Analytics for Change

Sign in

Analyzing change over time is the key to maximize social investments

Large investments are made to promote behavioral change. Companies invest in developing marketing strategies to increase sales by changing individual routines and needs. Governments invest on campaigns to maximize compliance with regulations or promote environmentally friendly choices. Health providers promote healthy lifestyles by facilitating or implementing behavioral changes projects. However, these investments are not always successful or sustainable over time.

Examples of measures of success of these investments can be rates of increase of sales of specific products, rates of increase of compliance with regulations (e.g., filing your tax return in a timely manner), rates of increase of electrical cars per inhabitant or rates of decrease of smokers. All these measures have one shared characteristic: they are measures of change, rates of increase or rates of decrease.

When companies, governments or health providers want to measure the success of their strategies, campaigns, or interventions, they are measuring change.

How do we measure change? The most basic way of measuring change is comparing our measure of interest at two time points; and the most logical way to decide which two time points we want to compare is before and after we have set up our marketing strategy, our engagement campaign, or our public health intervention. This basic design is named pretest–posttest design and it is one of the most basic experimental design. Scientists have been using these designs for more than a century and there have been amazing developments in the last decades. However, many companies, authorities and health providers still rely on the most basic experimental design and miss the great advantages of the new analytical techniques.

Data science allow us to consider more than two time points so we can capture how real change happens and account for the fact that change is not likely to be linear (i.e., there will be periods of increase, decrease or no change); and contextual factors can be also contributing to our probability of change.

· Change does not always follow a linear trend. Basic analytical techniques assume change is linear. This means that the rate of increase or decrease is the same per each unit of time (i.e., days, months). For example, if we want to know whether our advertisement campaign led to an increase in sales in the following month; assuming a linear change means assuming the same increase in the number of sales each day. This is rarely the case. Exponential growth is more likely to occur followed with a plateau stage, especially in digital campaigns.

Photo by Chris Liverani on Unsplash

· Change cannot be always directly attributed to your marketing strategy, engagement campaign or intervention. Several contextual factors can influence your outcomes. Some of these factors can be accounted since the start of your plans and some cannot, some can constitute a key transition point and some can be main contributors or barriers to your success. For example, changes in external policies, concurrent or competitive campaigns or a global pandemic can directly impact our activity.

Novel analytical techniques allow us to capture how change really happens and even predict early on what is working and what is not. This is key for effective decision making and essential to ensure our investments have a positive impact.

Then, why companies, foundations and health providers still rely on the most basic experimental design instead of leveraging from the most recent analytical developments? Recent surveys have found that the main barriers are: the lack of technical capability, the perception of a high degree of complexity and an unclear vision of the benefits of these innovative techniques compared to the most traditional ones. In other words, considering data science to inform these decisions is attractive but still out of our comfort zone.

The key to a successful incorporation of the most innovative data science developments in your projects is partnering with experts in making data science simple.