Data breadlines and data brawls: an excerpt from the “Looker Book”
June 16, 2016 — Santa Cruz, CA
[Editor’s note: Last week, Santa Cruz Tech Beat published The Looker Book: Winning with Data. TechCrunch has just published an excerpt from the new book by Looker CEO, Frank Bien, and Redpoint Venture Capital Partner, Tomasz Tunguz.]
This is an excerpt from Winning with Data, as published by TechCrunch:
When I think about the behavior of many business people today, I imagine a breadline. These employees are the data-poor, waiting around at the end of the day on the data breadline. The overtaxed data analyst team prioritizes work for the company executives, and everyone else must be served later. An employee might have a hundred different questions about his job. How satisfied are my customers? How efficient is our sales process? How is my marketing campaign faring?
These data breadlines cause three problems present in most teams and businesses today. First, employees must wait quite a while to receive the data they need to decide how to move forward, slowing the progress of the company. Second, these protracted wait times abrade the patience of teams and encourage teams to decide without data. Third, data breadlines inhibit the data team from achieving its full potential.
Once an employee has been patient enough to reach the front of the data breadline, he gets to ask the data analyst team to help him answer his question. Companies maintain thousands of databases, each with hundreds of tables and billions of individual data points. In addition to producing data, the already overloaded data teams must translate the panoply of figures into something more digestible for the rest of the company, because with data, nuances matter.
The conversation bears more than a passing resemblance to one between a third-grade student and a librarian. Even expert data analysts lose their bearings sometimes, which results in slow response times and inaccurate responses to queries. Both serve to erode the company’s confidence in their data.
Overly delayed by the strapped data team and unable to access the data they need from the data supply chain, enterprising individual teams create their own rogue databases. These shadow data analysts pull data from all over the company and surreptitiously stuff it into database servers under their desks. The problem with the segmented data assembly line is that errors can be introduced at any single step.
A file could be truncated when the operations team passes the data to the analyst team. The data analyst team might use an old definition of customer lifetime value. And an overly ambitious product manager might alter the data just slightly to make it look a bit more positive than it actually is. With this kind of siloed pipeline, there is no way to track how errors happen, when they happen or who committed them. In fact, the error may never be noticed.
Continue reading article here: https://techcrunch.com/2016/06/12/data-breadlines-and-data-brawls/