Articles

Five common mistakes of in-store testing

Featured image

Data drives decisions. Without factual and accurate information, companies base decisions on intuition. Over time, the lack of statistically sound testing practices leads to unhappy customers. Therefore, A/B tests are not only important to planning but also important for planning properly.

Traditionally, in-store tests can take weeks to plan and even longer to execute. With over 30 years of combined in-store testing experience, MarketDial commonly sees these five mistakes that prevent you and your company from properly planning in-store tests:

Lack of leadership support

Within centralized company structures, leadership teams make decisions for the entire organization. Often, buy-in from upper-management is necessary to plan, test and implement new initiatives. This structure makes a culture of testing harder to implement.

In-store testing efforts are not effective unless leadership supports the use of test results as the basis for decision making and is willing to adapt based on the results. No matter what is being tested, the insights will provide additional information to make more confident decisions and lead to successful initiatives across the company, both big and small. Without leadership support, the testing results are rarely implemented to full effect and the benefits are rarely realized.

Unclear hypotheses or objectives

Each in-store test needs a clear objective. A written, detailed hypothesis specifies the current business issue and the results you expect to experience. Hypothesis statements can be based on previous test information or current observations and business trends.

When writing a hypothesis, determine what defines success after the test is complete. The act of identifying a clear hypothesis early on can help an A/B tester to “learn about user behavior or boost your income,” says Brenda Savoie of TechWyse. Once complete, test results provide insight into changes in business performance and confirm whether your hypothesis was correct or whether additional testing is required.

Whether your hypothesis is confirmed to be correct or incorrect, the business will be in a position to make more informed decisions.

Using intuition instead of math for test design

Illustrated cogs interconnect with images of graphs, binary numbers, a clipboard, and a magnifying glass in each cog. The image overlays a blurred picture of a man in a suit pointing towards a graph in one of the cogs. Identifying mistakes with in-store testing can help make cogs with solutions fit together.

After forming the hypothesis, other test factors need to be determined. Common questions during this step include: How many stores do I need? Which stores should I use? How long should the test run?

The answers to these questions should be based on math and not intuition. Without employing the appropriate mathematical rigor in test design, the test will likely contain human bias and data error and prevent reliable decision making. Relying heavily on trends and statistics, with or without testing software, will ensure less internal bias.

Designing proper tests by hand is often time-consuming and requires users to have a deep knowledge of math and statistics. MarketDial can help. MarketDial’s in-store testing software does the math for you, reducing the risk of errors or human bias. In minutes, you’ll know how many stores to test in, which stores to focus on and how long to run the test.

Within the MarketDial software, you can also perform advanced optimizations for confidence level and test/control fit, giving you flexibility without risk of bias. Advanced statistical techniques are employed to ensure the most accurate, predictive tests are created, increasing confidence in decision making and ensuring that the highest value initiatives are prioritized.

Inaccurate and unpredictable tests

To measure business results, tests need to be accurate, discreet, and predictive.

Without a software solution to set up and manage tests, it’s common to find businesses running multiple tests in a single store location. This method doesn’t yield clear results and won’t help businesses make future decisions. Executing multiple tests at the same time, either in one or multiple locations, will ultimately skew results. The data from multiple tests quickly become unpredictable and unreliable.

Running a discreet test and using a robust and predictive control set sample will result in more accurate testing. “That allows an organization to iterate rapidly, fail fast and pivot,” said Ron Kohavi and Stefan Thomke from the Harvard Business Review. Being able to automate this process and run more tests (not at once) will assist in faster adaptation processes for your company.

MarketDial helps you automate the testing processes and the software helps you quickly plan tests and collect data each step of the way. The software manages complex tasks such as selecting the right stores to test in, the right time frame for the test to be executed and the right control sets to optimize the ability to predict results during full-fleet rollout. Each of MarketDial’s software features is designed to make tests more reliable and accurate.

Only testing big initiatives

Sometimes seemingly small tests can lead to the biggest learning opportunities. However, because of the time and effort required to properly set up and execute tests, some companies choose to only test big changes or initiatives. “Most winning tests are going to give small gains—1%, 5%, 8%,” said Peep Laja from Conversion XL. “Sometimes, a 1% lift can result in millions of dollars in revenue.”

In addition to testing small initiatives and saving big, a culture of testing must be created within the company to train employees to test frequently. Testing should be conducted consistently across all areas of the organization in order to see the impact of the results to the business.

Whether big or small, the tests performed need to be comprehensive and accurately set up. As the organization fosters this culture of comprehensive testing, comparability and consistency will improve, allowing for better, more accurate decision making.

Key takeaways

In review, it’s not only important to have leadership support in test creation, performance, and implementation, but also to have clear hypotheses, math-based calculations, and accurate tests. A/B testing has broad applications, all of which can prove successful to your company’s bottom line.

Want to learn more about productive in-store testing techniques? Check out these articles:
What is test and learn?
Enhance decision-making with test and learn software
A test and learn mindset: discover, adapt, and succeed

Ready to start experimenting?

Put us to the test. Let us answer all your innovative questions.