Moving Beyond ROI: Return on Adoption

With generative AI showing early promise of substantial productivity gains, organizations should start to think not just of traditional metrics like ROI but of ROA, Return on Adoption, as a critical measure of success in the future.

The reason for this is the widespread impact that generative AI solutions will have. It’s not about rolling out a new application in a single department but vastly improving the productivity of your workforce, comprising technical and non-technical knowledge workers. The strategy for the tech behemoths like Microsoft (w/ OpenAI) that are driving the adoption of AI is to tightly integrate AI into existing applications (See Office 365 Co-pilot, GitHub Co-pilot). For companies, widespread adoption is now feasible, and focusing on ROA highlights the opportunity cost of moving too slowly.

Typical ROI calculations include the investment in technology, including startup and ongoing maintenance and subscription licensing, while using some return assumptions related to cost savings, margin improvement, or top-line growth. For example, while prior application deployments would have people-cost components associated with training employees on a new solution, most expenditures would be technical (licensing, support).

The variables differ with AI, as rollout and adoption will require greater investment in people than we’ve seen with other technologies in the past. It goes beyond upskilling, which will be necessary, requiring new mindsets and accepting new ways of working to take advantage of these powerful and rapidly advancing capabilities.

The productivity potential is there, but the degree to which employees will adopt these new AI capabilities is very much a question mark.

In the past when new technology was deployed in an organization, users were accustomed to learning a set of functionalities and becoming proficient over time in applying this knowledge to their daily tasks. They would use a tool to accomplish a specific task (e.g., Salesforce to track pipelines or Excel to produce a dashboard). They were in control and responsible for the output resulting from their efforts. With AI the idea of a smart assistant comes into play, where users can now delegate entire tasks to their “assistant,” which begs the question of who is responsible for the output and introduces additional questions related to which tasks to delegate and to what degree one trusts the veracity of the work product coming from the AI.

In short, with AI users will not only give up some control but could also begin to dissociate or feel less ownership over the work product. As a result they may feel less pride in their authorship, which could make them resistant to adopting AI, believing it is just one more step to being replaced.

However, some promising early research (with the caveat that it’s not coming from an entirely impartial source in Microsoft) reveals attitudes from leaders and employees that may help spur adoption. Microsoft’s recent Workforce survey found that:

In addition, leaders are more focused on increasing productivity than reducing headcount:

The power and pervasiveness of AI make its potential impact orders of magnitude higher than anything we have experienced. In the past, new solutions impacted certain functional areas, like SaaS software for enterprise applications or AWS for infrastructure. Both had a substantial impact, but nothing has come along that could realistically deliver up to 50% or more productivity gains across an employee population. Wharton professor Ethan Mollick found that recent research indicates the gains could be real and substantial.

Therein lies the “Return” related to ROA and the significant upside potential for organizations. But the return is predicated on adoption, and realizing widespread adoption will require dedicated effort by organizations to create the conditions for change. So what will be needed to transition to this new way of working?

There are two core elements of the transition.

The first is shedding low-value tasks, which will drive productivity gains. If we use the standard definition of productivity as more output per hour, employees will produce more stuff (e.g., docs, presentations, reports, analysis). However, the concern leaders are likely to have with this is that more is not necessarily better. Focusing on the output volume doesn’t say anything about quality, which is essential to delivering real results.

The second component, working on higher-value creative and analytical work, is where organizations will realize the real value and where employees will probably face the biggest transitional challenges. They will be asking themselves (and their bosses!) what it means to be “more creative” or “more analytical,” particularly in the context of their current jobs.

Developing new mindsets and skill sets will be critical to breaking free from old work habits. But developing new work habits requires new behaviors, and making them stick will require anchoring these new behaviors to defined processes. Additionally, AI will automate not only workflows but also a variety of decisions, necessitating the development of a new discipline: decision process improvement backed by a disciplined change process.

So transitioning to an AI-ready organization focused on adoption will require leaders to execute across three dimensions:

Return on Adoption of AI will be fully realized when employees successfully leverage AI to improve productivity, become more creative, and develop greater analytic rigor. The latter two components will help create a culture of continuous improvement. Staff armed with the tools and the time will be better able to evaluate and improve business processes, discover new customers or markets, and create new business models. All of which will lead to growth, profitability, and market share without adding headcount. The critical first step for organizations is prioritizing AI adoption.