“In the beginner’s mind there are many possibilities;
in the expert’s mind there are few.”
-- Shunryu Suzuki
Confession: I am a total nut for metrics.
In fact, I have frequently driven my teams and colleagues a bit crazy by insisting that we measure and quantify everything. My career, my experience, and my MBA education all drilled into me that “you get what you measure.”
For the record, I’ve found this approach to be exceptionally effective in bringing order to chaos and driving execution results, especially in ultra-complex global organizations and when managing large, complicated multi-year programs.
Designing a metric requires a clear framework and well-defined questions and parameters. The process works admirably for things that are already well-understood. For example, if we’re measuring customer complaints, we can easily define metrics around the number, type, and severity of issues vs. the number of interactions or opportunities. If we’re running a help-desk, we can measure ticket aging and cycle time. If we’re running a planning organization, we can measure forecast accuracy. Running sales, we can measure opportunities opened and closed…and so on.
So where does Zen master Shunryu Suzuki’s maxim enter into the world of data, analytics, and metrics?
Things change when we move into the world of unstructured data and complex root cause analysis. The questions are no longer so self-evident. Neither are the answers. In fact, when we presume that we know enough to define and quantify all possible business problems—i.e. when we come to the data with the mind of an “expert” who has narrowed down the field of possibilities—we miss major opportunities to learn from the data.
In contrast, when we provide space, light, and oxygen to explore the data and let it tell its own story—i.e., when we come with the mind of the “beginner” without rigid preconceptions—we open an entirely new world of inquiry and value.
We’ve seen some compelling examples of letting the data speak for itself in deploying our qSuite platform for unstructured data and analytics.
In the IoT space, a capital equipment sales and manufacturing company came to us with a beginner’s mind. “We have all these technical machine log files. We don’t know what they say or mean, or what to do with them.” That’s just the kind of open-ended question that can be a great way to start exploring the value of data. Once we ran the log files through qSuite and were able to assemble, interpret, and visualize the data, we detected two patterns. Competitors’ consumables were being used on the machines and there were software training issues we were able to detect through software errors thrown into the logs by the machines. By letting the data reveal its secrets, we uncovered both a revenue opportunity and a customer experience improvement opportunity!
In contrast, the “expert’s mind” caused a proper change management problem with another client. We used our qSuite qSci module to help a global Pharma company reduce product discards due to short shelf-life dating. How their experts started the conversation with us was: “It’s less than 1% of sales…Planning’s not a perfect science…We’re always gonna have mismatches of supply and demand for products with expiry.” Basically, the planning team thought they already understood all the root cause drivers. So we started with some hypotheses from the experts about what we could improve. We also ran “unsupervised” machine learning on 2 years of their transactional supply chain data. We found that there were some obvious things that the experts got right: problems in master data setup and forecast inaccuracy, for example. Next came the “obvious” findings (at least, they were obvious once we discovered them): such as batch size variability and differences between component quantities ordered and received. Then, there were findings that emerged from our clustering analyses no human expert is ever likely ever to discern: e.g., for any component used in more than 23 finished goods, those products are twice as likely to have high discard rates. From all of the factors, we were able to construct a comprehensive diagnostic and predictive model that reduced product discard rates, driving significant P&L savings.
One final example.
Think for a moment about what question you’re asking when you search Google. Sometimes it’s very specific, sometimes quite general. Often, you’ll find brand new connections to adjacent topics, answering questions you didn’t even realize you had and generating new questions. A beginner’s mindset. This is exactly what happens when we deploy qSuite’s qFind module over unstructured data. Frequently, our clients just start with a need to find product data or customer data across multiple sources and within documents and pdfs. “qFind, show me all our marketing brochures for Product X.” The magic happens when all that data is together for the first time. Suddenly, engineers are seeing common features across products, saving R&D cycles and costs. Marketers are seeing that their public web-facing and internal materials are inconsistent and need to be fixed. Customer service reps are seeing all activity for a customer without having to say, “Hold Please.” This type of insight, enabled through Search and data interconnectivity, is a terrific example of the benefits of bringing a beginner’s mind to your data.
It’s been much observed and commented on in the AI press, and our experience has proven this time and again, that the combination of machine and human is more powerful than either by itself. We not only frame some hypotheses up front as experts to motivate the project, but also allow the unstructured data and implicit connections in the data to speak for themselves.
Listen to your data. It has important things to tell you. It wants to be heard. It wants to be your friend. Let your beginner’s mind become your data mind.