Data analysis is truly one of the greatest professional fields in existence! Particularly because, sometimes, in a very short period, you can give amazing new insights that businesses can benefit from right away. This not only leads to better decisions, but also allows for much faster decision-making. Also, the returns and cost savings can often be achieved very explicitly. Moreover, smart data analysis often leads to the development of new products and services. This forms the basis of important business innovations.
Many businesses are quite opportunistic in their implementation approach for analysis and / or virtualisation tools, because of the enormous hype surrounding analytics and self-service BI. They often have very high expectations for the returns. Unfortunately, actual practice is wilder than theory, and the disappointment is great when the results aren’t achieved. After all, no matter how low-threshold the current generation of data analysis tools are, you need more than a software product to achieve success with ‘self-service analytics’!
When I think about the data analysis possibilities for an organisation, I always look at the entire spectrum of data, processing, analysis and governance. You should develop four aspects if you want to realise the enormous potential of your data:
In my previous blog, I elaborated on the human aspect. What competences should a good data scientist have? This answer should be based on solid scientific research carried out in Europe… In other words, the quadrant on the bottom right-hand side. The quadrant on the bottom left-hand side pertains to the use of technology with which to unlock (big) data and tools like Tableau, Alteryx, Qlik or Board. A great deal has already been said and written on this topic. The upper quadrants are almost as interesting.
Many businesses have learned the hard way that merely having tools and competent people does not automatically make you a data-driven organisation. They are shifting their focus, to an increasing extent, towards the development of a specific vision and strategy for data analysis, which is then pursued at the most senior levels of management. They also devote attention to the adjustment of their organisation and business processes, to better accommodate self-service analytics. Don’t let this scare you off. By no means does it imply that major projects must be initiated right away. It mainly pertains to the process of raising awareness of how one can encompass and arrange data analysis in the veins of the organisation, in a way that is meaningful and well-structured.
I have headed many similar processes, which have enhanced my experience in this respect…
Data for everyone
‘Data is the new oil.’ A popular comparison that is widely used and abused. I have also used the analogy in the past, to explain the value of data, but I stopped using it a while ago. Oil is scarce and exhaustible, whilst data is abundant and the volume of data increases dramatically. Moreover, data in and of itself does not offer any added value to your organisation. Merely pumping out data is not sufficient for it to be comparable to oil. Here is a good explanation: ‘It’s not the data that makes the difference, it’s what you do with it.’
Who better to get the most out of data than those with the closest involvement? Nowadays, the quality and timeliness of the available information and insights make or break the success with which any position in a company is fulfilled. Most job titles have or will become redundant, because 80% of all employees are slowly becoming data analysts. Not just any data analysts, but data analysts with on-the-job expertise. By enabling employees (who possess business knowledge) to generate the necessary insights independently, without intervention from IT experts, they will simply be able to do their jobs much better.
This is exactly why analysis and visualisation should not be restricted to a limited group of experts in your organisation. Data – and tools that enable analysis of the data – should be available to everyone, to generate maximum value. From working better and faster, to entirely new products and services. In this case, data even becomes the enabler of new organisational and management ways (read more on Agile leadership here).
Thinking in terms of positions
You can turn self-service analytics – and who should do what exactly with data – into a very fundamental discussion. However, it is much easier to think of it in terms of demarcated positions and to clearly define and encompass (and organise) each position where it belongs. Examples of common operational analytics positions (subdivided into groups) are:
These positions all contribute, to ensure that the right insights are available at the right time. However, only the Analyst and Developer group are relevant for self-service analytics, and even that could still be two-fold. For example, the Report Developer or Business Analyst positions can easily be widely encompassed within your organisation, whilst the same does not apply for Data Engineer or Data Scientist. After all, you want to encompass those tasks centrally, because it involves more specialised work (for now). Organisation becomes much easier when you are aware of these subtle nuances.
Organising the data democracy
The point of departure for making data freely and widely available within your organisation, is often referred to as Data Democratisation. This is the opposite to the approach where all data, data manipulation, analysis and reporting take place centrally. A highly applicable definition, which should not be confused with data anarchy! After all, without a clear vision, strategy and processes for the release of data, the process will soon resort to anarchy. As is the case in a ‘real’ democracy, structure and rules are needed for the sake of order. Everyone can do as they please, if they follow the rules. In a data democracy, we also need a House of Representatives to adopt laws, a Senate to review the laws before actual implementation, and a Ministry of Security and Justice to monitor compliance with these laws and regulations. These bodies determine the playing area in which the (data) society can properly function.
The democracy analogy can easily be applied to your own organisation. I recommend introducing an ACE when you apply this analogy. An Analytics Centre of Excellence, as the backbone for anything and everything related to the unlocking of data. Many of the above mentioned operational positions in your organisation will be safeguarded within the ACE, in a centralised or decentralised manner, supplemented with several administrative positions. An ACE can be arranged in many ways, if the design corresponds with your organisation’s vision & strategy. However, always ask yourself these two essential questions:
- Where will you place the ACE in our organisation? This was usually placed under the IT department, but nowadays we are more inclined to place it under the CFO or Chief Information Officer.
- Will this team be organised in a centralised or decentralised manner? The choice increasingly goes out to compile the ACE from positions that are encompassed in a decentralised manner. However, not all positions are suited to the decentralised approach. It is best to encompass specialised positions centrally and to encompass positions like Report Developer or Business Analyst with the departments, as much as possible. This will result in a virtual team of so-called embedded analysts or key users.
The answers to the above questions will largely determine the structure and responsibilities of the ACE. If, with the second question, you opt for a decentralised approach based on data democracy, the ACE’s main tasks will be coordination, communication of the vision & strategy, and programme management. Other activities that may come into play with the ACE are:
- Data governance
- Coaching and training
- Thought analytics leadership
- Best practices
- Advice and consultancy
- Technological choices
- Central management of the (big) data architecture
In this case, think of the ACE as the House of Representatives, the entity responsible for determining the playing area for the business (with an Analytics Board, for example, as the Senate). The goal is maximum playing area, without your organisation deteriorating into data anarchy. The ACE is the driving force behind innovation, but is also responsible for the reliability, valuation and availability of data. You can imagine how difficult this might be!
Supervision versus flexibility
When we talk about big data analytics, we soon encounter two conflicting interests:
- Stabilisation, renovation and management, to ensure reliability and predictability
- Innovation and the need to experiment, to ensure renewal and agility of your organisation
Gartner introduced the ‘bimodal’ concept for this. A bimodal approach does not view these two interests as opposites, but instead tries to unite them, because both are essential to achieve change and to work together coherently. There is much debate about the bimodal approach, from various standpoints. Feel free to compare this opinion with this opinion. When we talk about data and data analysis, the discussion is simple in my opinion. Yes, you want to analyse data in a fast and clever way, to get ahead of your competitors, but without proper data governance guaranteeing the reliability of your data, the resulting risk factor is too high if one were to use incorrect information. This does not even take laws and regulations on data use into account yet.
In that regard, it is also worthwhile to delve into Gartner’s pace layer model. This model reasons from three layers, where the data is the point of departure:
- Systems of Record (run): improving execution
- Systems of Differentiation (grow): differentiation from your competitors
- Systems of Innovation (transform): new ideas
For this model, an approach has been established, ensuring that these layers connect to each other within an organisation. In my experience, this approach is practical in its application.
No size fits all
If, as an organisation, you want to get everything out of your data, you must look beyond data and tooling. I have outlined some of my experiences, but I know that every situation is different. Everything is determined by the sector in which you operate, the strategy that you have chosen, and your specific data-analysis goals. James Kaplan put it nicely: ‘There are no silver bullets.’ There is no single, fast, magical solution to ensure perfect implementation of self-service analytics. You must be able to identify technology, understand the impact thereof, be able to combine it and make smart decisions on how to use the technology. Moreover: you must know how to build operational processes and organisational structures around it.
By now, we can provide a good explanation of which ingredients are essential to also achieve success with data analysis in your organisation. The quantities and order in which the ingredients must be added, and whether they are all relevant, differs across the board, just like the best way to set up processes and organisational structures. It is and will continue to involve a customised approach in every situation, because no two organisations are the same. Also, as with any type of customisation: first think about your goals and your approach, before you start building!