The recent and rapid rise of AI presents every organization the ability to distinguish and protect their business.  According to McKinsey, AI has the potential to contribute ~$13 trillion in global economic activity by 2030.  AI has and will continue to produce improvements and transformations in efficiencies, insights, automation, and capabilities.  It has the potential to impact every industry, including energy, manufacturing, retail, financial services, healthcare, and logistics.

Most executives have already recognized the impact of AI.  In fact, 75% of executives “believe they risk going out of business in 5 years if they don’t scale AI,” according to a recent report released by Accenture.  Yet in the same report, 76% of executives said they are struggling to widely embrace AI in their businesses.  Successful adoption and implementation of AI has continued to be elusive for many organizations, and often, even knowing where and how to start can be daunting.

For an AI Transformation to start off on the right foot and be successful, the company must select and succeed at a few pilot projects to demonstrate value and build momentum.  Unfortunately, this is much easier said than done.  When partnering with our clients to select an early AI use case, we ask these 3 questions to guide the decision process:

Does the use case demonstrate measurable value to the business?

The biggest barrier that many organizations cannot overcome in their AI journey is getting their models and algorithms out of development.  They’ve hired and paid top-tier data science talent for many months, even years, without having anything to show to the business.  As a result, the organization has lost faith in AI while sacrificing resources and investment while their competitors, who have correctly implemented and adopted AI, have passed them by.

The best way to avoid this scenario is to ensure that your use cases are actually addressing critical parts of the business – not purely academic exercises or “chasing the shiny coin” but actual problems or opportunities that company executives are deeply concerned about.

The first step in determining value is to work with the business organization to understand and measure the change between the improved state and the current state of any business operation.

By selecting a use case of significant and measurable value, you ensure that:

  • Focus remains on the use case and resources do not get distracted or pulled off to the nearest fire
  • At the same time, appropriate resources and investment can be attained to pursue and complete your use case
  • Your AI team is knowledgeable of their value and will be held accountable – their modeling outputs should have benchmarks to meet or beat!
Helpful tips to ensure success:
  • Do your business value due diligence. Make sure that you have comprehensive stakeholder representation in your whiteboarding and discussions, from top-level leadership to the boots on the ground.  This helps ensure that you capture every important use case from every corner of your business.
  • Look for opportunities to either reduce costs (HINT: automation) or increase revenue. Examples for reducing costs are automating tedious tasks like searching /organizing documentation and optimizing your inventory management.  Examples for increasing revenue are pricing strategies and targeted marketing campaigns.

Is the use case a quick win?

Your early wins need to be double wins.  For most organizations, AI is a major investment and seeing any sort of return on investment years later is too risky and impractical.  Instead, ensure that you pick use cases that can be iterated on and produce value quickly (3-6 months).

Helpful tips to ensure success:
  • Leverage your data assets AND your in-house subject matter experts. Don’t forget to utilize and tap into all of your company’s sources of knowledge.
  • Pick a small handful of pilot projects to increase your likelihood of success and prevent a roadblock from delaying all progress.
  • Ensure that you are including your technical data experts in the use case selection and prioritization process. They should be able to speak to how technically complex each use case is.  Speaking of which…

Are your personnel positioned to succeed?

Often, companies do not have the in-house data science, engineering, statistical, or even programming expertise to adopt AI.  That is okay. Consider outsourcing AI talent to quickly get off the ground while avoiding risky long-term hires and investment. It can often take years to find and hire the right AI talent.

In early use case development, start small with self-operating teams (~5-9 resources).  A small and lean team will deliver in short sprints with rapid modeling and agility.

Your AI team should be in lock step and positioned side by side with their business stakeholders.  This allows your AI team faster access to subject matter experts as well as the data they leverage.  In addition, your business stakeholders should stay involved and take ownership of modeling outcomes since they will eventually be adopting and leveraging them for their work.

Helpful tips to ensure success:
  • Ensure that not just your AI team, but also your organization is aligned and regularly communicating on vision, strategy, and progress. The entire organization must be aware of what is and is not technically feasible, as well as engaged and committed to the success of AI use cases.  This also helps your AI team. AI is an ever-evolving field with emerging technologies every day, so it is important to keep early use cases grounded in the business and focused on delivering value.
  • Not all use cases require complex and costly infrastructure like Hadoop ecosystems or GPU clusters. Keep in mind when selecting use cases that many early modeling prototypes and testing can be done on a standard laptop.

 

In this article, we covered three questions we ask when working with our clients to help them choose an AI use case.  For more on how to attain AI excellence at scale, please review our eBook, Reaching the AI Summit.

Sign up to receive the latest insights.

We promise not to spam you and only send the good stuff!