Six key components that are important to ‘productionise’ analytics

Opinie

Six key components that are important to ‘productionise’ analytics

Embedding analytics and delivering business value

Developing insights or an analytics solution like a dashboard or predictive model does not suffice when an organisation wants to transform into an Insight Driven Organisation. Although having the data and technology in place and being able to deliver the analytics in a repeatable and consistent way is important, organisations should also work on strategy, people and processes to assure that analytics delivers value to the business.

Johan van der Veen & Bas Schmidt & Ingrid Lanting - 4 december 2017

Perhaps you recognise one of these cases:

Predictive model
Frank is a manager in a private sector firm. He has recently sponsored the development of a predictive model which is able to support in selecting the best new market segments. Although the model proved its predictive ability, still a lot of work needs to be done to start using the analytics solution. Until now, decisions have been made on gut-feeling, so a mind shift and cultural change will be required. Also, the analytics solution needs to generate insights periodically and therefore needs governance. Analytics has been an unexplored area for the organisation until now and Frank does not know where to start.

Dashboards
Alison is a strategy director for a public organisation where multiple dashboards are used. Well… ‘used’? The dashboards are being used less and less, while the intention is to make more decisions based on facts. Conversations with employees have shown that some dashboards do not comply with the needs and wishes of the users; for example, loading time is very long and the data is not up-to-date and in some cases even wrong. Other dashboards were not trusted anymore because of this so they simply stopped using them. Alison wants to change the culture of the organisation, but how?

In this article, we explain six key components that are important to ‘productionise’ analytics. With productionising analytics we mean embedding analytics solutions into the organisation so that they are used, improved and eventually, deliver business value. The key components that we will explain are Business process reengineering, Ready-to-use insights, Governance, Adopting analytics, Benefits Tracking, and Continuous Improvement.

This article is part of the IDO-blog series. In this specific article we will focus on the requirements for productionising analytics, assuming that the correct analytics solution or POC has already been developed. We will not cover the required steps to become an Insight Driven Organisation. We will focus on delivering business value out of that one example analytics solution.

1. Business process re-engineering

Even the best initiatives fail if the business users are not able to use them in his or her daily routine. The first key component of productionising analytics is therefore to redefine the business process which incorporates the application when making insight-driven decisions.

Redefining the business process starts with measuring the impact of the analytics solution on the current state. Determining the current state requires a definition of the processes in scope, employees and responsibilities, technology and information requirements [1]. For each of these components, the impact of the analytics solution should be determined.

Take the example of Frank, his newly developed model which helps to select the best new market segments will impact the current market segment selection process and thus how people work that are responsible for this process. In the current state, there is a team responsible for making the decisions, there are people in this team who are making these decisions, there is a process (defined or undefined) for market segment selection and there might already be some technology or information that support the current decision making. The predictive model will impact many elements, details of the impact should be made visible and should be validated as soon as possible during the project. Will the same team remain responsible, or does the analytics team also get the responsibility for maintaining the model? What competencies and training are required, is this also the responsibility of the business? How will the new step-by-step process for choosing the best market segment look like? How do we assure the Business As Usual is not/minimally impacted?

Next step is to create a sketch of how the impacted components would look like in the desired state, in which the insights from the analytics solutions are used to improve or automate decision making. Business process re-engineering is used as a tool for this. With business process re-engineering,

"companies start with a blank sheet of paper and rethink existing processes to deliver more value to the customer. … Companies reduce organisational layers and eliminate unproductive activities … they use technology to improve data dissemination and decision making.”  [2]

All activities for business process re-engineering should be verified with the business (or even be co-created with the business) and all results should be accepted by the business. Selecting a ‘product owner’ or key-user / champion in the business helps in reducing the number of stakeholders. 

2. Ready-to-use insights

Everyone has seen examples of, or even have been frustrated with, new business technology or a new tool that was too complicated to use. This could be the result of a lack of user guidelines, too much effort required to use it in daily practice, not seeing the benefits of the solution, etc. Analytics solutions also fall in the technology innovation corner and therefore also require a careful consideration of user experience.

For analytics solutions, as well as for all other innovations within technology, it is key to center the design of a solution to the requirements of the relevant consumer. Design Thinking is the field that is concerned with this methodology.

“The use of this approach in digital transformation involves understanding the specific needs of customers and even employees, then making sure a company’s offerings and tools address those needs”. [3]

Three elements of Design Thinking are crucial to make the insight ready-to-use: simplicity, sufficiency and reliability.

But what makes an insight ready-to-use? The following components explain the requirements: getting the insight is as simple as possible, understanding the insight does not require an extensive manual (and rather speaks for itself), the insight gives sufficient information for making the decision and the insight is reliable. Let’s zoom in on these requirements which make the insight ready to use.

The required insight should be only one or two clicks away, which can be achieved by embedding the insights in the workflow management system. However, this is often an expensive solution. Current technology developments (cloud platforms and APIs) make it easier to integrate workflow systems with analytics solutions [4].  Another way is to deliver the insight via a web portal using a third party tool or again via a cloud platform. The advantage of the latter is that no new server needs to be installed, everything is available to use and can easily be scaled.

Having quick access to a tool that shows the insight is one, having the required insight directly available in that tool is second. Big data ensures a high volume of data and therefore, it is very hard to get exactly the right data analysed and visualised in the correct way so that it can be understood directly [5].  Thinking about which data should be shown in which format is essential. Storytelling with data helps in centering the story that needs to be told to the consumer. With storytelling, the data is presented in a clear visualisation tool (such as Tableau or PowerPoint) so that the required insight can be obtained and understood by the consumer directly [6].

Finally, ready-to-use also indicates that the insight is always right. If there is only one case where the analytics solution gives a wrong figure, this might already lead to a lack of trust in the whole solution. This makes it much harder for consumers to keep making decisions based on the solution without verifying the numbers.

3. Governance

We like to take the example of a tree; retrieving fruit requires the tree to have solid roots. These roots are underground and are not visible most of the time. When something is wrong with the roots, it will reflect on the tree and its fruits. It is the same with analytics: successful analytics solutions require a good balance between analytics solutions and governance. Analytics solutions cannot be successful without having certain prerequisites being fulfilled underground. Our Insight Driven Organisation model translates this into five main roots that are required: Strategy, People, Process, Data, and Technology [7].

When determining the organisational strategy, it is necessary to pay explicit attention to analytical solutions: what role can they play in achieving our business goals? The answer to this question must then be reviewed periodically. This will continually legitimize the existence of the analytical solutions, and the requirements they meet must be determined again. The requirements for proper functioning are a stable and up-to-date IT environment, a properly managed data-chain - from source to user and profiles that have the proper skills. Only then can the use of analytical solutions drive competitive advantage, separating organisations that treat analytics as a collection of good intentions from those that industrialise it by committing to disciplined, deliberate platforms, governance, and delivery models [8] [9].

For the proper functioning of analytics solutions, it is also necessary that all processes are well-designed. The IT environment must be stable and up-to-date, including those which relate to management and (through) the development of the solution - think of the periodic refresh of data and developing new modules - and of course, the using process is also of great importance. This gives end-users access to the solutions, they know how the solutions work and get support when needed.

If the input of an analytical solution is incorrect, then the output will also be incorrect. Therefore, it is important to properly manage master and metadata management. This means in concrete terms that someone is taking measures to monitor the entire chain - from source to user.

Technical maintenance is essential for an analytical solution. Technological developments allow new applications and, in addition, monitoring the use of the solution can lead to adaptation. Also, good monitoring will show which solutions are not being adopted or where usage drops because relevance is lost.

The design, development but also use of analytical solutions, requires a different skillset. It is useful to identify at an early stage which profiles are present in the organisation and if previously mentioned processes can be guaranteed by these profiles. People should be trained and gain experience in their new role.

"Data is the only resource we’re getting more of every day,” said Christina Ho, deputy assistant secretary at the US Department of the Treasury recently.

Properly managed, it can drive competitive advantage, separating organisations that treat analytics as a collection of good intentions from those that industrialise it by committing to disciplined, deliberate platforms, governance, and delivery models [10] [11].

4. Adopting analytics

The next key component of productionising analytics is changing the mindset of consumers so that they start adopting analytics as part of their job. Next to the previously mentioned prerequisites for this, it is as important to manage the change with the insight consumers. A next article in this same IDO-blog series will go into more detail on how to realise this mind shift.

5. Benefits tracking and continuous improvement

We have explained that analytics solutions require maintenance and usage tracking. To assure that the expected business value of the analytics solution is reached, the business value should be clearly described and ensured that it is aligned with the business strategy. A next article in this same IDO-blog series will explain how value and benefits can be defined, periodically measured and optimised.

With a clear definition, the impact can be measured periodically. A simple framework or tool helps in a structured approach for measurement and monitoring. Having the business owner validating the value keeps the business aligned. To optimize the maintenance of solutions, discussions with the business should be started on shutting down the solution when it's not delivering any value anymore.

Continuously collecting ideas for improving the analytics solutions helps during adoption and in maximizing the benefits from the solution. Ideas for improvement can be collected for example by periodically sitting together with the consumers, opening a (digital) ‘idea box’ and by monitoring the data analytics solutions quality and efficiency.

Bottom line

Organisations struggle with acquiring the benefits from analytics solutions and because of that, analytics is often seen as a cost center or as a fancy toy of the managers. When embedded into production in a structured way, analytics solutions will definitely deliver the expected value for your organisation. Productionising analytics solutions require at least the key components as explained in this article. Deloitte has experience in supporting organisations to realise benefits by productising analytics solutions, both globally as in the Netherlands. Feel free to contact us if you want to know more about this topic.

1Components based on confidential change impact assessment approach, developed by Deloitte for a Dutch governmental institute

2Meena Jha, Combining big data analytics with business process using reengineering, available from: https://www.researchgate.net/publication/306925777_Combining_big_data_analytics_with_business_process_using_reengineering [accessed Jun 16, 2017].

3Using Design Thinking to Innovate, http://deloitte.wsj.com/cmo/2017/05/24/using-design-thinking-to-innovate/

4Five technology trends that leap-frog Artificial Intelligence, Deloitte NL https://www2.deloitte.com/nl/nl/pages/deloitte-analytics/articles/part-4-five-technology-trends-that-leap-frog-artificial-intelligence.html

5Dark analytics: Illuminating opportunities hidden within unstructured data, DU press, https://dupress.deloitte.com/dup-us-en/focus/tech-trends/2017/dark-data-analyzing-unstructured-data.html

6Why data storytelling is so important—and why we’re so bad at it, DU press, https://dupress.deloitte.com/dup-us-en/topics/analytics/data-driven-storytelling.html

7What is an IDO?, https://www2.deloitte.com/us/en/pages/deloitte-analytics/solutions/insight-driven-organization.html

8Krachtige analytics oplossingen vragen om stevig fundament, https://www2.deloitte.com/nl/nl/pages/data-analytics/articles/krachtige-analyticsoplossingen-vragen-om-stevig-fundament.html?id=nl:2sm:3li:eng_da_bus:governance

9Industrialized analytics, https://dupress.deloitte.com/dup-us-en/focus/tech-trends/2016/data-assets-and-analytics.html

10Krachtige analytics oplossingen vragen om stevig fundament, https://www2.deloitte.com/nl/nl/pages/data-analytics/articles/krachtige-analyticsoplossingen-vragen-om-stevig-fundament.html?id=nl:2sm:3li:eng_da_bus:governance

11Industrialized analytics, https://dupress.deloitte.com/dup-us-en/focus/tech-trends/2016/data-assets-and-analytics.html

Do you want to know more about Insight Driven Organisations (IDO)?

Please contact Ron Bergers via +31 (0)88 288 8740

Vond u dit nuttig?

Gerelateerde onderwerpen