Supercrunch Blog

Merlin Pappenheimer July 26, 2017 Big Data

Big data – big return?

Data science is all around us these days. It’s the new big thing and considered to be the solution to almost any challenge that companies face. Being a data scientist is supposed to be the sexiest job of the 21st century and the faith in the power of analytics seems to be the mantra of the new generation of business leaders. We at SUPERCRUNCH believe it too; we all love the seemingly endless possibilities of todays’ big data world. It’s challenging, it’s exciting, and it’s big fun. But working as an interface between science and the business user also shows some additional aspects which shouldn’t be underestimated when all of the hype is over. Data science projects will (hopefully) all come to a critical stage – when business users will start working with the outcome. We all have to face that fact at least for some more decades – as machines won’t ultimately take over too quickly.

One strategy to avoid frustration at the late stage of big data projects – the implementation phase – starts at the very beginning: A use case driven approach puts the user in the center of analytics projects right from the start. My colleague Stefan Roese just recently published a blog article on why this is so important. I want to add two aspects I’ve experienced, which are necessary to maximize the return on analytics investment (ROAI).

Integrate statistical results into day to day routines

Some of you might have experienced similar situations; no matter if you are a consultant in the field of analytics or a user yourself: Your data science team developed awesome algorithms to support business questions. But the way the results are presented and communicated to business users is just so far away from their habits in digesting information that they struggle – or refuse – to integrate it into their working routines and you end up with wasted budgets and nothing but frustration. One way to avoid such frustration is to integrate results from sophisticated statistics into your day to day reporting environment. Users will appreciate and use the magic of machine learning (et al) instead of running away from it.

Let’s take marketing mix modelling as an example: A very established methodology (according to a Forrester study used by 9 out of 10 marketing and analytics professionals) and proudly communicating its own return on investment, MMM basically delivers a contribution factor of marketing channels and touchpoints on sales results and is widely used for strategic budget allocation. But why use it only once a year or even quarterly if it could be integrated into reporting on standard status metrics on your marketing touchpoint performance, and thereby increase the return of such a costly project? We’ve recently managed to integrate such impact metrics and increased not only the return on analytics investment, but also established acceptance of data science on the business side in general.

Introducing analytics into an organization requires specific change management – or an ease of use solution

Another challenge we experience when implementing advanced analytics is the fact that the integration process is tough. But we see it as a positive: it’s a strong lever to maximize the return of companies’ analytics investment if it is done the right way. Deployment of science-based solutions is not only a technology thing, but a matter of change management. A change of mindset management. Management level support is certainly helpful too, but the key to success is a holistic approach; ideally driven by a dedicated (analytics) change manager. The best solutions are still those not requiring much change; those that seamlessly integrate into users’ routines and don’t require much of a mindset change: Simple to use with the science where it belongs – in the backend!

What are your experiences when rolling out advanced, science heavy solutions? Did you achieve a smooth integration into existing workflows? Which obstacles were you facing and how did you manage to overcome them? How did you maximize your return on analytics investment within the implementation phase? We’re interested in hearing your thoughts and discussing your best practice experiences with you!