The latest advances in AI are revolutionizing marketing—enabling personalized customer experiences, optimized pricing and promotions, and smart predictive models. Generative AI now commands the attention of global CMOs, as companies are exploring use cases to create competitive advantage. We spoke with Silvio Palumbo, founder and global leader of Fabriq, BCG’s AI-powered marketing personalization platform, about how AI is transforming personalization, why experimentation is key, and how tools like Fabriq can help companies incorporate GenAI into the customer experience.

Silvio Palumbo

Managing Director & Partner

New York

BCG: What are some of the common hurdles that teams face when implementing personalization programs?

Silvio Palumbo: We often see two recurring bottlenecks in companies: execution velocity and siloed operating models. It’s very hard to move at the speed of personalization. Consumers’ tastes, behaviors, and frequency of interactions are constantly changing, and traditional marketing campaigns don’t move at the same pace. That’s one bottleneck.

Second, effective personalization touches numerous parts of a company, requiring a model where the entire organization has a role to play. This concept can be at odds with the standard structure of many organizations. Not only do companies need cross-disciplinary talent in terms of different units working together internally, but they also need to maintain a unified dialogue with consumers.

Consider a large financial institution. Different units, such as retail banking, card services, wealth management, and deposit might each reach out to a consumer individually. This can create a perception that the consumer is dealing with four different teams. Internally it’s not managed cohesively, and so externally it comes across as uncoordinated.

We’re all consumers, so we all know what “good” looks and feels like from a consumer perspective. However, getting things right operationally is a complex task.

What role can AI have here?

AI plays a critical role along two dimensions. One part is taking care of time-consuming, repetitive tasks and streamlining workflow—essentially, automation. The other part is about scaling the marketers’ intuition. Imagine being able to tap into product recommendations, individual pricing decisions, and incentive optimization, all while conducting personalized experiments for each customer in your entire market. And all of this can operate on autopilot, with marketers driving the strategic decisions.

Many companies want to make the jump into AI but find that they struggle when it comes to data quality. What does the conversation around improving data quality look like with companies?

Data quality issues can vary considerably across industries—a retailer has typically richer and more structured data than an insurer—but generally, companies can focus on three key areas.

The first is being able to actually capture all the data you already have exposure to. The initial step typically involves recognizing that every interaction or touchpoint with your customers should be captured and organized deliberately. This starts with proper tagging all online experiences to collect critical data points. This goes beyond the basics like “did the consumer make a purchase?” and includes browsing behavior, time spent reading descriptions, and more.

Even touchpoints that might seem unstructured, such as a call center conversations or doctor notes, can be converted into structured data. For example, by transcribing and categorizing the call content, you can gather rich, actionable insights.

The second area is about creating the data you don’t have. This involves a different and much more strategic approach. One way to start is by experimenting. You can experiment on pricing, promotions, or even the language used in communications. Experimentation is a great creator of data, and it can help you create the data you want.

A third approach involves finding novel ways for consumers to connect with you. For example, a client, a global pharmaceutical company, found that one of the biggest hurdles for consumers to use their product was lack of information. The company created an unbranded website that offered helpful information to consumers and became a major source of data capture when prospective customers would look for a provider.

In a recent BCG survey of CMOs, 67% of respondents said they are exploring generative AI for personalization. What are the biggest use cases right now?

Interest in GenAI is genuine and soaring, and the exciting part is that it’s not just another passing trend. Unlike previous AI fads, practitioners are truly enthusiastic about the power and promise of GenAI, in large part because of potential for monetization.

What also sets GenAI apart from previous AI trends is that the technology has matured faster than the current real-world applications. Right now, it’s mostly being tested internally to create efficiencies across many domains, such as automation, creating images, creating copy, summarizing text, and identifying patterns and trends.

Unsurprisingly, we’re seeing a lot of interest and excitement in sectors that handle large volumes of data. Here it’s about helping internal individuals and teams to triage this data, whether it’s summarizing a financial report or providing a preliminary compliance or claim review. These are powerful examples of how GenAI offers valuable insights and speed, while humans remain in the loop and are the drivers when it comes to those crucial final decisions.

As companies become more adept with generative AI, how will marketing evolve?

I think in the future, applications will become more consumer-facing, and a good portion of that will come out of the box within the martech ecosystem. As we all become more comfortable with the technology and as companies gain a clearer understanding of the costs involved, GenAI could start playing a much bigger role in enhancing the customer experience.

This second wave of GenAI will be about helping consumers in novel ways. For instance, thinking a couple of steps beyond the shopping cart, helping with assisted search (not searching for an item, but searching for a solution to a problem), communicating and solving operational and logistical problems on the fly, tailoring imagery to individual preferences during real-time interactions, and simply surfacing a product to the usage, fit, or relevance of the product for that customer.

What is smart integration? And how can companies use it to create value?

Smart integration is the recognition that deriving value from AI doesn’t come from creating new AI solutions, but rather from effectively applying existing AI solutions within larger business processes.

Think about traditional AI, where a solution—such as a product recommender on a grocer’s website—provides predictions about a consumer’s future purchases. For example, it might predict a customer is likely to buy milk, eggs, and strawberries. But the true value doesn’t lie solely in these predictions; it’s how businesses leverage them to make strategic decisions. The advantage comes from smart integration, using these insights to inform promotion and pricing strategies.

Smart integration focuses on two key areas: experimentation and application. With experimentation, companies continually generate diverse, high-quality data to feed their algorithms, leading to better predictions. In the case of the grocery system, this could involve testing different factors, like browsing patterns or time of purchase, to refine recommendations. That capability is the combination—hence “integration”—of many components beyond AI.

Application refers to the flexibility to swap out new algorithms and solutions as business needs change. This requires a comfort level among leadership for ambiguity and adaptability.

How can tools like Fabriq play a role here for companies, in terms of orchestrating a more effective martech system?

Creating any type of AI-informed experience can be challenging, because there are many components that are making that system work. The AI is just one part of a bigger picture.

That realization is very important. Competitive advantage doesn’t come simply from going all in on AI. It comes from building a system where all the different moving parts work well together. How companies make these systems work effectively connects back to the organizational model.

Fabriq is BCG’s personalization platform and acts as an augmentation layer to existing solutions. Just as the name suggests, Fabriq weaves together different components in your tech or martech ecosystem, true to the smart integration ethos. Fabriq is not a single entity, but a set of modules that can work in isolation and/or as a collective, each adding automation or intelligence to processes like experimentation, campaign creation, media optimization, or measurement.

How do the three different dimensions of the Fabriq platform (identity, media, and personalization) help organizations address digital challenges?

Many organizations struggle to fully leverage their tech resources for value, and Fabriq addresses the need for speed, connection, and orchestration. The platform can be used as an add-on in areas needing focus or where IP ownership is preferred.

And given that the code base is accessible, Fabriq enables companies to create their own competitive advantage across three critical consumer-facing dialogues: knowing the customer (identity), acquiring the customer (media), and nurturing the relationship (personalization).

As the boundaries of personalization continue to expand, what emerging trends or technologies should businesses keep an eye on?

The rise of GenAI is quickly shifting how we will use tools. Vendors are set to embed GenAI solutions within everyday tools, which will take some of the pressure off organizations in terms of needing to create their own GenAI solutions. We’re seeing this already happening in offerings from companies like Adobe, Salesforce, Google, and Microsoft.

At the same time, the market is also experiencing a trend towards consolidation, with one-stop solutions emerging. This convenience factor doesn’t mean that businesses should put all their eggs in one basket. No one vendor can excel at everything, and organizations need to retain their own unique intellectual property or “secret sauce” to stay competitive. This might mean creating something that is differentiated and customized to their operations, even while leveraging AI-embedded vendor tools in other areas.