Most companies are using AI in some form or another, either for research or simple efficiency in crafting marketing messages or handling customer complaints. Some companies have figured out that getting AI Adoption right in their business is giving them phenomenal edge.
According to a survey by Harvard Business Review of 100 companies across various industries, top 25% of the respondents in the survey see 3.8x performance improvement compared to the bottom half of the respondents. The payback on investment in 6-12 months vs 18-24 months for the laggards of the group.
For most of the other organizations, that performance improvements and paybacks are still in the future. A lot of the organizations are still figuring out the vision and strategy of incorporating AI in their business. Over the past few years, we have helped clients adopt AI and other leading edge technology solutions in their business. Here are some of the things that we learned from these transformations:
AI Adoption is a strategic initiative:
Today, boards and C-suite at all organizations are actively including AI as part of their overall business strategy and have started planning for it.
What works: Successful AI adoption is an organization-wide initiative. This means that the priorities must be re-aligned, budgets and resources re-allocated and decisions are driven from the top with this vision in mind. In 2024, Publicis group announced a Euro 300 million investment over 3 years to leverage their in-house data and AI technologies including Generative AI to transform the company into Intelligent System company. Clearly, the board and C-suite level went through a lot of planning and alignment before this strategy was announced. The key point here is that they didn’t just talk the talk, they put the money where their mouth is. They are focusing the entire organization on this transformation.
What doesn’t work: Putting AI in a silo for few departments to work on while the rest of the organization continues to run the business is a sure recipe for failure. Companies need to learn from the past waves of transformation. Kodak missed the transition to digital and while it is still around, it is a shadow of its past self. Similarly, Blockbuster missed the ship-to-home DVD rental and the advent of streaming service and paid the price. The good news here is that it is still early in the AI adoption cycle for companies to adapt. But refusal to adapt or to take a “wait and see” approach could be a death knell waiting.
Set and measure goals
Early adopters of AI were experimenting and testing out the technology. Hence, frequently the goals were technology related, like user adoption, feature use, etc. To be successful at AI, align these goals to business goals.
A Deloitte survey of business and technology leaders found that the set of KPIs the two groups focused on differed with technology leaders focused on sales and NPS scores while business leaders focused on process improvements and efficiencies. Secondly, there is a 7% gap between the two groups on the KPIs of traditional AI. This gap for Generative AI KPIs widens to 37%.
What works: Successful organizations do not ignore their business goals. Instead, they keep them front and center and adopt new KPIs related to AI as leading indicators of progress towards the goals. Educate business leaders on the benefits of AI and how to use & measure success. Conversely, technology leaders need to understand the business drivers and help business leaders connect the dots. Share the metrics broadly with all your stakeholders and decision makers as they become available. Shining light on good or bad metrics helps organization understand how the transformation is progressing and the challenges that must overcome to meet the goals.
What doesn’t work: Business and technology having different sets of goals often leads to the two teams driving at cross-purpose. Misalignment leads to focus on wrong priorities and results in lack of progress. We recently helped a client where this misalignment led to $3M waste and doubled the payoff period.
AI Adoption is a team sport
Any transformation initiative requires a team effort; AI adoption is no different. The difference is that the skills required are new and not widely available. This technology is evolving faster and continuously. So, you need a diverse skill set both internal and external. Engineering teams alone cannot enable the transformation.
What works: Ideally, this transformation should be led by the business leaders who are most impacted by this transformation. Hold the business leaders accountable and empower them to make decisions. Stakeholders across all the departments from legal to customer service, from engineering to cybersecurity, from finance to HR, should all be active participants in the project. Educating the teams on what is AI, its benefits, challenges and impact should be a top priority as the project kicks off. Leaders who are active participants, learners and users of AI help inspire team members, reduce anxiety and roadblocks to adoption.
What doesn’t work: Putting the transformation project in a silo or viewing it as an efficiency improvement project is a guaranteed way to failure. It sends a wrong message to the organization. It appears as if this is not an important initiative, demoralizing to the entire organization in the process.
Learning new skills is a must
AI adoption involves new set of skills being introduced to an organization, not only on the engineering side but also on the business side. Engineering team will need new skills like prompt engineering, ML Ops, conversation design. Similarly, business teams will need new skills like content creation, content curation, ethical decision making.
What works: Identifying these new roles and conducting a skills gap analysis to fill those roles should be one of the first steps in the transformation efforts. The earlier you identify the gaps, the more time you have to fill those gaps. Training existing staff on some of these skills or hiring new staff is a time consuming process and needs to be actively managed. Relying on your partners to help plug some of these gaps is also a good idea.
What doesn’t work: Expecting your team to learn these new skills while still doing their day-to-day work does not work.
Change management should be front and center
Transformation projects cause a lot disruption to existing processes, job roles and tasks to be done. As such, they are a significant concern for employees with respect to job loss or changes in roles. Rightly or wrongly, AI has been touted as significant destructor of jobs across industries. This can lead to adverse employee behavior like lack of participation in knowledge sharing, refusing to use the technology or nitpicking flaws even when the solution is working.
What works: Informing and engaging employees from the start helps boost employee confidence in the initiative. A dedicated change-management team that can educate employees on technology, how it would impact their jobs can be the difference in the success of the project. A detailed change management plan including training plan, feedback loop and escalation process is another key to a successful AI adoption. The change management team can be the eyes and ears of the project team. They can share questions and feedback from the end users and impacted employees that can help improve the solution.
What doesn’t work: Deploying AI as just another software upgrade without adequate training for the user is a recipe for disaster. Lack of transparency about the impact of the transformation to the employee roles and duties leads to resentment, low morale and employee churn.
As you can see, there is more to successfully launch AI initiatives than a traditional software development project. This is a transformational project and hence the entire organization needs to prepare and actively participate in it.
Note: This article originally appeared in my LinkedIN newsletter on January 29, 2025. You can read it here. This post is a slightly modified version of the original post.
