Usage frequency is a critical variable to marketers. Frequent users of our products and services tend to look and feel different than lower frequency users. Understanding these differences can help guide strategy that maximizes revenue and profitability. The question at hand then becomes how do we measure frequency? There are two ways to get to the question of frequency of use. The first is to analyze your transactional data and create a suitable measure, such as the average number of visits in the last week, month, year etc. This provides a reliable benchmark, but it does not allow you assess the attitudes behind those behaviors.
Alternatively, we can ask our survey respondents how frequently they have visited a website or eaten at a restaurant for example. The advantage with this approach is that it allows you tie this frequency to other variables of interest, such as performance on key service factors or dollars spent. However there is a caveat, in that the farther back we ask respondents to recall the less accurate their recall will be. In short, if you keep the time horizon to the immediate past then the data will be more reliable.
Assuming we want to ask patrons of Moe’s Diner how frequently they visit the restaurant there are options in how this question can be asked.
Option one:
How frequently do you eat at Moe’s?
Very infrequently
Somewhat infrequently
Occasionally
Somewhat frequently
Very frequently
Option two:
In the last 30 days how often have you eaten at Moe’s?
Either of these approaches will yield workable data. However, the first option, with its scaled approach, is limiting in that we do not know what the differences are between categories. This alone makes it less actionable. The second option can be presented as an open-ended question or scaled (e.g. none, 1 – 2 times, 3 – 5 times, etc.). Open-ended will provide numerical data, scaled will provide ordinal data. Either of these is superior to option one.
Capturing frequency of usage in your customer satisfaction survey allows marketers to establish workable benchmarks. For example, in analysis we would check to see if there are differences in average amount spent by frequency of usage category. These measures can provide actionable areas for improvement. Again an example, if you know that moving a customer’s average usage from one time per week to two times will increase revenue by the average amount of the bill this provides a realistic target to work toward.
Lastly, if you have actual transactional data from a CRM or marketing data warehouse, it is advisable to calibrate these figures with survey data. Odds are the respondent’s recall will be off to some degree. Calibrating the two sources affords you the ability to create a correction factor based on real transactional data that can then be applied to survey results.