For many organisations data is the holy grail; the key to their future success. However, very few possess an effective means to collect and use their data. That’s why we urge our clients to define a clear data strategy: Using their available and relevant data to make effective decisions.
Back in March, when lockdown began, supermarkets suddenly experienced mass stock shortages. Amongst the public panic, consumers began hoarding essential household items such as toilet roll, pasta and tinned foods. In hindsight, the panic buying and hoarding was totally unnecessary. People have been left with a lifetime’s supply of tinned food, coffee, flour and other non-perishables, due to a predicted scarcity that never came to fruition.
An analogy can be made between the above and processes in the digital world.
In addition, they are unable to specify the exact data that will be of use to them in the future. This leads to them deciding to unnecessarily hoard (or ‘panic buy’) all the data that they possibly can.
As time goes by, the data hoard stash continues to expand and the cost of storing and managing it increases exponentially. When an organisation has eventually gained enough knowledge to start working with data-analytics and machine learning, they will unfortunately come to the realisation that the data volume has become too complex, even too outdated to be used for anything effective.
In conclusion; digital hoarding is a poor long-term strategy for data driven digital transformation. No long-term opportunities are created by postponing the problem in this way. On the contrary, costs will increase and it will be harder to extract value from the data available. Furthermore, it becomes very tempting to blag an unprecedented result from the collected data, rather than thinking from the needs of the user and asking: “What do we want to achieve?”.
Each individual business requires a unique set of information for optimal operations. Take digital giants Google, Amazon and Facebook, for example. All have very different business models: The data focus at Google is on search behaviour, whereas at Amazon it’s on purchase behaviour and at Facebook on social behaviour. These differences influence how each individual business determines their data strategy and pinpoints the relevant information within it.
Admittedly, the aforementioned organisations possess the luxury of extensive knowledge and resources. They are able to collate as much data as possible and obtain valuable insight from it; not something the majority of businesses can boast. However, what all businesses do have in common with these organisations is the need for making decisions. They start by formulating a digital strategy. After determining the objective, the appropriate analysis type must be specified. Only then can they select from the correct data.
For now we are assuming the company has decided on the right goal.
The first factor to consider is the appropriate form of data analysis and the necessary depth of information needed.
Below are the four primary types of data analysis, ranging from the simple to the more complex. Generally, the value increases with the complexity of the analysis. Whether complex insights can also be profitable, strongly depends on the organisation and their business model.
1. DescriptiveIn short, descriptive analysis is a look into the past. The simplest form of analysis uses diverse historical data sources to answer the question, ‘what happened?’. Use it to see how much turnover a specific product has delivered, to determine its historical trend, or, what share of total turnover it has represented within the entire range, over a certain period of time.This traditional form of analysis definitely has worth, but results from the past don’t offer any guarantees for the future.
2. DiagnosticCombines the historical data with other (external) data to find out why something has happened. With diagnostic analysis you can, for example, discover why certain sales targets have been missed, despite investing in marketing. For diagnostic analysis a combination of internal and external (for example Linkedin or Google) data sources are used.As diagnostic analysis’ goal is to explain past events, it provides interesting insights into factors that must be taken into account for the future, in order to grow as an organisation.
3. PredictiveThis analysis is used to determine possible outcomes. It combines the results and trends derived from descriptive analysis data, and the information taken from diagnostic analysis and uses it to predict future trends.This form of analysis is very useful for forecasting. It has recently become more powerful due to advancements in techniques like machine learning. However, it must be said that these predictions are always estimates. First of all, they assume a stable market environment. Secondly, the quality of the prediction strongly depends on the quality of the data, and how complete the insight into influencing factors is.
4. NormativeThis form of analysis focuses on making recommendations. It offers insight into which measures could be taken in certain situations. Normative analysis can be seen as a more accurate version of predictive analysis, whereby a number of options have been pre-calculated. It allows you to, within predetermined parameters, select the best available option. Normative analysis can be used, for example, to prevent future problems in business operations, or to take full advantage of a particular market situation.
In the first case, you use operational company data and in the second, the emphasis is on customer data and sales transactions. Normative analysis is an important means of optimising business operations for efficiency and growth.
Now you’ve decided on the best analysis type for your situation.
The next step is to determine exactly which data you need to base your research on.
This can be complex. Take data hoarders for example, their data is widely available but not in the correct context. It’s like trying to automatically count cars in a car park: building weight sensors in the ground is much easier than developing a machine learning algorithm to define objects as cars, and then release it to the archives of camera images of the entrance.
This may seem obvious, but recent research shows that data scientists spend up to 45% of their time cleaning data before it can even be analysed at all! A huge waste; when these specialists could be using it to find the answers to really complex questions.
A reason for this is as part of their digital transformation, data-hoarding organisations look for value in their motley collection of unstructured data. Such as video, audio and image files or the content of emails, help desk interactions or social media posts. Although relevant insights can be deduced from this, the question is often whether these are also insights that are in line with the digital strategy, or have any relevance at all.
Which data should you keep and which should you dismiss? If you recognise the value of data in general, it would be extremely counterproductive to make your decisions using a gut feeling. If you have doubts as to whether certain data is or will be relevant, get rid of it: Use the time and resources that you save as a result to invest in an effective strategy. Resulting in an outcome that is certain and, in the long term of greater added value.
This article was previously published on Consultancy.nl, written by:
Founder & CEO
We support The Plastic Soup Foundation
Sumatrakade 91019 BJ AmsterdamThe Netherlands
4-5 Bonhill St, ShoreditchEC2A 4BX LondonUnited Kingdom