New Analytic Strategies: lots of new toys plus Big Data – what to do now?
The key objectives of Retail remain the same. Measures like recency, frequency and value are what defines the profitability of the different retail ‘entities’ such as SKUs, lines, ranges, single stores, regions, chains, brands, campaigns, seasons, segments, channel and even individual customers.
But the increasing availability of Big Data information resources makes it hard to figure out what to analyse for and how to go about doing so. The many new shopping technologies and new supply chain technologies mean that there are too many new data sources and associated methods of using them to make it a simple choice.
The business outcomes are the same. But how can you connect these outcomes to the appropriate analytics approach, and the analytics approach to the vast choice of data that could be used?
This problem is related to part of the fragmentation problem below. When I work with retailers on customer data projects a really common barrier is fragmentation of data sources and multiple versions of the truth. I think that’s what’s behind some of the issues in Econsultancy’s report on what is holding back social marketing.
But I don’t get it. OK, there are lots of sources of data, internal and bought-in, but they are mostly known and its not a new problem so there are lots of IT integration platforms.
So what’s the problem? I think its because some firms do not really understand what the actually use of the data is – i.e. its not well specified. Then the integration (in all senses of the word) problem becomes a moving target. Which makes the problem complex.
You cannot take data at face value. Before you can use it you need to know its context – its history, provenance and the circumstances of how it was generated. You need to have permission. But permission depends increasingly on what you want to use it for and there are so many different things that a single piece of data can be used for. Right now and in the future.
The challenge is not a technical issue. It’s a matter of knowing what data you actually need and have, what analytical approaches are available to you and what strategic questions you can ask. The pattern of data, analytical processes and strategic questions can be configured in so many different ways that it makes it hard to choose. It also makes it hard to know you have made a good enough choice.
Its like the ‘chicken and the egg’ question – which came first? If you have this data what analytical process can you feed it into and what will this tell you? Or, starting at the other end, if you have these strategic questions then what analytical process will give you an answer and what data will that require?
I think the problem is about getting to grips with choosing New Analytic Strategies. A lot of confusion is caused by too many new analytical approaches that can now be used to deliver the usual business objectives. It can be unclear which approach to use and what data to feed into it.
Solution: as with all complexity problems its always best to start with the outcomes that you want and then work backwards. Working backwards damps down unmanageably large numbers of options. Here, it can systematically map out the flows of cash, transactions, customer decisions, queries, attention and awareness in a sort of ‘reverse funnel’.
Don’t start with what data do you have – you don’t know what you want it for yet and you can always manufacture more or buy it in. And certainly do not start in the middle with sexy new analytical fashions that are just begging to be used.
Use a ‘reverse funnel’ to investigate what your analytical processes needs to end up looking like. You can then compare what different analytics methods and software are able to deliver – and what data they need to do so.
Think of your analytical strategy as an array of Business Intelligence telescopes that tell you different things about your customers, your business and your environment. Here’s how Google+ and Facebook are starting to configure their very different arrays of Business Intelligence telescopes on their members.