Data ART… not Science
Data science gets a lot of lip service, but ART is the true foundation for your business. A data warehouse that addresses ART first is the surest way to assure success.
Accuracy, Reliability, and Timeliness (ART) - are the three pillars that will forever dictate the success of your organization.
Balancing these three requirements will provide the mechanism for all future data science and analytics at your company. It will also increase the likelihood that these efforts will be actionable.
Accuracy
Your data must be accurate. Not just close. Analytic problems aren’t solved with Data Grenades. Although, data problems are often lobbed over cubic walls with explosive consequences.
It is equally not nirvana. Data Perfection — a term no one uses for good reason — is not a highly achievable outcome either.
So how accurate does your data need to be?
The answer should be driven by the job it is supporting. Growth companies, start ups, robust marketing analytic shops are most often looking for outcomes showing 10–20% growth/lifts. The data to support that analysis better be able to box within 5–10% of your system of record.
More mature companies, those with deep market penetration, finance forecasting teams, and analytic teams engaged in optimizing large pools of risk, capital, or transaction expense, will be gunning for 2–5% lifts. The data to support that effort will need to be 99% accurate or better.
Reliability
Accuracy is normally judged on the long term, over week, months, and quarters. Reliability is more event driven. Tracking reports, alerts, new product launches, and a myriad of other data functions can be thrown into utter disarray by these events. Events include outages, wild but temporary inaccuracies, and other general nonsense.
The fewer Data Events, the better.
Poor reliability will degrade the value of analytics, destroy forecasts, corrupt models, and generally cost your company a lot of money (and opportunity). These events are the sort of problem that causes the CFO to lose sleep and the CAO to lose their hair.
Timeliness
Finally, even if the data is accurate and reliable, it needs to be timely. Data, like food, is best when it is fresh. Fresh Data is more predictive, more useful, drives faster decisions, and generates higher returns.
This is NOT about Real-Time.
Not all data needs to be real time. Fresh food is great but it is a rare restaurant indeed that would harvest their meat or vegetables right in front of you. It is just not practical. It would also be ridiculously expensive — every restaurant would need to be half farm.
If our aspiring farmer/restaurateur truly sold out to freshness, you might expect them to harvest your potato directly from the soil to your plate. The meal would quickly become one of dirty vegetables, gritty mussels, and briny seafood. Data often requires cleaning and preparation, too.
Some data actually requires a little aging to be best. Like fine cheese and wine, the data needs to build structure, complexity, and character to be best. Even Beaujolais Nouveau, a wine that is ALL about freshness, needs a few weeks to be ready.
So remember — the ART of Data comes first!
Happy Beaujolais Nouveau Day!— November 19th this year
Quintessentially is an article format created by Corsair’s Institute to increase the reader’s comprehension of key concepts by providing several distinct views on a central theme. For more articles from Data, Quintessentially — click here.
For more information on the author visit his profile on LinkedIN — George Earl
