How Marketing Leaders Build Better Data Strategy

1. Why Marketing Has to Lead: Dave argues that marketing cannot remain a “luncheonette,” producing whatever product teams or business units request. Modern marketing needs data for three core jobs: developing insight, making decisions, and proving impact. Yet the required information often sits with agencies, sales teams, e-commerce groups, Salesforce environments, or other functions. Marketing therefore has to advocate for access, linkage, and ownership. This is not an attempt to replace the CIO. It is a business leadership responsibility: if executives expect marketing to improve growth and demonstrate returns, they must support the data foundations that make targeting, budget allocation, attribution, testing, and AI-powered personalization possible.

2. Business Value Comes Before Technology: Ari, who works on omnichannel personalization at Apple across online channels, more than 500 stores, and contact centers, recommends beginning with outcomes everyone accepts: revenue growth, lower costs, stronger customer experience, or improved NPS. Personalized recommendations can then become a defined route to those outcomes, with unified data positioned as an enabler rather than an abstract technology goal. Dave reinforces this logic because investment will not follow a platform-first argument. Paul adds urgency: an IT-led Adobe program was implementing a view of marketing that was already two years out of date. Marketing must articulate valuable use cases early, before technology teams lock outdated assumptions into the architecture.

3. Data Days Create Executive Momentum: Dave recommends a quarterly data day or half-day in which marketing pauses routine work and examines the data behind its business goals. The team should assess quality, access, robustness, ownership, measurement, experimentation, and the help required from other functions. At one client, the CMO converted the output into an hour on the C-suite agenda with support from the COO. The CEO valued the exercise because it exposed why ambitions for AI, digital operations, and personalization were stalling. Sam’s experience shows that teams may first need basic education: colleagues jumped directly to desired tools, insights, and staffing without identifying the inputs needed to produce those outcomes.

4. Architecture Maps Expose Hidden Waste: At the first data day for one of Dave’s clients, a half-day session ran past lunch because nobody understood how the systems connected. The company had a Snowflake data lake, Tealium CDP, CRM, and an Adobe implementation, but teams used inconsistent data and disputed numbers with sales. Even the CIO organization lacked an integrated view. Mapping the systems revealed both capability gaps and potential savings from simplification. The client, a major manager of apartment rentals, also collected information separately at lead, sale, and resident stages. That fragmentation forced customers to answer the same questions and complete forms again instead of continuing one coherent relationship.

5. Customer Journeys Clarify Fragmentation: Ari suggests mapping data architecture through the customer journey because employees can readily understand one customer moving across moments and touchpoints. The approach reveals where a lead becomes a customer but also enters an entirely different system. It reminds teams that customers do not perceive ten departments, platforms, or data owners. Alex Chong described the consequences of ignoring entry points at a multi-billion-dollar company. After buying a customer data platform, the business discovered customer records with ten last names and 20 email addresses. Newsletter sign-ups could be repeated using several addresses, leaving no dependable source of truth. Fixing the resulting mess took years.

6. Instrumentation Turns Creative Into Data: Dave distinguishes measuring an entire asset from learning which components make it effective. At Aetna, every outbound item was tagged against 25 attributes, including color, font size, background image, and the presence of specific words. Platforms such as Adobe can support this, while newer AI tools can tag batches of creative automatically. The opportunity grows with generative AI: teams can produce five variants quickly, but those variants generate useful learning only when their differences are recorded. James offers evidence of the payoff. His organization produced roughly 12 research assets annually, but only one mattered, and about 80% of even that asset was noise; two images in six pages drove interest.

7. Measurement Changes Internal Power Dynamics: Granular measurement can challenge work backed by large budgets, agency confidence, or internal reputation. Ari notes that attribution and telemetry may show that a heavily supported tagline, campaign, or asset performs poorly. James encountered the same tension when researchers learned that most of their output did not influence customers. The answer is not to retreat from measurement but to position evidence as a method for improving future work. Paul warns that delivery culture can still defeat learning: one Adobe implementation team rejected new data requirements because it had April milestones and suggested waiting 18 months. Organizations need success measures based on actionable learning, not schedules alone.

8. Customers Can Supply Better Signals: Marketing can create valuable data instead of only integrating existing databases. Dave contrasts indiscriminate retargeting with direct questions. After viewing drill bits on Grainger’s website, he received daily Gmail promotions covering unrelated areas such as security and cleanup, and clicking a promoted product led to a general catalog page. Grainger could instead ask what he needs. At Aetna, Dave and Tim used five simple “Is this more important than that?” questions in Medicare marketing. The conjoint-style exercise captured customer trade-offs and produced personalization data more powerful than other available sources. This zero-party data can also strengthen attachment when the brand visibly uses the answers.

9. Call Centers Are Marketing Sensors: AI-enabled call center systems can digitize and interpret customer conversations, but their output often remains inside service operations. Dave identifies two useful levels. Aggregate analysis can reveal competitor offers, recurring product confusion, and topics requiring education. Customer-level analysis can flag attrition risk, difficulty using a product, or a life change that makes a cross-sell relevant. He cites DataOrb as one company working with telecommunications providers, while stressing that a special tool is not always necessary because major contact center platforms increasingly include these capabilities. Marketing’s job is to specify the signals it needs and create a route into CRM, targeting, and content workflows.