Insights into the modern marketing data stack with Slalom

In the last year, we’ve seen marketing transformed by three interconnected forces: AI, data gravity and privacy. The new edition of the Modern Marketing Data Stack report explores the ways in which these forces continue to evolve based on shifts in technology, market conditions, consumer expectations and the needs of marketers. The report includes insights from industry experts, including leaders at Slalom, one of our trusted services partners helping marketers unlock the full potential of data and AI. Below, we’ll take a deeper dive into their insights on the latest trends in stack technology, why this evolution matters and how marketers and advertisers should be navigating it all.

Accelerating at the speed of AI

The forces of AI, privacy and data gravity have had a massive impact on a range of tools and technologies across marketing workloads. But in 2025, what single adtech or martech development had the most impact on the industry? According to Heather Roth, Director of Digital Strategy at Slalom, it was the consumer-grade application of AI across marketing tools. “Not just gen AI content creation but agentic, embedded intelligence that personalizes journeys, predicts behaviors and automates decisioning,” Roth says. “The expectation shifted from insights to actions, redefining how data is used in real time.”

This widespread adoption of AI has been so impactful that not only are AI/ML technologies now incorporated into every layer of the stack, but a new dedicated category for LLMs has also been added to the modern marketing data stack. The marketing data and AI core and the new LLM layer are closely integrated, powering a range of marketing functions, and organizations are increasingly experimenting with finding new ways to deploy LLMs in their marketing efforts. This new level of experimentation goes far beyond chatbots and customer service. The use cases are increasingly unique, from teams leveraging them to fill in missing data points in customer profiles to powering early versions of autonomous agents to execute multistep campaign tasks.

However, while marketers have made huge strides in developing advanced LLM use cases, there are still challenges. We haven’t yet reached the point where marketers can interact with them as true strategic partners, since most of them are still generalists rather than specialists. As more LLM vendors crop up and models continue to evolve, developers are feeling pressure to stitch together an AI stack — easier said than done when they also have to contend with new vendor onboarding, siloed support and complex integrations. Throw in the additional complication of when security, governance and observability are managed separately for data and for models, and it becomes incredibly difficult to create the transparency needed to deploy AI with confidence.

One way companies are working to address this is by moving the models to the data, rather than the other way around. This helps to create a foundation of trust by protecting customer data, not using brand property to train external models and controlling AI outputs for bias. Creating a unified data foundation where the same rules used to govern an enterprise’s data are extended to the AI learning from that data is crucial for differentiating experimentation from truly lasting, foundational change.

A massive potential of AI that has yet to be fully realized is the truly autonomous AI agent. While this has been a hot topic this year, agentic AI still hasn’t matured to the point where it’s ready to start planning your next marketing campaign and handling multiple tasks at once. When that time comes, AI agents will be capable of coming up with their own workflows to complete tasks, solving problems and working probabilistically. In the meantime, organizations can start leveraging AI agents now by giving them small tasks and repeatable processes they can own, and setting clear guardrails to foster trust and transparency.

So should marketers proactively seek new AI tech, or wait for the vendors in their stack to embed it? According to Roth, it’s best to let vendors carry the risk but to also know when to lead. “Embedded AI in platforms is maturing fast, but smart orgs are piloting bespoke AI models where differentiation matters most — such as media mix modeling, intelligent next-best actions and personalization models. AI success lies in balance,” Roth says.

While organizations are considering the best path forward to upgrade and optimize their stacks, it’s crucial that they approach any martech modernizations as not just a tech project, but a human change project and initiative, according to Logan Patterson, Managing Director for Marketing, Advertising and Customer Experience at Slalom.

“The most sophisticated stack you can comprise will ultimately fail if marketing, business and tech teams aren’t trained, incentivized and understand how their current way of working is going to evolve into new ways of working that are more efficient,” Patterson says. “It will allow them to do the work that they would love to do rather than the work that they’re required to do by automating manual processes with AI to accelerate their workflows and give rise to what I call the human creative renaissance that we can have when we couple human creativity with technology and AI. Leaders really need to make sure they’re overinvesting in enablement and culture rather than just a procurement of new technology.”

The pull of data gravity

AI is magnifying the impact of data gravity: the desire and increasing tendency to unify all enterprise data on a single platform and bring the work to the data, not the other way around. The sheer amount of data that companies are accumulating today is making it riskier and more expensive to move it from one enterprise application to another. It’s also critical that companies eliminate data silos and have a single source of truth to help ensure that the right data — and only the right data — is incorporated into the AI models that are increasingly being adopted across entire organizations. This means that the work, which includes the tools, applications and even AI models, should come to the data.

This development has had an impact on a number of martech categories, but it’s been especially significant for customer data platforms (CDPs). While CDPs originated to bring together data to incorporate into marketing campaigns and programs, the value they bring to marketers is changing. Some CDPs are being incorporated into larger products — often through acquisitions — but many others are evolving and moving into the customer engagement space. Jennifer Fleck, Senior Principal Consultant of MarTech and Digital Strategy at Slalom, recommends that organizations should approach CDP market consolidation strategically, and even cautiously. “Consolidation signals market maturity — and like every martech wave before it, it’s an opportunity for organizations that lead with strategy and customer needs. But for those chasing shiny new tools and big names without clear goals, it’s a costly risk in disguise,” Fleck says.

Prioritizing privacy

Meanwhile, privacy concerns aren’t going away. Even as new laws and regulations are put in place to protect consumer data, consumers’ concerns about data privacy continue to grow. Statistics from the Pew Research Center aren’t pretty: 77% of Americans don’t trust social media executives to keep their data safe, and 70% don’t trust AI companies to protect their privacy.

Between consumers’ growing distrust, a trend toward deregulation and advancements in data collection and monetization, marketers and advertisers are finding it difficult to track where privacy trends are headed. But one thing is certain: Prioritizing privacy is necessary for building trust and protecting customers — and it gives companies a competitive edge. “Consent, transparency and responsible AI are not just checkboxes, they are competitive differentiators,” said Roth. “Organizations must embed privacy and governance into every layer of activation, AI and personalization.”

That can feel like a daunting task in the face of murky situations and constant change, like the one surrounding Google’s shifting policy on third-party cookies. After years of taking a hard stance against them in the name of privacy, they announced in April 2025 that cookies aren’t going anywhere. But for Nick Miller, Senior Director of Marketing & Advertising Strategy at Slalom, this shouldn’t change how companies view their privacy strategies.

“In short, nothing changes for our clients and the recommendations we’re providing to them. With third-party cookies staying — at least for now — it offers organizations a temporary reprieve but not a reason to slow down their shift toward privacy-resilient strategies,” Miller says. “We see it as a chance to double down on first-party data, durable identity solutions and consent-based personalization while reevaluating reliance on third-party signals. Yes, advertisers will still be able to retarget, extend audiences and measure cross-site, but at what cost in the future if they don’t look forward in parallel? Future-proofing is still essential, as regulatory pressure and platform shifts will continue to shape the ecosystem.”

Marketers should build privacy strategies and practices to withstand uncertainty around privacy — and focus on building trust-based customer relationships through responsible, transparent data practices.