FOLLOW

FOLLOW

SHARE

Are You Focused On The Right Analytics?

The analytics continuum: descriptive analytics to predictive analytics to prescriptive analytics

21Jul

The Analytics continuum: Descriptive Analytics → Predictive Analytics → Prescriptive Analytics. Are you focused on the right one?

Let's assume I am a marketing analyst for one of the largest discount retailer in the market. My company has invested heavily into one of the largest analytics, (actually, I meant descriptive analytic tools in the market), name ends with a U. I look at trends as soon as data is loaded into the tool. My end consumers include the VP of Marketing, the VP of Category Management, and also the Finance team. The Sales team also has access to my reports. I am the front line for all of these folks.

Every quarter, we have a planning exercise which starts with all of the above participants getting together in a room and talking through plans for that quarter. I start by providing each team its most updated and refreshed dashboard. Based on sales and performance seen in these reports we determine products, sub-segments, and promotions for the new quarter. The only problem is that once we have some conclusions, we then ask the predictive analytics team to run some scenarios for us. That is complicated because they have to update their data sets and then run a few different simulations around some decisions we make. Most of the time we have to change one or more variables and re-run simulations. There are about 6-10 people involved in this exercise. Sometimes we get to the specific answer that we all have agreement on, but it requires a lot of interactions between the different teams.

Sometimes, because of the complexity of all these steps, we rely on our experience to come up with the plan more than any prescriptive analytics!

We claim to be data driven, but it’s a little complicated sometimes. So we 'kind-of-use reports', but really we are using our own business acumen to make the final call.

The problem with the above analytics continuum for this marketing analyst is that it is setup for failure. The core reason why people spend time on analytics is not to have intimate access to the latest reports! While knowing what just happened is always important and useful, in reality, it tells you how well things happened, or did not, it gives you an overview of financial performance in the past. However, what it often fails to comprehensively do is to focus on what you should do tomorrow. A lot of companies are spending a lot of time on descriptive analytics, not on the rest of the continuum. If the core goal of looking at analytics is to plan for the future, then this is where people should be focused! If the most important process is to come up with the plan for the future, next month, quarter or year, the prescriptive analytics part should happen first. In fact, the entire continuum should be inverted – and perhaps collapsed into one step. Tell you what to do first.

This is the continuum: Imagine a sales exec, she is using Salesforce, there is an embedded analytic scenario modeler right inside her interface, the exec, wants to focus on certain business goals, she is able to run different scenarios right from Salesforce, by changing a few variables in the widget. She gets her recommended plan first. There are predictive models being invoked by the widget, there is machine learning being applied, data sets from sales, social media etc. are all being combined and yet the end user has no idea about the complexities underneath. She has a simple executive scenario widget, which she can change variables on. This is the change in the analytics continuum. This way we focus on the plan for tomorrow, start with the prescriptive analytics and by the way, if you want to see previous quarter's, or year’s performance all that is accessible through this widget as well. In a sense, we have inverted the analytics continuum and collapsed it from three costly steps to one efficient, cheaper and easy to use step: the end recommendations module!

Comments

comments powered byDisqus
Turkeyss

Read next:

What’s On Data Analysts’ Plates This Thanksgiving

i