Looking into Integrating AI? Think Beyond Business Value Alone

Any business leader, analyst, or engineer involved in deploying AI must examine a number of essential requirements before diving in. Defining a use case solely based on business value could put you in perpetual purgatory.

Let me explain by first defining business value in its simplest terms. Business value represents the tangible and intangible benefits that a company realizes to its overall profitability and success. This can include increased revenue, cost savings, market share growth, customer satisfaction, brand reputation, and more.

However, just as beauty is in the eye of the beholder, business value is in the eye of the stakeholder. What’s considered business value to one business leader may not be to another. For example:

• What’s important to the head of manufacturing is different from what’s important to the head of supply chain.

• The goals of a regional president in North America are different from those of a regional head in China.

• One function may be focused on growth, while another focuses on efficiency.

• One business unit prioritizes market expansion, while another prioritizes consumer loyalty.

The list goes on!

Due to the diversity of the stakeholders you serve — their competing strategic priorities, how they are compensated, and their personal agendas — it’s nearly impossible to please all by picking use cases based on business value alone. If applied as the sole reason, business value will set you up for trouble.

So how should you investigate how best to proceed with your AI deployment? Begin by exploring complexity and criticality. You need to establish a measure of how complex the deployment will be. You also need to establish a measure for how critical the objectives are to the longevity of the company.

In terms of complexity, consider the following:

1. Stakeholder involvement - Is the person you’re doing the work for a good partner? Will they be engaged and involved? Will they have regular meetings with you and your team? Will they give you active support when needed?

2. Resource availability - Will the resources you need be available to help or are they being pulled into too many directions? Will scheduling meetings to align on direction and decisions be difficult to coordinate? 

3. Data accessibility and integrity - Is the data you need for this use case available, accessible, and of relatively good quality? The last thing you need is to take on a data project while delivering a use case. 

4. Dependency on other functions - Do you have dependencies on other small teams, functions, and groups that are hard to work with? …that have low attention spans? …that move at a different pace than you?

5. Infrastructure - Do you have the basic infrastructure to do the work or do you need to put that in place?

Go through each of your requirements and rate them in terms of the five complexity considerations with your team. Why? Because, your success is determined both by value and your ability to deliver with excellence — which means on time, on budget, and with business outcomes.

Most new projects fail to launch in production if teams take on requirements beyond the organization’s maturity. They end up spending too much time putting in place the fundamentals. This means that you must be realistic about where you are as a company and select a project that “bends but doesn’t break you” as a company.

Your feasibility to deploy and ability to deliver is just as important as the value that your AI implementation will bring. So reduce (if you can’t remove) as many of the barriers to success as possible.

As for the AI deployment’s complexity, consider the following:

1. Competitive threat - Take a look at the competitive threat from both established and customized businesses that are fast approaching. Which of your concepts strengthens your market position and posture?

2. Market consolidation - What markets and distribution channels are consolidating versus expanding? Which scenario supports these channel strategies?

3. Government regulations and fines - Government regulations, data privacy laws, and federal/state/local acts are being established with fines and penalties if not followed. Which of your concepts has a financial implication if not compliant?

4. Exposure and press - In today’s social media culture, everyone has an equal voice. Take a look at your use cases and see if any of them can be provocative in nature or debated in public.

5. Business value - Establish which concept focuses on increased revenue, cost savings, market share growth, customer satisfaction, and brand reputation, and see how it aligns with the company’s strategic priorities — not just those of business leaders.

Go through each use case and examine these five complexity considerations. Rate them with your team. Why? Because when you define value you want to make sure you’re not isolating leaders or getting involved with politics, and that you’re taking into consideration the macro-landscape as well as the mico-landscape to position yourself as an objective and fair thought leader.

Both criticality and complexity are objective measures to determine which AI, data, technology, or business use case has the highest chance of succeeding. Not only should it bring value, but your organization should actually have the ability to deliver it within the needed time frame.

Nothing takes the wind out of sails more than a project that drags on endlessly. Teams lose steam, the company gets fatigued, and stakeholders get bored. Make sure you thoroughly ponder all the important elements beyond business value so that you don’t end up in deployment Purgatory.

About the Author:

Sol Rashidi is Head of Technology for Amazon North America and an influencer within the AI, data, and technology space. Forbes has called her the AI Maverick and Visionary of the 21st Century. She’s been named Top 100 Thought Leaders in AI, 50 Most Powerful Women in Tech, is on the Global 100 Data Power List, and more, and she holds eight patents.

Rashidi’s new book “Your AI Survival Guide: Scraped Knees, Bruised Elbows, and Lessons from Real World AI Deployments” (Wiley, April 30, 2024) was named among Best Books in AI by CEOWorld and made it to the bestseller list in less than one month.