Fixing Predictive, Making Anticipatory Work

Thanks for coming along… to refresh your memory: three part series

The concept? Predictive analytics was badly implemented, but new technologies and knowledge have made it obsolete.  There’s a new model that can be used, an interesting new model: anticipatory analytics.

This third and final post in the series will show you how to make it work – the rest had been more theoretical.  I want to really explore how to make the art of the possible work now.

How is this possible? It involves five elements:

  • Fast data management – the technologies that we use to manage data have changed dramatically in the past five years or so. From in-memory processing that does not require storage to understanding of unstructured data to faster throughput and better connections between systems we are able to manage data magnitudes faster than in the past – leading us to real-time (or near-real-time) use of data to let computers automate decision making. This is what allows us to create in near real-time customized offers that would have a far greater likelihood of being accepted.
  • Sharing of data – consumers, in the age of the customer, are sharing more and more data today thanks to social channels, social networks, and online communities. Information that was before hard to obtain or to understand it is today freely shared in public places for organizations to leverage and use to deliver solutions of mass personalization. This highly customized and attractive offers can only happen when data about sentiments is freely shared and analyzed.
  • Loads of data – there is no doubt that we are seeing more and more data being generated, but the counter side to that (organizations being able to manage all that massive load of data) is more interesting. Long gone are the days when using more data meant more expensive storage to accumulate it and welcome are the moments when either the storage is so cheap that is not an issue, or when the processing that happens in near real time makes storage obsolete and unnecessary. The volume that was suspect of bringing down performance before becomes a load of welcome data when properly filtered.
  • Outcome driven – as more and more consumers and customers are pushing to co-create and collaborate we see more organizations willing to focus on the outcome from the perspective of the consumer, not the company. After all, the outcome for the consumer also benefits the company as it usually relates to more products and services being sold. An ideal outcome for a consumer would undoubtedly necessitate more of a product or service that the organization, or one of its partners in the ecosystem, can offer and benefit from selling. Focusing on the outcome from the consumer perspective is a win-win situation for both parties.
  • Experience based – not just a generational shift, but also a societal move that reflects the advances of the middle class around the world, more and more consumers are shifting away from single interactions to complete experiences. This move, and the demands they make on organizations to deliver against them, is making more and more brands and providers consider ecosystems or complete experiences. In the case above, the airline could offer complete experiences versus a simple trip to create a competitive advantage. Knowing what consumers want for those experiences requires a complete predictive model that leverages all data and sources from all partners to deliver better experiences to consumers.

Of course, none of this happens automatically – organizations must undertake the process to understand what they have to do, setup the ecosystems, implement analytics, and optimize as they go. There are three steps that organizations must take to embrace anticipatory analytics:

  1. Understand Your Customers.  This is past the concept of demographics and who they are, but more around of their expectations, wants, needs, and demands.  What are they really asking? How they expect to get it? When? What is their impression of your duties and responsibilities towards them? This type of understanding is what drives the analysis and decision making you will undertake.  If you don’t know what you need to deliver, you don’t know how the data can help you anticipate their needs (and, btw, make sure you have processes in place to update those needs and wants as your customers are wont to do).
  2. Tie to KPIs.  If you are going to invest time, money, and resources into making this work you better be prepared to show how it supports the ever-moving goals of the organization.  We are not talking about numbers or metrics that are relevant to the process, but how those functions tie to the numbers that tell the story of how the organization is evolving.  The outcomes you are seeking better have a tie-in with the numbers that show the health of the organization – if not, find it.
  3. Evolve. There is no simple way to get there from here.  There is no silver bullet or magic potion that will give you an understanding of your customers expectations and how they evolve, nor is there a way to understand how the specific actions are tied to KPIs.  Further, it is likely that the technology you have you already implemented – no one will sell you an “anticipatory analytics engine in a box” to deploy.  This is about making commitments to see outcomes become realities and the best way to implement the technology you already have to support the new goals you are setting.  Leverage, repeat, learn, and evolve.

Well, that’s it for this series – thanks for reading along… what do you think? something you could do? See your organization doing?

Would love to hear what you are doing or thinking…