Embracing data is not enough. You need to be driving wholesale adoption of data science across the business

The current state of organisational data science

Today, companies all around the world are expending huge resources on the collection and analysis of data to drive and scale data-driven decision making throughout the business. There is a massive increase in the amount of data made available to organisations, and huge upside for those companies that learn to apply knowledge gleaned from this data towards greater business value.

However, many are struggling to actually leverage these opportunities and achieve mass adoption of data science and analytics in order to fundamentally change the way they operate across the business.

There are various reasons for this stagnation:

• A misunderstanding about the value data science can (and should) provide
  We have already spoken at length about the value a data team brings to an organisation and how to leverage this value. Problems arise when organisations wish to jump aboard the data train, employing teams of data engineers, scientists and analysts with virtually no idea about the processes they need to have in place first in order to fully take advantage of the value these teams provide. This is a strategic mistake - pouring resources into data aggregation and new teams without yet knowing why or even how to apply the analyses and insights generated about the business.
• Lacking in data science infrastructure
  This leads us to to our next problem, which is the (lack of) sufficient data infrastructures within an organisation prior to employing these data scientists and data teams. Is the business already collecting and aggregating data for use? If so, has the data been transformed into a useful format for analysis by the data engineering team? Do the data scientists have tools available to them for publishing and sharing their insights with the relevant stakeholders? The stage at which a business is in its data pipeline - from initial data collection, all the way to predictive analytics and reporting - will greatly impact both the type of work data teams can carry out and the effect said work will have on daily decision-making.
  
• Fragmented data science efforts
  Due to these issues and, in many cases, a lack of direction from middle management and the higher-ups, companies continue to have several fragmented data science efforts going on across the business. Individual departments and pods take it upon themselves to drive new projects and pilot new tools as and when needed. This piecemeal approach leads to disconnected learning from these various initiatives and a general desynchronisation of organisational data science.

In the next section, we will go through some of the steps we at Kyso think organisations should take in order to fully reap the benefits of data-driven approaches and business-wide adoption of data insights in decision-making.

Steps to becoming a data-powered business

1. Understanding the business problem space

‍
The data team needs to first and foremost have a good understanding of the problems the business faces, before they delivering possible solutions. For this to happen, communication channels need to be opened up between the data team and the other stakeholders, so these actors can help the data scientists understand the most impactful projects to work on, to ask the right questions of the data, and ultimately to solve the most important business challenges. By improving communication between data scientists and stakeholders, better outcomes for all involved will be the result.

2. Aligning goals between different stakeholders

‍
Due to the disconnect between non-technical stakeholders and the data team, expectations about the outcomes of data-based projects and initiatives tend to differ between the two sides of the business. A lot of data science and analytics projects don't deliver value to the business, and oft the problem is not the quality of the insights generated, but rather the focus of the analysis in the first place, because the data team is lacking direction from project leads and other stakeholders across the various departments

However, as domain experts, business agents - the non-technical stakeholders - should be able to work closely with the data team in order to ensure that data-based projects are focused on business outcomes. In this way, the data team knows what is expected of them, and there is a more effective feedback loop from the other side of the business. Aligning their projects with the individual departments' objectives also allows the data team to influence decisions made around the business at all levels of operation.

3. Democratising data insights across the organisation

‍
Ok, so your business has successfully opened up communication channels and the data team's projects are business-orientated . Now they need to ensure their results are shared with the relevant stakeholders. If the ultimate goal of collecting data, modelling, and making predictions is to help everyone to contribute to organisational business objectives, then everyone needs access to this knowledge to make more informed data-driven decisions.

So, how does your data team publish their results? Typically, the answer is they don't. They share insights upon request by email or over Slack. There are two problems with this type of results-sharing:

                  1. These insights become siloed within different sub-groups, undiscovered by other, potentially relevant, business agents.

                   2. These results will likely only ever be used once, meaning analyses that COULD have repeat value are generated over and over again upon request,                        as they become lost to the archive history, never to be discovered again.

            Encouraging wholesale publishing of results is important because such a system prioritises discovery over sharing. Why? Because knowledge becomes             consolidated and transparent to everyone across the business. Because it empowers all employees to apply data insights in their daily decision-making.             And because it encourages a more data-driven culture. Allowing people to discover your data team's work organically is really really important - there are             huge benefits for different types of stakeholders across the business having the ability to discover these results and use them in their own work.
‍

‍4. Publishing actionable intelligence

‍
We've discussed before the optimal way to write up data-based reports, in which we look at several ways to best educate stakeholders on analytical findings and how best to formulate understandable business recommendations from those results, which shouldn't be ambiguous.

Results are shared in order to turn insights into action all around the business. Knowledge is what drives business value, and data science is the process through which this knowledge is created. However, the models used and outputs created by data scientists are often quite complex, and these reports need to be digestible for non-technical audiences - those that will be applying the data insights to their respective roles.

Each project's potential and estimated impact on any given aspect of the business needs to be structured in such a way that employees around the organisation view the work of the data team as pertinent to their own business objectives, and not simply as interesting learning opportunities. It is in this way that data science will be become the driving factor in the company's future successes.
‍

5. Driving lasting collaboration

‍
Organisational data science is not simply a one-time effort, but rather iterative, whereby the data team should be continuously looking to experiment and derive feedback from the non-technical side of the business in order to improve the impact of their endeavours over the long term.

Ensuring that data-science initiatives involve input from business agents, closing the gap between technical and non-technical teams, and democratising access to data-based knowledge will allow your business to fully leverage the value created from its own data-science initiatives.

We dove into these ideas in more detail in a recent article on the impact of empowerment loops in organisational data science - the idea being that bringing everyone into the discussion empowers all employees with the required knowledge that will ultimately empower them to carry out their jobs effectively and enable the business to stay ahead of the curve.

Future success will depend on wholesale adoption

The above are just some of the steps that we at Kyso will help organisations derive the most value possible, both from their data teams and from everyone else around the business. A lot of the points we have made may sound intuitive to some of those reading this article. However, the fact remains that many organisations today continue to overlook the negative outcomes of their unstructured data science efforts.

We believe that achieving data-driven success requires a centralised, top-down effort, with input and collaboration from all aspects of the business. Success breeds success, and as more and more organisations improve upon and build out their data strategies, competitive advantages will continue to accelerate, meaning those organisations not keeping pace will simply be left behind.

‍