Page 88 - The Real Work Of Data Science Turning Data Into Information, Better Decisions, And Stronger Organizations by Ron S. Kenett, Thomas C. Redman (z-lib.org)_Neat

Moving Up the Analytics Maturity Ladder 79

Figure 16.1 An example of a control chart tracking individual measurements (upper chart tracks the
measurements, lower chart their moving range representing variability). The right panel provides
summary statistics. Source: Figure from SPCLive system by KPA Ltd.

Ambitious companies move up to the next level of the data analytics maturity curve, quality
by design, when managers extend this forward‐looking thinking into the design of products
and services. This stage requires statistically designed experiments, robust design, and other
methods to ensure the product/service meets customer needs and performs well, even when
the raw materials are of uneven quality and environmental conditions vary. Genichi Taguchi
and Joseph M. Juran played key roles in developing this approach. In the 1980s, Taguchi intro-
duced the West to methods he developed in Japan in the 1950s (Taguchi 1987). Juran described
a structured approach for quality planning that started with understanding customer needs and
ensuring those needs were met in the final product (Juran 1988).
In quality by design organizations, the data scientist understands the role of experimental
design and is becoming proactive in the planning of interventions. This is a precursor of A/B
testing (Kohavi and Thomke 2017), where web application designers direct customers to
alternative designs to understand which works best, using data such as click‐through rates.
In service companies, which offer human‐intensive services, there is natural predisposition
to perform quality by design in a more qualitative fashion, because of the following:

• A/B testing is not allowed for compliance/regulatory reasons, or there are policies requiring
that all customers must be treated equally.
• Customer relationships are an important aspect of the service, which is provided by
combining several products and/or services.
• Customers are so different from each other that A/B testing is too difficult.

For more on behavioral big data, see Shmueli (2017).
Neither Taguchi nor Juran anticipated the big data era, with data coming from all quarters,
including social media, web clicks, “connected devices” (e.g. the IoT), personal trackers, and
so forth. This new age poses both new challenges and opportunities and suggests to us a fifth