Data analytics undoubtedly means different things to different people in different places, but there’s probably a common understanding at some level, which involves examining data in such a fashion as to generate actionable insights from that data. What that examination entails, and how those insights are generated and presented, are unquestionably book-length (or longer!) subjects. In theory, analytics involve the quantitative and/or qualitative analysis of aggregated data, the presentation of those analyses in a format that includes some level of visualization, and the expectation that outputs, particularly predictive and prescriptive analytics, can in some way drive, affect, or inform organizational actions.
The actual practices of analytics may be, shall we say, a little murkier. Our clients report that they want to put more data in front of their analysts, that they want those analysts to ask more questions of and about that data, that they want those analysts to generate timely and informative responses to those questions, and that they want those responses to be presented visually, and preferably interactively, for decision makers to review and potentially act on.
Each of these desires is predicated on assumptions, many of them not fully unpacked, and some possibly not even examined. For example, why do we think more data is better? What kinds of questions are these analysts posing, and how were they developed? Does aggregating data for display in a chart or graph actually mean that useful or valuable work has been performed? Where, exactly, do the insights lie, and how are data consumers going to recognize and act on them?
There’s been a lot of hype about analytics over the years, and throwing AI into the mix may have simply increased the volume and volubility of the hype. We’ve heard from a lot of people over the years that their organizational vision for analytics is more like witchcraft than science, since many at the senior level who may secure funding have a hazy idea at best of what's involved. We imagine they understand analytics as something like putting a lot of data into a pot, and then doing something like boiling or stirring (but with a special stirrer, maybe?), and then expecting intelligence and insight to simply rise to the top.
To get more data in this notional pot and to heat up the stock and to provide the special stirring often means increasing the number of people dedicated to data analysis, and upgrading the technology stack around data. Yet for many organizations these steps alone do not usually solve the problem of not being able to generate and act on insights derived from data.
If your vision for analytics involves pointy-nosed crones and a bubbling cauldron, then adding AI to the coven probably won’t make things worse! Give AI access to all your data, supply it with some minimal prompting about generating insights for your business, and see what happens—if nothing else, AI’s processing power and facility with language ought to come up with something interesting.
Interesting, of course, is not the same as meaningful, and even when you have meaningful information you may not be able to act on it. Figuring out where the meaning lives in your analyzed data almost certainly requires figuring out the meaning of the data you capture in the first place. If we have a shared understanding of the data we collect, where we store it, and how we use it for operational purposes, then we can identify reasonable questions that haven’t yet been asked, or that we haven’t devoted sufficient resources to answering. If our culture values data, if our employees seek to use data in their regular work, if our leaders share publicly how data helps guide their decisions, then analytics can likely grab a firmer foothold.
That's a lot of ifs. And we observe that most organizations still don't have a shared terminology for data, or a complete catalog of data assets, or something of a culture of data. This leads us to a certain skepticism: not any given organization’s interest in analytics, but that organization’s ability to succeed with analytics. It’s tough to produce reliable information from ungoverned data for operational reporting, much less sophisticated analytics. And while we think the potential for AI to help you with data management needs is high, we suspect the potential for mischief when you set it loose without structure and guardrails is also high.
We expect to come back to this topic again and again. Among other posts, we want to take a look at practical data quality in the age of AI, and we want to offer up some thoughts on what AI-generated or AI-supplemented analytics might look in a data-enabled organization. So stay tuned!
Hope you found this blog post beneficial. To access other resources (blog posts, videos, and recorded webinars) about data governance and data intelligence feel free to check out our data governance resources page.
Feel free to contact us and let us know how we can assist.
(Image Credit: StockSnap_XBL8EKLLAK_ManDeepThought_AnalyticAI_BP #B1291)