality they are distinctly different elements of the method. The prediction specifically determines the logical consequences of the hypothesis, or the proposed proofs of your hypothesis. These will likely become the basis for your future tests.
Example: Based on your hypothesis, you predict that by making the background color orange, you will be able to increase conversations. There can be several predictions made against one hypothesis. For example, another prediction could be that increasing the font size will also support the hypothesis.
4) Conduct An Experiment The next step is to conduct a controlled experiment to prove/disprove each prediction of your hypothesis. This will be done through the use of specific tests derived from your specific needs. Running a controlled experiment is very important here. Each test should be compared to baseline metric, and not necessarily each other.
Example: An A/B style test where you compare each prediction to the current version of the page. Use Google Website Optimizer, Google Analytics, or Optimizely to measure and test your results.
5) Analyze Your Data The most exciting step of the whole process is to analyze the data that you've collected. This involves determining the results of the experiment, and using them to decide on the next actions to take.
Example: Your analysis shows that both of your predictions resulted in more signups. This results could lead you to implement one of them, or conduct another series of test based on a new hypothesis – "By increasing the font size and adjusting the background color we will gain the optimal level of signups."Too often with our marketing techniques, we prefer to make a guess and then pray for results, rather scientifically planning and testing our ideas. This is sloppy marketing, and the scientific method is basis that we need for a practical change.
The History Of The Scientific Method In MarketingThe use of science in advertising and marketing is not necessarily new. It has been around for longer than you realize. In 1923, advertising legend Claude C. Hopkins wrote the seminal book on the subject entitled Scientific Advertising. In the book, he introduced concepts like A/B testing (sampling) and using specific campaigns to test his own hypotheses and how they might affect customer response rates. He was well ahead of his time. Hopkins believed that the best test of an advertising man was to sell by mail because "there one learns that advertising must be done on a scientific basis to have any fair chance of success. And he learns that every wasted dollar adds to the cost of results." To make this point, Hopkins loved to illustrate his work with Pepsodent toothpaste, which according to him, was successful only because of his strategic testing. He completed a series of small tests using a direct mail campaign that he iterated over several weeks. Each week, he would test a new version of the ad and carefully measure the results (sales). Based on the results, and their connection to his hypothesis, he would make a small adjustment in the campaign and begin the test again. After several iterations of this, he finally put a big spend behind the 'perfected' ads resulting in an immediate nation-wide demand for the product. Original Source: theguardian.com Hopkins, however, wasn't the only advertising giant to lean on this method. In the mid 1960s, David Ogilvy, the so-called father of advertising, was famous for saying "nobody should be allowed to have anything to do with advertising until he has read this book (Scientific Advertising) seven times. It changed the course of my life." The book was even referenced on the first season of Mad Men. How To Use The Scientific Method With Your Marketing Techniques Things like marketing techniques, social media, and business often feel far too emotional or ephemeral to measure in a scientific way, but that is not the case. The scientific method provides us with a logical baseline by which we can measure and improve all things. We would be crazy to leave it out of the equation. Here are a few practical ways that you can implement the scientific method into your marketing right now: Improving Email Campaign Open Rates I've often said that there should never be an email that doesn't include at least one A/B test. Most software makes testing two variations super easy.
- Question: How can I improve the open rate of my email campaigns?
- Hypothesis: Adjustments to the subject line of my emails will make a noticeable impact on the open rate.
- Prediction: Using a question rather than a statement in the subject line will result in a better click-through rate.
- Experiment: Begin by A/B testing headlines that ask a question with headlines that don't. Do this over several weeks.
- Analysis: Do questions ALWAYS perform better, or do there seem to be other factors at play?
- Question: How can I get more retweets on Twitter?
- Hypothesis: By adjusting the key words I use in my tweets, I can get more people to retweet my links.
- Prediction: Using words like 'Please RT' or 'Retweet This' will result in a noticeable increase in retweets.
- Experiment: Pre-schedule a series of test tweets and let them run. Change only one variable (or keyword) in each message. The key here is to minimize as many variables as possible by sending each variation on similar dates and times.
- Analysis: Measure the average response rate for each keyword used. Were there some that performed better than others?
- Question: How can I improve the quality of the leads coming from my website?
- Hypothesis: Many of the leads coming from our website are not applicable to our product offering. By adjusting the questions we ask, we believe that we can improve lead quality while reducing the number of of emails we receive each month.
- Prediction: Adding more specific questions related to our industry will improve the overall lead quality.
- Experiment: Add specific questions, one at a time, and measure the number of requests that come in and the overall quality. You will need to devise a system to rate overall message quality.
- Analysis: Track your measurements month to month, based on your system, and see if there is a noticeable difference that the added questions are making.