There are many influencing factors that ultimately end up playing a part in the cost optimisation and value creation associated with AI applications. It is absolutely possible to reduce costs and create value with AI applications even in the short term, but it’s also easy to veer off the track of ROI if the wrong business case is selected, staff skills are not in place, or models are not built for scale.
Even if all these elements are in place, delivering ROI with AI can still be a somewhat daunting task for companies in the earlier stage of AI adoption. In AI ROI, the difference between followers and leaders is stark, compounded by the fact that so few companies can call themselves leaders, and most companies are still at relatively early stages in their AI journeys. Yet the fact remains that executives are highly conscious of AI’s part in business growth amongst the competition.
According to Accenture’s AI: Built to Scale study, 88% of executives in the United Kingdom polled said they won’t achieve their growth objectives without scaling AI. In addition to this, in the same study, 84% of executives believed they risked going out of business in five years if they didn’t scale AI, and 84% of executives acknowledged that they knew how to pilot, but struggled to scale AI across the business.
This ability to scale is what is defining the big difference between AI followers and leaders, and at the same time, driving ROI in AI investment. How can followers and those that are still advancing do in order to scale and to increase ROI?
Understanding the AI Maturity Journey
Every journey has to start somewhere, and many companies may start their AI journeys with short-term goals. For example, insurance companies may implement AI to automate claims processes, or to reduce fraud. Other initial goals are often centred around increasing staff and customer engagement, or other ‘low hanging fruit’ projects that may not necessarily be scientifically complex, but can end up saving businesses hundreds of millions of dollars. It’s about establishing tangible value from a few initial use cases and laying the foundations to scale.
However, to continue growing ROI, companies need longer-term goals as well. This may be AI-driven longevity, or better managing cash solvency or liquidity. The key is mixing the quicker, low-hanging fruit use cases with longer-term wins, expanding usage of AI across the organisation to spread to all departments and functions.
Establishing Longer Term AI Business Use Cases
We’ve talked about the low hanging fruit and what to do in order to make sure it runs smoothly, but once that has been conquered, it’s time to think about progress. Recent research from ESIThoughtLab called Driving ROI through AI showed that AI leaders have adopted AI across large parts of their enterprises. These leaders have no doubt got their data requirements sorted, and are probably experts in operationalisation as well. However, it’s also likely that they have put a lot of research and thought into the strategic business use cases they have selected.
As the research says, as a total group including all followers and leaders, the companies surveyed had made the most progress in implementing AI for customer service and IoT, followed by IT operations, customer analysis and data security.
Firms overall had made the least progress on applying AI to distribution and logistics, finance and auditing, supply chain, R&D innovation, sales and business development, risk management and fraud detection.
Yet here’s the thing: the research also indicated that it’s these areas where the leaders are ahead, and showing the most ROI from AI. So, while every industry is unique with its own use cases, those specific to the business’s operations are likely to become the most important, and may offer the most AI-driven ROI.
Get a Business Translator for all Your Data Needs
Use cases are nothing without data. Data sourcing, simplifying and improving aspects of data quality, data labeling, and connectivity: there’s a lot to do when it comes to getting and preparing the right data.
Many data leaders are beginning to embed “business translators” into their organisations to translate business needs into data needs to make AI pervasive. According to Gartner, “this could be a business-savvy data scientist or citizen data scientist, an analytically minded business person or a process engineer (process modelers or business analysts focused on process design) who is mindful of business optimization opportunities derived from analytical assets.”
What’s the benefit, specifically? Business translators can serve to identify data requirements, oversee data workstreams, and act as mediators and points of contact for members of the development and operationalisation teams as well as to executive stakeholders. It’s this type of capacity that can begin to transform a follower company into a leader in AI, building in more agility into the transformation and delivery capacity required for large-scale data projects.
Operationalise with MLOps
By providing frameworks for moving models into production, data science tools can reduce costs associated with model maintenance and monitoring. MLOps, the standardisation and streamlining of machine learning and lifecycle management, takes the concept of operationalisation up a level on the AI maturity model.
Good MLOps practice will be a critical component to scaling machine learning efforts, and in helping companies go from one or a handful of models in production to tens, hundreds, or even thousands.
MLOps best practices will enable teams to keep track of versioning, understand if retrained models are better than the previous versions (and promote models to production that have more optimal performance.) This best practice also ensures that model performance will not degrade in production.
As firms continue their AI journeys, it’s important to remember that AI maturity is also about people as well: it’s essential that AI is supported by a multi-person team of analysts, business makers, IT – people who have different perspectives on the business to use as a starting and constant reflection point.
Once firms determine the business problems they are trying to solve, they need to move on to track the value generated in order to truly test both AI-driven ROI and maturity.
Learn more about Dataiku here: www.dataiku.com