Thought Leadership

Data-driven vs data-led decisions

By May 11, 2021No Comments

The use of data to make business decisions isn’t a new concept. We’ve been doing it for a while. Today, businesses face the difficulty of choosing a data-driven or a data-led approach, which will result in better business decisions. Yolanden Moodley, [Senior Manager: Business Improvement] at Altron MS, says there is a place in different parts of most companies for both. He goes on to clarify, “A data-driven approach uses data to guide the decision-making process where there may not be 100% confidence levels to allow for automation to provide a fully automated or intelligent response. Data-led decisions are where data pull people as opposed to being pushed by the data.”

Moodley says, “Businesses may adopt a hybrid model where they shift between one and the other depending on where most suited.”

He cites the example of using an online search to find a restaurant close to your home. That’s a data-led decision. A data-driven decision would be where a tech firm sees an increase in call volumes and realises their user base lacks specific technology skills. Hence, it decides to deploy specific training. “Companies deploy both approaches for different applications; they can and may co-exist.”

He cites the example of learning and development within the business and its impact on data-led and -driven decisions. “Historically, business’s approach to learning and development has been broad-brushed, providing an overview of the specific learning and development area.

“However, with data as a key tool in the arsenal of learning and development departments, there has been a move toward a more personalised approach to training, with a focus on the needs of the delegate and the business outcome. In the past, delegates would attend a course that provided a broad overview of the topic at hand. Today, data enables us to identify the most common challenges that the individual is likely to encounter in the field and incorporate those into the training.”

This has also enabled a continuous approach to learning, where the learning continues after the initial training to include follow up courses on specific pain- or value-points that may arise upon course completion. “Learning is continuous as new challenges come to the fore,” says Moodley.

Previously, the success metric for learning and development was primarily subjective or one-dimensional, based on a competency-based assessment. Data-based learning has resulted in measurable insights into the effect of the training on business metrics. “Data can tie in the business metrics and determine things like how often we achieve our SLAs or how often we repair the same device, enabling us to tie that back to the classroom.”

XHEAD: Drivers to data-led and –driven decisions

Moodley goes on to list six key business drivers behind organisations using data analytics:

  • Accelerated operational excellence: with a data-based culture, more information is available to identify opportunities for optimisation within current operating models.
  • Access untapped revenue opportunities both within existing and new customers. Data-based decisions bring the articulated and unarticulated needs of your customers closer.
  • A customer experience game-changer. The ability to understand key-touch points in the customer’s experience in their interactions with your products and services, all the way from them buying your services, to the experience they feel when they consume, to the reasoning on why they are letting go. Looking at the opportunities within data-led drivers in this space, the power of influence comes to mind when complemented with AI-based recommendations which nudge customers on their next purchase.
  • Obtain excellent insights into employee experience to make critical decisions to improve the employee experience.
  • Speed to value outcomes in decision making.
  • An unbiased view – people, have beliefs about why an issue arises, but the data unveils the truth.

Making the shift to data-based decisions requires, first and foremost, buy-in from the top. “Business domain level experts such as supply chain, operations, finance and marketing need to be aware of the benefits and capabilities of using data analytics to make business decisions and solve their problems easily and effectively. Then the technology experts need to understand the business and work closely with the subject matter experts to ensure that they’re solving relevant business problems.”

A collaborative culture must be encouraged, and the shift enabled by the organisation and not just the IT department. IT must be integrated into the organisation and not viewed as an isolated entity. Sharing measures of success milestones and metrics organisation-wide to counter subjective responses to the process is vital. It’s also crucial to improve data literacy at all levels of the organisation and to personalise the value of analytics to consumers in their different roles. The outcomes must be relevant to the user’s needs in their business area to see how this will benefit them. Finally, target value-driven data sources – it’s impractical to process all available data using high-value data as a data source. The business value of the source becomes the qualifier for prioritisation.

Good data-led or –driven decisions require context, integrity and validation. “It’s important to have context around the information, such as data origin and capture meth. You also need to establish the data integrity levels and have visibility and awareness into the metrics of success or failure of decisions based on that data.”

“It’s important to allocate time to comparing predictions or data-led automated decisions against those made manually. Such decisions need to add business value, and that needs to be measurable.”

XHEAD: Pros and Cons

He goes on to discuss potential challenges and how to overcome these. “Buy-in from the top is critical. Data insights provide new layers of visibility, and the company culture needs to allow for these insights to come to the fore – as well as the appropriate actions to derive the desired outcomes.”

The technology designed with big picture thinking in mind – has to be designed to grow over time. Data will continue to grow, as will the number of users, while queries will become more complex. “Your design needs to be flexible, as do your technologies, to allow for that growth without continuous redesign.”

As mentioned previously, speed to business value matters. It’s fundamentally essential to translate the business value drivers into critical deliverables for the technologists. It’s effortless to take on tech projects that eat into man-hours but don’t deliver direct business value. “You need to simplify cumbersome technical processes to improve the speed to value. More often than not, if the information is not at hand when needed, by the time it’s built for consumption, the decision has been made, and it may be too late. It’s paramount that over time robotic process automation and intelligent responses take these insights to actions that will ultimately result in the desired outcome (value).”

Then there’s the issue of skills, which have become increasingly difficult to attain and retain, particularly when specialised, such as data scientists or data engineers with industry experience. There is no one-size-fits-all skillset in the world of data skills – building a successful team will require a combination of data skills.

It’s paramount to find a combination of technology expertise with subject matter expertise. Industry-specific expertise over and above qualification is vital, says Moodley.

He also advises being wary of the never-ending project. “You must have defined scopes upfront and have the end in mind when commencing with data-led projects. It’s advisable to develop piecemeal and then review as you go along, instead of only reviewing at the end, resulting in large additional scope agreements.”

Security considerations should always be at the forefront, with security teams enabling growth and expansion whilst meeting the necessary security requirements. “You need to secure the data without hindering your growth, progress and expansion.”

Finally, Moodley says data literacy for data consumers is a cornerstone to success. “This is a difficult metric to measure but can significantly reduce the challenge of developers having to figure out requirements, not to mention the many re-do’s that come with that. As literacy improves, people know what to request. Improving general data literacy also improves data consumption and decision-making.”