We shed light on what data and analytics really involves, why it’s different and the key challenges to making the most of it.
Data and analytics: evolving from business intelligence and reporting
It’s always been important to use good information to understand what’s going on in your business and run it successfully. The idea of information isn’t new, but its nature, use and value has moved on a great deal.
Data and analytics is an evolution – a step on from business intelligence, which in turn evolved from traditional reporting.
Increasingly, information is seen as a competitive tool to drive growth, not just a necessary tool to control and sustain a well-run business.
Traditional reporting: pushing past information
Traditional reporting focuses on pushing information in fixed formats to defined audiences. Monthly finance reports are a classic example. The emphasis is on capture and control. It’s static, historic, and a matter of record. It’s essentially about the past: what’s happened.
Business intelligence: pulling information together
Business intelligence made static data more interactive. Fixed-format reports became interactive and online, with more ability to slice-and-dice, drill down and explore. Information became easier to digest with dashboards and other data visualization techniques.
But the information was still essentially backward-looking, based on a defined area and amount of information, and typically sourced from internal systems.
Data and analytics: predictive information
Analytics looks not only backwards but also forwards: considering the future.
It often involves predictive models, showing future scenario options and forecasting the outcomes, to help you understand how best to drive your business forward. It is inherently dynamic, and its power and purpose lies in being a live operational and strategic tool, enabling you to use information far more effectively than ever before.
If traditional reporting is a rear-view mirror, analytics is a head-up display. It is front office and front of foot, with a strong bias to action – analyzing what-ifs to identify therefores.
Moreover, the nature of the information is fundamentally different. It involves not only much larger volumes, but also greater variety – looking not just at your business but at your customers’ data, for example. The data sources are more disparate and, often, externally sourced. There’s also a great variety of unstructured data, such as social media.
Three key differences at the heart of data and analytics
With data and analytics you can look at all the data – every last bit of it, rather than just samples.
This can lead you to see some very different and valuable insights, as well as avoiding drawing some very wrong conclusions! Whenever you look at a sample of data, you run the risk of some sort of unintended bias in the sample, which colors your analysis.
In the 2008 presidential election, for example, many of the opinion polls called the results wrong. It turned out that the polling was done by calling people on landlines, which excluded people who only had a cell – younger, typically more liberal people. The polling was skewed and therefore failed to predict the result of the election. The fact that you can look at and make use of all the data, rather than just a sample size, is an exciting step change in the world of applied information.
Historically, business intelligence has focused on looking for causal links: finding out why. It’s a very left-to-right, linear, process-based way of thinking about information. Data and analytics swerves this by tending to focus on correlations, rather than causes: the what, not the why.
Arguably, data and analytics may say don’t waste your time and energy trying to understand what’s happening; the fact that it’s happening may be interesting and valuable enough. This has some very interesting consequences and practical applications.
For example, flu vaccine stocks are moved to areas where there is a strong likelihood of a flu outbreak, partly based on Google search query correlations. The correlations are hard to fathom – but they work well as predictors. Spotting and acting on correlations without wanting or needing to spend time and energy understanding their causal links is an interesting change of mindset. If you just go with the correlation, no matter how odd it seems, you can end up doing some very different things with highly effective, counter-intuitive results. It’s a catalyst for innovation.
As another example, Walmart identified that just prior to a major storm in the North East, sales of Strawberry Pop-Tarts increased in line with more obviously storm-related products like batteries and flash-lights – and were able to manage stock and increase sales accordingly.
Data and analytics is increasingly open to everyone – from start-ups on a budget to multinationals with deep pockets. This is helped by another differentiator for big data, which is the growth of on-demand, cloud-based services such as Amazon’s RedShift and Google’s BigQuery.
Indeed, these pay-as-you-go services reflect a key change, provoking one of the big questions surrounding data and analytics. With such large data volumes and potentially lots of sophisticated computing power required – should you do it yourself? Or are you better off buying it as a service?
Using a service can avoid major capital investments in buying big databases and the necessary number crunching power. What’s more, the services are inherently dynamic and elastic, fitting in with and supporting the ‘perpetual beta’ (i.e. constantly in development) culture of data and analytics.
But using an external service may present a range of other questions and challenges, such as data provenance, ownership, security and data transfer times. So data and analytics is presenting some big questions on build versus buy – and the best route for you will inevitably depend on your particular situation.
Tackling the data and analytics challenges
Data and analytics clearly offers many great opportunities across both strategy and operations. But as we know from talking with senior information leaders across a variety of leading organizations, it also comes with big challenges.
The right balance of data and analytics, traditional reporting and business intelligence
Both traditional reporting and business intelligence continue to have important and valuable roles. It’s a mix – one that’s determined by the needs and ambitions of each individual organization, but also one that increasingly sees energy, resources and value shift more towards analytics. A major challenge is how best to strike the right balance across all three.
Many organisations want to redirect effort (in both headcount and funding) away from reporting towards more analytics, but there may not be a simple way to do this. The skillsets and culture are very different, so you are unlikely to be able simply to re-use or re-assign the same people. It may require a deeper reorganization.
Creating an effective organization structure for analytics
Data and analytics can be a major headache when it comes to structuring your organisation. What kind of structure do you need? What sort of people should you employ? What skills and capabilities should they have? Where do you put them in the organization (should it be owned by Marketing, IT or neither)?
One key insight here is that you shouldn’t expect to be able to do analytics well by simply reusing or extending your reporting team or your business intelligence centre of excellence. The differences inherent in new data sources typically require different skills, culture and management approaches.
You may well need to employ fresh people – such as data scientists, analytics operations experts, data visualization experts, and those with sufficient understanding of the statistical analysis fundamentals.
And critically, you may be better off placing them with people at the sharp end of your business – your marketers, your supply chain experts, your product development whizzes, for example. Don’t put them in an information team silo; sit them side by side with business folk. That way, together, they’ll be able to make the most of data and analytics’ great potential to drive insights and actions that transform your strategy and operations.
Old data versus new data
The information used for data and analytics may be quite different to that used for traditional reporting, which has big implications beyond organizational and structure skills.
For example, effort that is typically invested in cleansing, modelling, structuring and aligning data used for reporting may be better invested in innovative querying and insight generation techniques for analytics.
Put simply, rather than spending a lot of effort tidying your loft so it is easy to find things, you might be better off simply paying someone who is very good at finding the things you want in a messy loft. The true value in analytics, however, may often come from combining old and new data sources.
Mobilizing and managing data and analytics projects
Making a business case for investment in analytics is hard. By helping you find the questions you didn’t even know to ask as well as discover new, often counter-intuitive insights, it is by its very nature somewhat speculative. There may be gold in those hills, but it’s hard to say just how much before you start to dig. This makes it important to start small, demonstrate the value, and then be prepared to develop in many small, evolutionary steps.
Analytics projects are different: inherently open-ended, iterative, fluid, hard to pin down and constantly evolving. The very scale and dynamism that makes data and analytics so powerful can also create big headaches for traditional project management. And no one is ever forced to use an analytics solution purely to execute their business process; they have to want to use it because they see the value.
All this means different management approaches are required together with the right change management. And focusing on end user take-up and value realization is essential.
Plotting your route forward on data and analytics
There is no easy answer on plotting the best way forward with reporting, business intelligence, data and analytics.
We recommend developing an overall enterprise information roadmap that highlights which information capabilities will really drive value, and a rough timing and sequencing for developing them.
We’d also recommend thinking as carefully about the organization structure and management approach for analytics as you would the technology and data approach. The order of development needs to take into account the underlying data dependencies as well as the business needs. Once the roadmap is set at a high level, the analytics capabilities need to have flexibility to iterate and evolve within that framework – but with just enough discipline and rigor to maintain control.
Whatever course you plot should ultimately serve the aim of harnessing the power of information to reveal valuable insights and actions for your organization.
Discover more
The author
