As enterprises move beyond early AI experimentation, one challenge is becoming increasingly clear: AI transformation depends on helping employees understand how AI changes decision-making, workflows, collaboration, and business outcomes at scale.

For many organizations, the real challenge is building enterprise-wide fluency while ensuring AI investments translate into measurable business value.

In the second part of a three-part interview series, Jeff Goldman, VP, Enterprise Data Science at Procter & Gamble speaks with Donna Medeiros of Data Society Group, about how the organization is approaching AI education, scaling business adoption, measuring impact, and identifying the initiatives most likely to succeed across the enterprise.

Part 1 of the conversation covered how P&G built a multi-layered AI organization that combines data science, AI engineering, and what the company calls its “AI Factory” model.

Building AI fluency across the enterprise

According to Goldman, building AI fluency at scale has been a long-term effort rather than a short-term initiative tied to generative AI hype. He explains that P&G began investing in AI education years before the rise of tools like ChatGPT, initially focusing on upskilling its large analyst community.

One of the key initiatives, called “Friends of Data Science,” is a 15-week upskilling and certification program designed to train professionals in applied machine learning and analytical AI.

“We’ve run hundreds of analysts across the company through it to focus on building a critical mass of people across the organization that have a fundamental understanding of AI,” says Goldman.

As AI became more central to business operations, P&G expanded its efforts further through a partnership with Karim Lakhani, Professor of Business Administration at Harvard Business School.

Goldman shares that the company’s internal AI ecosystem also accelerated significantly after the arrival of generative AI. P&G rapidly expanded its internal generative AI ecosystem following the rise of ChatGPT, launching ChatPG alongside additional tools like ImagePG for imaging and video, AskPG for enterprise knowledge discovery, and InsightsPG for data analysis.

As adoption grew, the company recognized the need for broader employee enablement. To support that effort, P&G launched Formula AI, a decentralized upskilling program delivered through a network of AI champions across the organization.

Goldman says the initiative has already educated more than 20,000 employees on how to use AI tools effectively in their day-to-day work.

Why business sponsorship determines AI priorities

As organizations scale AI adoption, prioritization becomes increasingly important. Goldman says that P&G maintains a business-first approach when selecting AI initiatives. “The business problems we solve are almost always directly sponsored by a business leader, and their investment makes pursuing that problem possible.”

At the beginning of P&G’s data science journey, many breakthrough ideas originated from technical teams exploring new possibilities. However, that dynamic has evolved as business teams became more AI-literate through internal education efforts.

Goldman notes that prioritization happens across multiple levels of the organization, balancing global transformation efforts with regional and category-specific needs. Ultimately, he says, funding and measurable business value become the key filters.

Why measurement is critical for scaling AI

When asked why many organizations struggle to move beyond pilot programs, Goldman points first to measurement. He explains that measurement planning frequently takes as much effort as the transformation work itself.

Without that layer, organizations risk endless internal debates about impact and value.

“Whenever anyone’s reviewing a project with me, it’s usually the first question I ask: ‘How are you going to measure this?’” says Goldman.

He acknowledges that some projects may still move forward even when ROI is difficult to quantify, but those efforts become significantly harder to sustain.

Digitization comes before AI transformation

Beyond measurement, Goldman says another major factor separating scalable AI initiatives from stalled pilots is process digitization. Processes with too many manual steps create barriers to algorithmic transformation.

Goldman also points to the size of the business opportunity itself as another defining factor.

“As you look at the business decision you’re trying to drive, is it a compelling enough business problem that you’re going to have the leadership and sponsorship to push through what is often a difficult transition?” he says.

Measuring the impact of AI on teamwork and innovation

One of P&G’s more ambitious AI measurement efforts came through a large-scale experiment conducted in partnership with Harvard Business School.

The organization ran a hackathon involving roughly 750 employees to study how generative AI affects collaboration and performance.

Participants from research and development and commercial teams were divided into four groups:

1. Individuals working alone without AI
2. Teams working together without AI
3. Individuals working alone with ChatGPT
4. Teams working together with ChatGPT

The findings were revealing. Individuals working alone generally delivered weaker results, while collaborative teams consistently performed better. However, individuals using AI tools were able to outperform teams working without AI support.

“The best ideas disproportionately came from teams partnering with AI,” Goldman says.

He believes the experiment reinforced how AI can fundamentally reshape collaboration, product innovation, and research workflows across the enterprise.

AI acceleration and business impact

Goldman says the long-term impact of AI will ultimately appear across multiple operational dimensions, including product development, media optimization, supply chain performance, and speed to market.

He points to a recent Pantene initiative discussed at the Consumer Analyst Group of New York Conference (CAGNY) as an example.

“They were able to go from concept to market on an entirely new product concept in record time because they had the tooling of AI behind them to accelerate every aspect of that campaign,” Goldman says.

For Goldman, these examples reinforce that AI transformation is not defined by a single ROI metric.

He adds that organizations must continue developing creative ways to measure business outcomes as AI becomes more deeply embedded into enterprise operations.

CDO Magazine appreciates Jeff Goldman for sharing his insights with our global community.