The Good, the Bad, and the Ugly of Sampling Automation: Part 1 (The Good)

By JD Deitch, P2Sample

Automation is proving to be one of the most powerful trends of this century. It is affecting nearly every industry, including market research. Most of the attention it has received in our industry has been directed at making the data collection and insight generation process more efficient. This is only part of the research equation though. The impact of automation for sampling is equally important—and, in some cases, downright distressing—for the research process.

For at least five years, since the arrival of two very prominent DIY data collection platforms on the marketing research conference scene, market researchers have been discussing the broad impact of technology and automation. Overtones of these issues were present in the 2009 bombshell report by Grey Matter Research. This documented the extent to which the tail of technology and automation, from routers to river sample, had begun wagging the dog of sound sampling practices.

Despite this, and despite literally years of articles and conference papers, there is still little understanding of, and attention being paid to, the implications of sample automation. Over a series of three articles (this being the first), and with apologies to Sergio Leone and Clint Eastwood, here are the Good, the Bad, and the Ugly aspects of sample automation as things currently stand.

What is being automated?

The word “automation” is used so frequently that it suggests a concrete and shared set of principles and practices. Nothing could be further from the truth. Let us first define automation by what it is not: it is not DIY (Do-It-Yourself). By automation, we mean the absence of humans in the fundamental execution of certain steps. Automation enables both experts and non-experts alike to leverage technology for greater efficiency.

While the extent of automation varies by company, most automation efforts for sampling are focused on four things:

  1. The exchange of quota information for a bid/project (tied to the questionnaire)
  2. The extraction of sample data from a database and its provision for a project
  3. The “collision” of a questionnaire and participant
  4. The monitoring of field status (completes, disqualifications, overquotas, abandons) for a project

Fewer companies are automating the following:

  1. The transfer of data
  2. The recruitment of participants to maximize reach as well as other things like re-contact rates
  3. More sophisticated field tools like quota balancing, click balancing, and throttling
  4. Fraud prevention

The Good

Few dispute the efficiency gains of sample automation throughout the insights supply chain. Suppliers benefit from faster execution, lower operating costs, and reduced error rates. Clients feel these benefits as well. (Regrettably we can see that most suppliers are not sharing some of the cost savings with participants in the form of better incentives, though we will get to that later!)

We begin to see differences among suppliers when it comes to field monitoring. In principle, automation should empower a host of capabilities, from real-time problem detection to mid-stream supply corrections to live insight into survey design issues at any hour of the day. Yet few firms (to our knowledge) have this in their DNA.

Tune in on Wednesday to find out more about the bad aspects of sample automation as it currently exists in the industry.

By JD Deitch, P2Sample

Share: