Implementing AI: The benefits, uses cases and measuring value

08 July 2024 6 min. read

AI is being presented by many as a panacea, an advancement to all kinds of matters. But truly getting value out of AI is easier said than done. Experts from AND Digital explain how the right use cases can be identified and what metrics should be used to safeguard a positive return on investment.

Measurable value should always be the end goal from any technology adoption, and artificial intelligence (AI) is no exception. The key to AI success ultimately lies in making sure the technology adds value, to the business, stakeholders and/or society as a whole.

If implemented successfully, the benefits of AI can be significant, or event ground-breaking, with a recent study finding the technology could add trillions in economic value annually to economies and sectors.

Implementing AI: The benefits, uses cases and measuring value

What is AI good at doing?

A key starting point in the AI journey is to pinpoint where AI truly adds value. The truth is that there are some areas where AI truly excels – and can outperform a human – and there are others where human intervention is definitely still required. To see value quickly, it’s wise to focus on AI’s strengths and consider how they can be applied within an organisation.

  • AI is great at doing boring tasks like data tagging, metadata tagging, and pattern recognition. Due to the boring nature of these tasks, there’s a higher propensity for error when humans are doing them, it also takes people a lot longer
  • AI is excellent at processing large amounts of data and providing insights quickly, it can free people up to focus their efforts on higher value activities
  • AI is fantastic at identifying patterns, applying it in the right context can yield insights and opportunities that may not have been possible otherwise

Consider the business challenges

Once an understanding of what AI does well is gained, leaders can consider the challenges that their business is facing. A strong use case is built on the alignment of technological capabilities and genuine organisational needs. After all, leaders don’t want to end up with a “solution in search of a problem”.

So how can leaders identify the right use cases for AI? Five key factors:

(1) Identifying the right problem
The first step is understanding what problems or opportunities exist in the organisation that AI could address. These could range from customer service enhancements, recommendation engines, machine learning models that predict outcomes or simulate real world scenarios.

(2) Feasibility and impact
Once a potential use case is identified, it’s essential to evaluate its feasibility and potential impact. This involves asking questions like: Can the problem be clearly defined? Is relevant data available to train AI models? What is the expected ROI? This helps in prioritising use cases that offer the most value.

(3) Integration and scalability
Consider how the proposed AI solution will integrate within existing systems. It’s important that the solution can scale and adapt as the business grows and changes. If an AI tool cannot integrate well with existing processes – or is too rigid – it might end up creating more problems than it solves.

(4) Ethical and legal considerations
AI deployments must also be reviewed thoroughly to identify any ethical or legal implications, especially when it comes to data usage and privacy. Ensuring compliance with regulations like GDPR is crucial in maintaining trust and integrity in the use of AI.

(5) Continuous learning and adaptation
Finally, successful AI implementation is not a one-time project but a continuous journey. AI systems need to be continuously trained and updated to adapt to new data and changing business environments. Organisations need to be prepared to invest in ongoing training and development of their AI systems, and ensure their people have the skills necessary to drive value.

Measuring the ROI from AI use cases

While the business value of AI can vary by use case, there are several metrics that can be used to project and monitor the value ahead of artificial intelligence initiatives:

Pre-implementation AI metrics:

  • Estimated AI model development costs – include the costs of data labelling, model training, testing, and more
  • Estimated ongoing maintenance costs – the projected costs to update an AI model over time as more data is collected
  • Time spent on current manual processes – Track hours specific roles spend on tasks that can be automated
  • Error/rework rates of current processes – track errors or rework generated from manual tasks like data entry
  • Customer satisfaction with current processes – conduct surveys to understand pain points customers are experiencing with current processes

Post-implementation AI metrics:

  • Actual AI model development and maintenance costs – track spend against estimates to determine the variance
  • Time savings from automated processes – track the hours saved by different job roles from reduced manual tasks
  • Increased productivity – Track # of claims/orders processed, tasks completed by roles per unit/time
  • Reduced error rates – track errors from automated tasks compared to manual performance as either a percentage or count
  • Improved first contact resolution – track if AI helps resolve queries in initial contact versus multiple touches
  • Decreased support costs – track reduced calls/emails to support teams in terms of both volume and/or cost savings
  • Increased customer satisfaction – repeat surveys to measure any improvements in experience with the new automated processes
  • Growth in relevant KPIs – track metrics like order value, customer lifetime value, churn rates over time, aligned to business goals

Accelerating value through AI

A recap of the key things leaders need to know about building an effective AI use case to deliver demonstrable value for the organisation:

First, start with clearly defining a specific problem statement and a desired quantifiable outcome. This will give a “North Star” to guide the project. Then, focus on solving strategic business problems for the long-term, not just implementing technology for its own sake in the short-term. The focus should always be on business value, not on “shiny objects”.

Finally, think carefully about relevant quantitative and qualitative ways to measure ROI and business value over the long-term.