In September 2014, Gartner’s Annual Big Data survey reported, “Looking only at those organizations that have invested in big data, 32% report having deployed, up from 25% in 2013(1). ” Equally worryingly there is a general rule of thumb within the industry that 80% of a BI project cost is in getting the data prepared.
There was a high profile big data failure during the last US election. The Romney campaign developed a big data solution to help ensure voter turnout in marginal areas. Even Google got their Flu tracking predictions badly wrong.
As Internet of Things (IoT) embraces big data as one of its key enabling technologies, then clearly the reasons for these failures need to be examined before embarking on an enterprise-wide IoT deployment. Looking at the reports it shows that there are a couple of key reasons for the failures.
Firstly, enterprises try a 'big bang' approach, investing in very expensive systems before they understand the value that can be gained. Secondly, they do not take time to understand what they are trying to get out of the big data investment and therefore have no metrics to measure success against. CGI advocates a 'start small, but start' principle in IoT that avoids these pitfalls. In the Netherlands millions of euros was spent on smart meters only to be derailed when it was found that smart meters violated article 8 of the European Convention for the Protection of Human Rights and Fundamental Freedoms (ECHR).
'Start small but start', means that you put together a small project or pilot to prove how IoT can add value to your business. The project, once successful is used to build upon, like lego bricks, incrementally adding value in each step. The 80/20 rule remains valid that 80% of the value is from 20% of the data.
Starting small has a number of advantages.
- The outcomes can be clearly defined, agreed and tracked as delivered use cases rather than technology blocks.
- Costs can be controlled by bounding the project scope at each phase.
- It is easy to keep the stakeholders appraised of progress, which keeps then on line.
- A small project team can be used which saves the issues with building a large team from scratch. The risks of standing up large teams are well known.
The ‘start small’ paradigm also enables some secondary benefits. A small team, working on a pilot naturally works in an 'Agile' way, so you should end up with a project that has Agile methodology baked in. IoT thrives in an Agile environment where small incremental enhancements and new use cases are added in regular releases. This is because of the way IoT democratises insights and makes them available across the enterprise. As more people see the insights, then more opportunities to gain additional business value appears.
Facebook lives by the motto 'move fast and break things' in their development, but most enterprises cannot work this way as they are trying to deliver against key benefits. On the other hand an Agile development methodology can be incorporated. Agile is a methodology by which requirements are developed just in time and development is done in short 'sprints'. By keeping the sprints short, it allows for refactoring of parts of the solution ‘as you go’, for the cases where problems are found. This means non-optimal modules can be fixed without being stuck in the codebase for years.
In a real development this would end up with the development cycles (called sprints) being of 1-2 months. Whilst a sprint is ongoing the use cases for the next sprint are being prepared.
But is it possible to start small and grow without having to re-architect? Historically small pilots and proof of concept work, cuts corners in order to deliver, which end up limiting the expansion potential. Areas such as security are often completely missed and there is not enough time to architect the functionality in an enterprise grade manner. This is where you need to choose the right platform for your development. CGI has invested heavily in a platform for IoT enablement that radically reduces the project setup time for IoT and enables scalable implementation from day 1. This platform has already been successfully used to develop small pilots for our clients, an example being the ThyssenKrupp Elevator demonstration. In just a few weeks we were able to build an end to end demonstration where all elements were re-usable moving forwards.
Starting small no longer means a ‘quick and dirty’ pilot that then needs to be discarded. You can have your cake and eat it with a low risk pilot that then expands as needed.