Enterprise AI Delivery: Bridging the Gap Between Vision and Execution
In the excerpt below, from the book The Challenger Sale: Taking Control of the Customer Conversation, widespread support is identified as the indisputable top criterion in the process of choosing a supplier.
“Of all the things that decision makers care about, topping the list is widespread support for the supplier across the organization […] senior decision-makers simply aren’t willing to go out on a limb for a supplier on a big purchase – at least not on their own.”
As we question why that would be, additional insights point to the fact that strong team buy-in inevitably leads to an ease of doing business conducive to higher chances of success. The book goes on to say:
“[…] while we might have assumed that things like price and willingness to customize would top the list for decision makers, they’re significantly less important than widespread support and ease of doing business.”
“When it comes time to decide, the decision-maker wants to know he’s got the strong backing of his team.”
As I reflect upon those findings from what is considered a highly successful sales process, I believe they parallel closely with what happens in the delivery and adoption of enterprise AI projects. What I have seen, across many organizations, is the paramount importance of getting widespread support – in particular, the need to bridge the gap between vision and execution.
The Disconnect Between Strategic Vision and Enterprise AI Delivery
We can probably agree enterprise AI projects are tied to an organization’s strategic vision of where it needs to go and to its innovation agenda. We can also probably agree that those plans tend to be sponsored by senior executives. Senior executives need little convincing that a data and insights-driven decision-making process is crucial for any modern enterprise. This notion certainly matches what we see in many of our client engagements in terms of who the sponsors are and how they speak about the role they envision for AI in their organizations.
However, it is not necessarily true that senior executives would have secured widespread understanding and support for their innovation agenda—and, more specifically, for their vision around the key role that enterprise AI technology can play in bringing that innovation agenda to life.
When internal advanced analytics AI teams or external enterprise AI consultants are faced with that scenario, a lot of pain tends to follow. Internal teams may become isolated and external consultants unable to achieve their commitments and deliver value. Heavy dependence on escalations, constant slowdowns and restarts are all signals that this might be happening.
I realize this might come across as common sense, especially when for years we have seen big consultancies talk about the need for business and technology partnerships and about the importance of organizational change management to implement large digital transformations. The fact is though, that statistic after statistic tells us the percentage of enterprise AI solutions actually being operationalized sits below 20%, which obviously means there isn’t just one thing going wrong, but most likely several.
Factors Contributing to the Importance of Early Engagement
In my opinion, when it comes to enterprise AI projects, the need to drive engagement earlier is even higher. There are different factors that contribute to that, for example:
- Fear: AI is often pictured as the “monster technology” that will come to take jobs and replace humans. In fairness, because many use cases do point to autonomy and automation, it is reasonable to understand how that fear could have originated. Most people do not know the vast majority of AI use cases in industry are about augmenting human intelligence and require a human in the loop. The hype around large language models (LLMs) and ChatGPT more recently will potentially exacerbate this fear as there are some very valid reasons for better controls to be established around using AI responsibly.
- Low Literacy: Contrary to traditional software, the concepts behind AI and machine learning are not well understood and the knowledge, even if just at the business level, is not as widespread.
- Higher Uncertainty: Once again, contrary to traditional software, enterprise AI projects—even the most applied ones contain an inevitable level of uncertainty and lack of guarantee on results, given their heavy dependency on the existence of good data and the experimental nature of the data science process.
- Lower Explainability: Certain techniques used make it very hard to explain why a model would produce the results it does. That causes anxiety on the part of the stakeholders engaged.
As I work with my teams, and as we work together with our clients, we believe helping our sponsors bridge the gap between vision and execution is a key part of our mandate and one of our key responsibilities. That means breaking down barriers, educating, being humble and showing empathy to the teams who have day jobs—sometimes in critical operations—with the common goal of bringing everyone along on the journey towards AI adoption, increased operationalization and value growth.