In 2025, it’s already evident that AI is delivering value at scale. Nearly half of technology leaders in PwC’s October 2024 Pulse Survey
With AI reshaping businesses on such a vast scale, the question becomes: how do you build a sustainable AI-driven organization? According to Doug Sutcliffe—an accomplished product management expert, MIT Sloan Fellow, and former Director of Data-Driven Marketing at L’Oréal US and marketing leader at L’Oreal Canada, Moen, and AIMIA — sustainability in AI comes from reliably generating innovative opportunities powered by AI, all while maintaining cost levels that the business can continue to support. Achieving this balance requires two critical elements: data readiness and a culture of experimentation.
Robust data systems enable faster, more cost-effective testing of new concepts, reduce time spent on data preparation, engage data science talent more effectively, and allow successful solutions to scale quickly for maximum impact. These qualities create and compliment a strong experimentation culture by fostering an environment where it’s safe to propose, test and implement new ideas, and where the ones that work can have much larger impact.
In this article, Doug explores how companies can fast-track AI adoption and drive AI-powered marketing, whether through centralizing data, restructuring teams, or rethinking metrics.
Unified Data, Accelerated AI
AI projects often take an extraordinary amount of time to launch due to inconsistent and misaligned data. Before any meaningful analysis can begin, a substantial amount of time is spent on data cleaning and preparation—a process that frequently takes far longer than the data science itself. This creates three major challenges: project costs and timelines balloon to accommodate unplanned data pre-processing work, solutions become overly specific to individual business units and are difficult to scale, and most critically, attracting and retaining top data science talent becomes harder. Professionals spending the majority of their time on repetitive data cleansing tasks have fewer opportunities to engage in challenging data science work they usually pursue.
A solution to this issue lies in building a well-documented core architecture that standardizes, cleans, and organizes all incoming data. By ensuring every new source connects directly to this centralized system from the moment it begins generating data, companies can eliminate the need to reprocess and clean information for every new project. This approach not only streamlines individual projects but also enables businesses to scale successful data science initiatives across the organization far more efficiently.
“If you identify key data classes—such as customer, sales, and product—and integrate all new sources into the same canonical (bronze, silver, gold) data flow as existing data, you will have a single source of truth that remains clean and ready for use at any time,” Doug explains. “Any new sources will seamlessly integrate into existing pipelines, automatically enriching reports, automations, and tools without additional effort.” For an even more robust system, Doug advises building all applications off this central flow and integrating AI in the data onboarding and data governance workflow to easily maintain high-quality alignment of the data to this strucutre. This guarantees that any required updates are made only once when new information is added.
By adhering to these standardized design principles, you can ensure that future enhancements would integrate seamlessly with the streamlined system. Any data-driven initiative, such as analytics, targeting, or personalization, will follow a consistent process. Simplifying the data path also makes it easier for new team members to onboard and use the system effectively. In addition, a unified pathway improves stability and accuracy, accelerated deployment timelines and delivers cost-efficiency.
Empowering Teams for Innovations
Launching AI initiatives is not just about data—it’s about assembling a team with the right technological expertise. But ironically, AI and ML readiness often depends less on data science knowledge and more on the quality of the underlying data and team structures, Sutcliffe emphasizes. When data is well-organized, intuitively presented, and effectively documented, modern platforms and Large Language Models (like Chat GPT) empower users to rapidly develop scalable AI and ML solutions.
With robust data systems in place, AI teams can shift their focus from managing technical complexities to delivering meaningful business outcomes. This approach ensures that enough ideas are assessed to uncover truly innovative solutions while providing teams the bandwidth to refine and evaluate strategies based on sound business value.
Doug notes that from a team perspective, success relies on the ability to rapidly test multiple concepts and allocate resources to scale the most promising ones. Achieving this requires a shift from skill-based to deliverable-focused structures. In traditional marketing departments within large organizations, projects are often fragmented across multiple teams, each responsible for a specific stage of the process. This siloed approach leads to inefficiencies, as handoff delays force teams to plan their tasks much later than necessary to avoid last-minute disruptions. Consequently, projects that could be completed in weeks are stretched out over several months.By consolidating teams into single units responsible for managing particular projects end-to-end, reallocating budgets and renegotiating vendor agreements, companies can significantly accelerate the launch of AI-driven innovations while simultaneously reducing costs. Such an approach can enable the delivery of dozens or even hundreds of groundbreaking solutions every year. By empowering these teams with secure, offline language models like customized implementations of the latest Llama, these results can even be achieved without growing team sizes as such tools provide incredible abilities in automating administrative tasks such as system documentation.
Actionable Metrics for AI-Driven Marketing
Once the restructuring and redesign of the data architecture is complete, it paves the way for accelerating the shift from Data-Driven Marketing to AI-Driven Marketing. However, it requires more than advanced algorithms, Doug explains. It must be designed to deliver impactful insights in a format that is both actionable and practical for decision-making.
Rethinking metrics is what is essential to unlocking dramatically better results. “Too often, leaders rely on metrics inherited from previous leaders or carried over from prior roles or companies,” Doug shares. “But one of the biggest opportunities for growth is reassessing your chosen metrics with fresh eyes to ensure they are clear, actionable, and aligned with your objectives.”
For example, in the world of Consumer Relationship Management (CRM) many organizations separate consumer acquisition and retention, or at best, track active buyers with a single metric, analyzing acquisition and retention strategies within each purchase channel in isolation. However, by shifting focus to maximizing lifetime revenue efficiently, companies can recognize that it’s not just about optimizing within channels—it’s about reallocating resources toward efforts and platforms that deliver the highest lifetime value.
Doug identifies two significant changes most organizations need to adopt. First is bringing more data together. Metrics like customer acquisition cost (CAC) from paid media should be analyzed alongside retention, return on ad spend and more, enabling comparisons across strategies rather than just within them. Second is adopting a lifetime value (CLV) mindset. Without a reliable proxy for how current decisions impact customer relationship longevity, it’s impossible to effectively allocate budgets and other resources.For many companies, evaluating customers based on CLV is standard practice, but applying this approach to segments, business units, or portfolios is often uncharted territory. Doug recommends a step-by-step progression to quickly and accurately optimize efforts:
-
Create an estimated CLV (eCLV): Multiply each customer’s annual spend by the historical average lifetime across your customer base. This will provide a quick an easy estimate for broad applications like getting a general picture of consumer spending or an early indicator of how it’s evolving. While this is only an estimate and doesn’t actually measure the spend over a whole customer’s life, changes in this metric can be measured quickly.
-
Leverage Machine Learning: Use time series modeling to forecast individual customer behavior and create a predicted CLV (pCLV). This metric is more sophisticated and can be used to identify high-potential customers earlier on and engage with them more effectively for more of their customer lifetime.
-
Integrate Costs: Factor in acquisition, targeting, and promotional costs to calculate predicted Consumer Lifetime Profit (pCLP). This metric brings everything together to understand not just which customers bring in the most revenue, but actually create the most value. This allows you to invest more in building deep relationships with truly loyal customers without diminishing brand value with excessive market-wide promotion.
Doug also advocates for adopting a unifying metric like the Net Present Value (NPV) of Lifetime Revenue, underpinned by the pCLP metric.
By incorporating costs and benefits while balancing short- and long-term impacts, pCLP can align purpose and action across the business.
“While this approach is standard in finance, I believe it has transformative potential for sales and marketing teams,” Doug notes. A single unifying metric like this can revolutionize how businesses prioritize technology development, set campaign budgets, and allocate resources.
Engaging the Modern Consumer
When discussing marketing trends, Doug notices that today’s consumers increasingly demand engaging and immersive experiences, expecting marketing materials to reflect their own identities. Research consistently shows that users respond more positively to images featuring models who resemble themselves. While personalization and media tools like Dynamic Creative Optimization can adapt content for specific consumer groups, only Augmented Reality offers a truly personalized experience—whether it’s trying on clothes, testing makeup shades, or visualizing furniture in your home.
However, as more companies adopt these solutions, how can brands differentiate themselves? According to Sutcliffe, while AR tools can drive 2-3x higher engagement and conversions, their growing prevalence makes it essential to actively promote their availability. It’s not enough to develop these tools; consumers need to know where to find them. To achieve this, brands should leverage new AR advertising spaces (e.g., Snap and Meta) to offer interactive experiences directly where their target audience is. This ensures every ad becomes 100% personal, allowing customers to deeply engage with products without leaving their browsing environment.“Push the technology further,” Doug emphasizes. “With so many basic AR tools available, it’s crucial to advance the capabilities of AR while keeping the customer at the center of the story.” Leading innovators are now integrating AI-powered product recommendations into purchase experiences, offering consumers an interaction akin to a trained shopping advisor. With the rapid evolution of language models like ChatGPT, combining conversational AI with AR can create a tailored, high-touch service experience—even in a virtual setting.
AI also enhances the customer journey at multiple touchpoints. It can maximize engagement, improve data capture, and boost retention with solutions like virtual guides and interactive manuals. Imagine scanning a code on a product to activate a virtual assistant that guides you step-by-step—like identifying the next screw for an IKEA project and visually showing you where it fits. Similarly, informational overlays, such as player stats hovering above athletes at a live sports event viewed through your phone, can transform the consumer experience.
Still, looking ahead, the most significant value of these technologies lies in their potential for AI-driven data governance. By automating data management, companies can reduce the burden on technical teams while ensuring data is ready for rapid testing and scaling. This capability will be critical in optimizing consumer experiences and maintaining a competitive edge.
