This is part of the Operations topic cluster, which belongs to the Business Expertise Triad.

 Members only

Executing on Becoming Data Driven: The Politics

Feature image for Executing on Becoming Data Driven: The Politics

Sign up for the Newsletter

    Once a week. Three links. No spam. Unsubscribe anytime.

    Every analytics project consists of two parts. The technical part, and then the 'get the organisation on board' part. We talk about why the latter is the real challenge.

    Note: this is Part 7 in a series of blog posts about becoming data driven in business. You may want to read the prior parts before reading this essay.

    I’ve been saying for awhile now that the Becoming Data Driven series is going to wind down temporarily, as I start putting the ideas in the series to practice.

    What does this actually mean, though? What does it look like?

    The answer is boring, and perhaps painfully obvious to those of you who have worked on data projects before: it means instrumentation. It means collecting data, building a reporting mechanism, and then iterating until I get to a point where I have a causal model of the business in my head. As a process, this is going to take a few months.

    Two Parts to Every Analytics Project

    There are, broadly speaking, two parts to every analytics project. The first part is technical: it is instrumentation, data storage, and reporting. Someone needs to write the code, plug in the data pipelines, and set up the dashboards. The second part is actually getting the organisation to use the data. It is — to use language we’ve already covered in this series — to help management gain knowledge.

    My belief is that this second bit is much harder. For a few years now, my bugbear has been that data consultants, data tool vendors and data thought leaders focus almost exclusively on the first bit — the instrumentation, storage and reporting pieces. They give long talks about technical best practices. They write long screeds on corporate blogs and plaster LinkedIn and Twitter with posts about data infrastructure. But the truth is that it is relatively straightforward to accomplish all of this. Yes, I am aware that straightforward does not mean easy. Data work consists of many moving parts. The landscape changes. Machine learning is a thing now. There is much jargon. Businesses used to build OLAP cubes on local servers, and then they shifted to columnar data warehouses hosted in the cloud. (Unless you work in the industry, you are not expected to understand that sentence). Building a minimum viable data stack requires actual engineering; none of this comes cheap.

    But my contention has long been that it is much harder to change organisational culture.

    A data leader once told me that he believed there were only two ways you could change a company culture to become more metrics driven: “either the CEO leads the charge, changing executive behaviour from the top down, or every leader who isn’t data driven gets kicked out of the company.”

    I thought this was funny at the time — if consistent with my experience — but the assertion has haunted me ever since. I have searched for counter examples. As of this writing, I have found none. It seems a little bizarre, but every successful change that I know has occurred through one of the two paths (though, I should note — I’m still looking). And there are publicly available stories that illustrate this dynamic. Here’s one of them.

    Originally published , last updated .

    This article is part of the Operations topic cluster, which belongs to the Business Expertise Triad. Read more from this topic here→