Collection No 6
An imaginative look at principles for testing physical spaces in the future workplace.
It is early January 2018, and you have just returned from the winter holidays. Are you satisfied with your team’s performance and optimistic about the year ahead of you?
If not, follow these three simple steps to track and improve your team’s results using workplace optimization split testing:
1. Choose the right “workplace analytics” solution.
Market-leading workplace analytics solutions use sensors to measure social engagement, power dynamics, and group cohesion. Some also integrate workplace analytics with digital communication analytics, which track digital and verbal communication, subconscious facial expressions, and keystroke patterns to create a holistic data set for your analysis. Top-tier solutions offer a wider set of measurable factors that allow you to test how personal diet, clothing style, and hygiene habits influence group collaboration.
When choosing a vendor, consider employee privacy. Privacy concerns are the biggest barrier to the adoption of workplace analytics, as people do not always feel comfortable having their choice of breakfast cereal cross-referenced with their group engagement metrics. The most common vendor strategy to overcome this resistance is the “opt-in” participation model, which allows employees to choose if they want their data to be part of the split tests. Because it is now easy to profile employees in a very personal way—for example, using computer vision that analyzes motion patterns or restroom visit frequency—it is important that all employees understand that “surveillance” is optional.
You will also want to consider the type of sensors your potential vendor uses and what level of real-time feedback they provide. The latest solutions offer connected contact lenses that measure stress levels and emotional states. These lenses also display feedback in the field of view of the employee, which allows the wearer to adjust their collaboration strategies in real time. Overall, you want a sensor set that is comprehensive, providing value for your employees and a holistic data set for your split tests.
2. Define a clear and relevant test hypothesis.
Once you have chosen your workplace analytics vendor, you should define the correct hypothesis to test. This is where most companies fail.
First, you must define a business objective that is empirically measurable. Most businesses want to boost productivity, but, depending on your industry, this might not be an easy thing to measure. Prior to the advancement of sensing technologies, call centers were one of the few industries that offered a great set of quantifiable productivity metrics, such as the number of issues resolved and the average handling time for each call. Now that companies can measure detailed biometric information, test scenarios are easier to define. But even without the luxury of clear, quantifiable success metrics, companies should look for quantifiable factors that are proven to significantly influence productivity, such as increasing social interactions or minimizing distractions.
With the right business objective, you can brainstorm workplace variations that might impact your hypothesis. Common variations range from team composition to collaboration format to office layout and design. A test hypothesis might sound something like this: “Variations on team composition by personal diet will increase social interactions in the workplace, thereby increasing productivity.”
3. Run tests that will yield meaningful results.
Having the right hypothesis is not enough; you must also choose the right type of test to yield the results you require.
Think of A/B split tests as a way to obtain a go or no-go decision by testing a single larger hypothesis against the status quo. Multivariate split tests are a way to iterate on a larger issue with greater variation, by testing a variety of options and gaining incremental optimization.
Savvy business leaders often start with A/B split tests, as these require a smaller sample size and produce results quickly. For the hypothesis above, an A/B test could measure the level of social interaction within a team whose composition was not determined by personal diet, and then compare the results to the productivity level of a team composed with respect to personal diet. If the variation in the results was statistically significant, further tests would be warranted because personal diet had been shown to have some impact on the level of social interaction. If not, the hypothesis should be modified.
Once you have run A/B tests and validated that a specific scenario is having a significant impact on your business objective, you can begin multivariate tests on the winning variation. These can help you to test more variations at the same time, but they require larger sample sizes and more time to establish statistically significant results. Following the earlier sample hypothesis, at this point you might start testing how grain eaters work alongside three different types of non-grain eaters, or how four different lactose replacements impact the team’s average level of social interaction during coffee time.
With the right tool, a clear hypothesis, and the correct test format, you will improve your team’s performance and inform the decisions you make for your workforce. While your specific workplace analytics solution will define test parameters—such as the optimal sample size and duration for statistically significant results—it is important to remember that large tests generate impactful results quickly. So make sure your employees’ privacy concerns are addressed, ensure that sensors are properly fitted and comfortable to wear or ingest, and then test big and test often.
Siddharta helps organizations design for a better future using collaboration and design tools from start to finish. His areas of expertise and interest include workshops, design research, experience design, and fun times.