How AI and data activation deliver unforgettable customer experiences

At the September 2025 MarTech Conference, moderator Len Devanna, CMO of Cortico-X, convened a practical panel on how to turn AI and data into customer experiences that matter. 

He was joined by Annette Franz, CEO of CX Journey Inc.; Sav Khetan, VP product marketing, at Tealium; and Jiaxi Zhu, head of analytics at Google. The discussion spanned data collection, activation, integration and culture, but one message was clear: technology alone won’t deliver unforgettable CX — people and process must lead.

Hard lessons on data and experience

The panel opened with “hard lessons” from practice:

• Coverage beats convenience. Zhu stressed that surveys and small samples don’t tell the whole story. Without broad, representative data, decisions risk building bias into customer journeys.
• Recency can rival history. Khetan has seen strong outcomes from focusing on the correct recent data rather than hauling every data point into analysis.
• Data isn’t the finish line. Franz reminded the audience that insights must be synthesized, socialized and operationalized to influence employee and customer experiences.

The takeaway: collect the correct data, use it quickly, and turn it into action.

Where marketers still struggle

Two live polls during the session underscored persistent challenges.

• The first asked, “What’s your biggest data activation challenge?” Accessibility was the runaway leader, followed by actionability, collection and trust.
• The second poll focused on martech stacks: integration was by far the weakest link, followed by activation, then ingestion and analytics.

These results mirror industry trends: even as AI adds new capabilities, marketers still wrestle with siloed systems and inaccessible data.

AI-proofing your stack: outcomes first

Khetan warned against chasing architectures before defining outcomes. Too many teams design for a “single AI brain” in the middle of the stack, only to find it slow and costly. Instead:

1. Start with business goals. Anchor architecture choices to clear outcomes.
2. Use models where they add value. Edge models often beat central ones for cost and flexibility.
3. Design for change. Expect smaller, specialized models to multiply; build for interoperability.

These results mirror industry trends: even as AI adds new capabilities, marketers still wrestle with siloed systems and inaccessible data.

Franz added that some firms take the opposite stance — treating AI as a threat and locking systems down so tightly they can’t adapt. Both extremes are dangerous. Success requires guardrails and openness to new tools.

Zhu emphasized grounding AI in specific use cases. When teams can point to concrete value, much of the fear and confusion fades.

Dig deeper: Get an audio overview of the September 2025 MarTech Conference

Culture defines the ‘right’ data

When asked how to focus on the right signals, Franz said the answer is contextually relevant data — information that explains or predicts moments that matter in the customer journey. She outlined three steps:

• Start with the experience, not metrics. Map where customers win or fall away, then select data that clarifies those points.
• Tie every data effort to outcomes—for customers, employees and the business.
• Lean into predictive and prescriptive analytics to anticipate needs and design for them.

Zhu added that coverage, consistency and embedded decision-making are essential for data quality. A simple change in survey scale once shifted his company’s CSAT score by four points, proving how fragile data comparability can be.

Khetan pointed to the explosion of event data — hundreds of millions daily for some firms. You’ll drown downstream if you don’t enforce hygiene and context at collection.

Breaking silos without losing control

The polls confirmed integration as the top pain point. So how do you break silos responsibly?

• Leadership sets the tone. Silos are a leadership failure, Franz argued. Customer-centric organizations design for collaboration and data-sharing from the top down.
• Every team owns consent. Khetan warned against assuming one group can manage everything; data moves too fast.
• Unify the customer view. Zhu described how sales, marketing and support often run conflicting surveys, creating a fragmented truth. A cross-journey analytics team can standardize definitions and insights across functions.

Devanna reframed it succinctly: don’t just sell “customer experience” to leadership; frame it as growth. Growth resonates in the C-suite, even if the levers are CX-focused.

Where AI is already making CX better

The panelists cited practical, high-impact applications:

1. Mining unstructured data. AI can analyze call transcripts, chats and reviews to surface actionable themes — turning noise into structured insight.
2. Journey orchestration. Predictive and prescriptive analytics let marketers anticipate churn or intent, then trigger timely, relevant interventions.
3. Reducing friction. Generative AI bridges humans and systems, making it easier to query, extract and act on insights without technical overhead.

These uses succeed because they combine customer benefit with business measurement.

Who analyzes the data?

Audience questions turned to organizational models. All panelists agreed: in larger companies, dedicated teams are essential.

• A data engineering group owns pipelines, governance and definitions.
• A cross-journey analytics team handles CX data holistically.
• Embedded analysts within marketing or product ensure insights drive real decisions.

This structure avoids dueling dashboards while keeping analysis close to action.

AI personas: promise and pitfalls

On using AI personas for market research, the panel was clear:

• Talk to customers first. Data averages can blur nuance; interviews build empathy.
• Then codify with AI. Personas can scale knowledge across departments, but must be grounded in authentic customer voices.

Speed-round advice

The panel closed with one-sentence advice for teams early in their journey:

• Franz: Build a people-first culture. “You can’t AI your culture.”
• Khetan: Start small experiments to build confidence.
• Zhu: Embed AI and data in decision-making, tying them to revenue and CX.

5 steps to act on now

1. Pick two outcomes and three moments that matter most; align data and AI projects to them.
2. Establish a minimum viable integration of identifiers and events across systems.
3. Enforce quality at collection to prevent expensive cleanup later.
4. Apply AI to unstructured inputs like transcripts or reviews for weekly CX insights.
5. Create a cross-journey huddle where marketing, product and support align on shared CX scorecards and experiments.

The bottom line

Unforgettable experiences aren’t delivered by AI alone. They emerge when organizations align culture, outcomes, integration and consent — with AI and data as accelerators. As Devanna summarized, technology is just one piece of the stack; people and process determine whether it actually delivers growth.

Listen to an audio overview of the September 2025 MarTech Conference

Use the player below to get an AI-generated summary of the conference.