How to build a great Data & AI organisation Part 2: Aligning the business, strategy, and Tech-Stack to deliver
8/13/2024 | nmb@konfitech.com
Although we are amidst some economic headwinds and many focused on cost optimization, the data and AI organizations that are leading the market are frugal yet in a growth-oriented mindset. Improving innovation and lowering turnover times, such as time to market or improved products seems to be a main priority. While operating lean and fast with data and AI is just the norm.
The goal of the expansion and growth with data is to expand sales and service channels. As we have seen companies like Klarna do with great success in their service channel, saving 40 million US dollars by implementing an AI chatbot to their service channel. Which as a side-effect saves enormous amounts of time in terms of productivity when leveraging data and AI.
Then for most companies' data strategy and the corporate strategy needs to be aligned. However, what is differentiating those that are successfully implementing it versus the rest? They dug deeper than the rest, going into how it would look like to deliver on the various strategic initiatives in the business. Links such as:
- Establishing a single source of truth
- Democratizing data
The single source of truth is absolutely a priority for many companies, they focus on data management. Having a single source of data where the company can pull out various types of data, and then democratizing accordingly. Going through the exercise of considering what types of data from this source does the various business users need to get the insights today and for the future. This makes a centralized management of the data, with a one-to-many relationship in distribution. While distributed data storage makes both maintaining one source to pull data from and therefore harder to scale data and AI to the business users.
According to MIT in a study published, they investigated who are the best for master data management. Apparently at least half of the data services of those high performers surveyed ran their data infrastructure in the cloud. The performers had credited this due to a focus on the basics of data management; reduce duplication, ease of access, etc. However, more likely this came because of creating a cohesive data strategy. Thinking through in the mid term how this would look like per workload basis, and investigating if the next steps with AI and data would align with business requirements and outcomes. Looking for alignment in strategy, IT vendors, and creating an according implementation plan.
Much of their success has also come from developing great use-cases in production for Data Analytics and Machine Learning. Many companies want to skip the foundations and go straight into full machine learning without mastering the basics. The likelihood of there not being enough use cases is not there, but the stakeholder engagement in the business is difficult in these cases. With AI and data, there is not a lack of talent in the greater ecosystem, but bridging the gap between finding the right ones to deliver on business capabilities is hard.
This makes it generally hard to scale the use cases for ML in the business. The business users need to own it, as data scientists would be the builders of the algorithm, the output of it needs to be aligned and operationalized properly by the business. To do that one needs excellence in handover to the business, as gaining buy-in to the business would only be through displaying financial gains like ROI. Which can kickstart the development of an AI and data culture in the organization.
Make sure to also give us a follow on: