More and more businesses are taking the decision to put data at the heart of what they do. But what does it mean to be data driven? How is that any different from being Data informed or any of the other buzzwords we hear so often? What do data centric businesses do that the rest don’t? And once you’ve made the decision to become data literate, how do you get everyone else on board? It’s a confusing and fast-moving world where “Data” prefixed terms pop up left, right and centre, but you don’t want to be left behind. So, let’s slow it down and find out what’s going on.
Before we go any further it would be best to define what these various terms mean and how they differ from each other. They are all too often confused but, when laying out your data strategy, it’s important that you’re clear.
Data Driven – Data, statistics and evidence underpin decision making at all levels, overriding human intuition due to the high level of trust in the data, calculations and algorithms. This can save vast amounts of time but relies on clean, accurate and timely data that can be difficult to achieve.
Data Informed – Data, statistics and evidence are utilised in discussion and debate at all levels. This does not completely disregard the human in decision making; more that businesses take an augmented approach that merges human intuition with data insights. Sometimes known as Data-led (I do not like this term as it suggests that data does the driving).
Data Centric – Data architecture is considered above other business systems such as ERP and CRM. Data architecture is protected so that it will outlive other business systems.
Data Democratic – All data is made available, ideally to everyone. All business systems should make their data accessible and staff are encouraged to utilise it. Often, all data is pooled into a data warehouse or data lake *.
Data Literate – The ability for humans to handle, interpret and argue with data. Helps to make informed decisions and share insights with others. Assists with all the above so you can choose which paths to take.
* More data buzzwords! Both are approaches to storing data, with a lake containing unstructured, often raw, data from multiple sources and a warehouse containing structured, transformed data. As data grows and becomes increasingly difficult to manage more companies are migrating to data lakes.
Which One Should I be?
As you might have guessed, there is no single best practice, though, data literacy is certainly the starting point which will help inform your decision on where to go next. Let’s look at the advantages and disadvantages of each.
Advantages – In a nutshell – speed and clarity. When you trust your data and let it drive decision making you take out the hesitancy native to human reasoning. It might start by trusting data over human intuition when faced with a report but for more mature businesses it can go as far as automated decision-making using AI and machine learning which frees up time for humans to get on with other tasks. Example
Disadvantages – It can be a huge and daunting cultural leap to take. It relies on highly cleansed, timely data so investment in data architecture and staff are a must. For those looking into AI and machine learning, data scientists need to be employed. Above all, it removes the advantages that human intuition can bring, such as spotting trends before they play themselves out in the data.
Advantages – Decisions that consider human knowledge and intuition combined with data are well balanced and rounded. Each can be used as checks against the other to guarantee accuracy and uncover insight that might have otherwise been missed.
Disadvantages – Often results in a slower decision-making process than using either data or human intuition on their own because the two need to be compared. Relies on clean, accurate and timely data so that the data it can be trusted and a good level of data literacy to interpret it.
Advantages – Whilst ERP, CRM and other applications are ephemeral, data is here to stay. The more you keep, the more history you have to inform trends and forecasts. Therefore, putting data architecture above all other architecture protects this history. It’s longevity also means that connections to it are well established. Add to it Master Data Management (MDM) and you have a well governed, easy to understand set of data with no silos.
Disadvantages – Can be a big and long project; cataloguing all your data sources and owners, setting up data flows, additional security and mapping business terms/aliases. There’s also the cost of additional data storage (including staging areas).
Advantages – With access to more data sources, you can compare data to make for richer insight – e.g. comparing footfall, store sales and weather data to find that your store performs best when it’s raining, and people take shelter in your store. Staff who consume data don’t have to wait weeks for IT to approve access to different data silos.
Disadvantages – Unless coupled with a data centric approach it can be difficult to maintain connections to all the different data sources. Relies on having a data literate audience who won’t be overwhelmed by the amount of data. Less secure, as consumers have access to all, potentially sensitive, data.
Advantages – Staff are empowered to make balanced decisions by augmenting their own intuition with data. They can cut through the noise and “fake news” to spot flaws in data and they can construct compelling arguments with the evidence they build from their data.
Disadvantages – Staff may need to be trained and the culture change required may be considered with suspicion by so called “data doubters.”
When drawing up a data strategy it would be wise to consider each of these 5 buzzwords and see how far you could take them. They don’t have to contrast, rather, you can pick elements from each to build a rounded, flexible strategy. Some of these elements will be a ground-up approach with analysts and developers leading the change whilst others will be top-down with C-level staff inspiring others to bring about change. A typical strategy might combine them in the following way:
- Data Literacy training to improve staff confidence with data
- Staff encouraged to use data in meetings to make data informed decisions – reviewing reports and dashboards.
- Staff who are comfortable with data are given access to more of it under a governed data democracy.
- As the demand for data grows, data is pooled and standardised in a data centric architecture to make it easier to access and interpret. Silos are brought in slowly, one-by-one.
- For a small number of specific business areas data driven decision making is exercised to enable rapid decision making. g. in a call centre certain staff are automatically assigned back office work whenever the call queue drops below 2 and assigned calls when it’s over 5.
So, which one should you be? Why not be all of them?
NB: For more on culture change and data literacy check out my interview with Jordan Morrow – Global Head of Data Literacy at Qlik.