I recently joined Novarica as a Principal after a brief hiatus from the insurance industry during which I founded and eventually sold a SaaS data analysis and business intelligence company in the health & wellness space. For most of my career I’ve been thinking about how to best leverage insurance data, and after spending so much time deep in the weeds of designing and selling a data analysis solution, I have a new perspective on actionable analytics and BI.
The idea of “actionable data” has been a minor buzzword for at least a decade, surrounding all of the many advances and evolutions of data projects and services. We all know intuitively what it means: make sure that we can take action from our data. And we all know examples of projects that failed to heed this advice, where after much time and money spent on bringing different data sources together and building reports against it, nothing in the business changed as a result.
But how do you actually get actionable data and what kind of action do we mean? After having spent so much time at a company that provided online business intelligence reporting to clients, is it perhaps surprising that I often don’t like reports? Or, rather, I frequently feel a report is just a stepping stone to a better business process. Clearly there are users, typically at the strategic or executive level, who need to be looking at big pictures and trends. But for most operational users, poorly designed business intelligence acts as a dead end rather than a call to action.
Now whenever I see a report, often with excellent data and filled with tables and charts and graphs, I ask three questions about it:
- What is the key take-away from this data, and if it can be distilled down to one or two important facts then why is the whole report even needed?
- Can these one or two important facts be worded in the form of an action or a “to-do” list?
- Can I get these “to-do” items directly into the hands of the person who needs to do them, either via e-mail or some other process integration?
As an example, my data company had a very popular report showing a manufacturer’s product distribution across retail stores. It contained a lot of trend and sales information at a macro and micro level. For an industry with a lot of middle-men and where manufacturers often don’t know where their own products are being sold, this was a very important report. But when we asked the above three questions, we got the following answers:
- The key take-away is a list of the top five or ten stores where a manufacturer’s full product line is not being sold, ordered by how much potential business is being lost.
- The action is to call those stores and sell them on carrying the additional products.
- The best way to get these actions done it to split them up by region, and have a list of stores to call (with phone numbers and product information) sitting in each sales person’s inbox on Monday morning.
The head of sales and marketing wants to look at the big picture and be able to analyze the data, and for her the original report is still the correct source. But for most users, by answering those three questions, we’ve just taken a potential confusing dead end and turned it into a driver of their ongoing behavior. (And for a startup that relied on a monthly SaaS subscription, getting data integrated directly into a client’s operating process whether or not their employees ever logs into the website was a great way to become invaluable.)
Every report in your business should be put to the same three questions. For example, any time a data analysis solution helps an underwriter make a risk decision it’s the result of taking what was once large reports with multiple tables and lots of data, boiling that down to the key rating, and then putting that rating in front of the underwriter right within their toolset.
To be clear, I’m not anti-report. Sometimes insurers’ services aren’t integrated or modern enough to support this kind of distribution across tools and systems. But if you can answer those three questions about a report, even if the kind of automated distillation and distribution of the data won’t happen for a long time, then you can better train users on how and when to use that report to take the action that makes your business better.