With accelerating AI investments, a question many executives have is whether these investments deliver returns that align with the strategic objectives of the business. Many businesses might be adopting AI for technology sake but the lack of a vision for tackling real business problems actually holds them back.
To ensure there is a positive return on investment, there are four key areas to consider in implementing AI that clearly addresses a business problem.
- Business outcome orientation
Many companies begin with the question on “what they can do with AI” instead of “what business outcome they need to move the needle on.”
This requires identifying a specific outcome such as “reduce the average customer support call handle time by 30%.” With that, work backwards to determine what AI capability is required. The clarity enables the business hone in on what to look for from demos and build excitement for pilots that get buy-in.
The clear business case with well defined outcome(s) helps to incorporate AI where it would be most helpful and select the technology that would realize that objective.
- Solid data foundation
Reliable AI output builds on good data. Deploying AI using fragmented and poor quality data is wishful thinking that AI could fix data issues.
Data readiness is a pre-requisite to scale any AI initiative. This means businesses need to audit data quality, review accessibility, and understand the data life cycle. The AI model’s accuracy relies on the data used for training.
Businesses do not need to embark on a multi-year data transformation effort but do need to be ruthless about cleaning the data and centralizing data for the specific needs of the AI deployment.
- Workflow redesign
For AI deployment, the most common approach is to train employees on AI fluency. Training alone is insufficient to deliver optimal results. Businesses need to redesign workflows, roles and decision authorities around which AI operates.
By mapping the end-to-end workflow where AI touches, it identifies what decisions could be made faster, lower in the organization or automatically. In addition, redesigning roles around the new AI capability makes the workflow and responsibilities more logical. It also removes the perception that AI is an add-on to existing workflows.
Using AI in broken processes generates slightly faster broken processes. However, buy-in and results could be boosted dramatically when processes involved are redesigned.
- AI performance KPIs
Similar to other technology deployment initiatives, it is important to define clear ownership of performance so there is an organized approach for follow-up and escalation when there are problems.
A business owner is an individual responsible for AI’s performance standard. This person oversees when human review is triggered and when to update the model as conditions change. Inconsistent outputs erode the trust needed for broader AI adoption.
While IT can still be the technology owner, it is the assigned business owner who steers the business to focus on business applications that matter.
These four areas support a logical flow for effective planning and execution of AI initiatives. In fact, the same applies for any technology investment. The shift away from technology focused to business-outcome focused makes a difference in the final return on investment.
Check out A Pragmatic Approach to AI Adoption for ideas on identifying opportunities for sound returns.
Leave a Reply