According to a recent Boston Consulting Group (BCG) report, while there is much hype around artificial intelligence (AI), the value is hard to find. Based on recent research involving more than 1,000 companies worldwide, only 22% of companies have advanced beyond the proof-of-concept stage to generate some value, and only 4% are creating substantial value.

The Massachusetts Institute of Technology echoed this concern and wrote, “For all the talk about artificial intelligence upending the world, its economic effects remain uncertain. There is massive investment in AI but little clarity about what it will produce.”

Why is it so hard for companies to measure the return on investment (ROI) of AI? There are simply too many pilots and proof of concept projects. Even when there is a more ambitious program, there is frequently a lack of well-defined goals for such AI initiatives. Then poor data quality and challenges with gathering detailed dataon desired outcomes compound the difficulty of calculating ROI. Then the difficulty of integrating AI initiatives with existing IT systems adds complexity to the mix.

In their research, BCG identified six characteristics[i] of leading AI companies:

·      They are more ambitious.

·      They invest strategically in a few high priority oppor­tunities to scale and maximize AI’s value.

·      They focus on the core business processes as well as support functions.

·      They integrate AI in efforts both to lower costs and to generate revenue.

·      They direct their efforts more toward people and processes than toward technology and algorithms.

·      They have moved quickly to focus on Generative AI.

To successfully build such characteristics, organizations are advised to consider the following tactics:

·      Shift management attention to value creating processes

·      Deploy both Generative AI and Analytical AI for maximum impact

·      Focus on data

·      Measure what matters to customers

Shift Management Attention To Value Creating Processes

Shifting management attention from a vertical, departmental view of the organization to an “outside-in” horizontal view is very much needed for success with AI. Viewing end-to-end business processes at the enterprise level – from both the customer’s and the company’s point of view – is needed for success. That requires the senior leadership team to have pictures of how the organization develops, makes, and delivers its products and services. In other words, high level, easy to describe process maps. By so doing, leadership teams can gain a shared understanding of a big picture view of the business which is needed for a more ambitious agenda and strategic investments in a few high priority projects.

For example, as Dr. Robert Cooper has outlined, a focus on improving new product development with AI requires both a process and a product view with balance on what matters to customers and the company. Similar principles apply to other core processes such as order to delivery and inquiry to resolution.

Deploy Both Generative AI And Analytical AI For Maximum Impact

While Generative Ai has huge potential, much of the progress with AI in improving customer experience has come from predictive analytics, now increasingly called analytical AI. AI can be deployed to analyze customer data and provide organizations with a 360-degree view of the customer, including past interactions and individual preferences. This allows companies to provide personalized recommendations and send targeted communications. [ii]

IBM has worked with both Starbucks and Amazon in developing AI enabled systems to improve customer experience. It’s Deep Brew initiative uses AI to elevate the coffee business and in-store customer experience. The tool uses machine learning and predictive analytics to personalize marketing messaging, which drives retention and hopefully increases loyalty.

Amazon’s AI system has revolutionized e-commerce shopping by making personalized product recommendations to customers. The data-driven system analyzes customer behavior, purchase history, cart history and more to understand what purchasing behavior.

Ther is evidence that evidence that analytical and generative AI are “better together” and organizations are well advised to apply each to their respective strengths.[iii]

Focus On Data

AI's effectiveness depends heavily on the quality of data and integration with existing systems. Consistent and standardized data reduces inconsistencies that can degrade AI performance. Therefore, it’s important to define clear standards for data formats, and appoint data stewards and/or teams responsible for maintaining data quality. Also, it’s necessary to regularly evaluate datasets for accuracy, completeness, consistency, and timeliness.

As high-quality data ensures that models receive reliable inputs, identifying and eliminating duplicate, or erroneous entries in the dataset is needed. Since AI models require ongoing access to fresh, accurate data to remain relevant and effective in rapidly changing environments, real-world feedback can be used to identify and fix data quality issues.[iv]

Measure what matters to customers

Specific, measurable goals for AI initiatives are needed to accurately measure the ROI on projects. Measuring what matters to customers, be they external or internal, is a fundamental first step. The best set of such performance measures is a balanced view across quality, timeliness, cost, and volume. Customer journey maps depict customer touchpoints such as website visits, key interactions, and support calls as well as bottlenecks or inefficiencies in the customer journey and the emotional states of customers at different stages. Customer journey maps can be an effective means to identify customer needs, pain points, and opportunities for improvement. [v] 

 Summary

The importance of large volumes of high-quality data to optimize the performance of AI is well known. The need to shift management attention, the importance of measuring what matters to customers, and the need to better understand both analytical and generative AI are perhaps less well known. Yet, taking action on all four of these is needed to improve the ROI performance of AI.

[i] Boston Consulting Group. Where’s the Value in AI? Boston Consulting Group, October 2024.

[ii] Watkin, Rick. "Machine Learning Meets Customer Experience: How AI Is Reshaping CX." Forbes, February 13, 2024.

[iii] Davenport, Thomas H., and Peter High. "How Gen AI and Analytical AI Differ — and When to Use Each." Harvard Business Review, December 13, 2024.

[iv] ChatGPT was used to generate and/or verify key points in this section

[v] Ibid

Andrew Spanyi is President of Spanyi International. He is a member of the Board of Advisors at the Association of Business Process Professionals and has been an instructor at the BPM Institute. He is also a member of the Cognitive World Think Tank on enterprise AI.