Why New York Life's CMO Built a Data Foundation Before Deploying AI

New York Life’s CMO Amy Hu spent three years building data infrastructure, in-house analytics capability, and measurement culture before deploying AI at scale. The sequence worked because organizational capability determined what the technology could deliver.

The Failure Rate Nobody Talks About at AI Conferences

Most organizations approach AI as a procurement decision. Find the right platform, negotiate the contract, train the team, expect results. RAND Corporation research tells a different story: more than 80% of AI projects fail to deliver expected value, with data quality and organizational readiness as the consistent root causes (1. RAND Corporation, 2025). The technology works. The organizations deploying it usually haven’t built the foundation to make it work for them.

Amy Hu, CMO of New York Life, made a different first move. She invested in data. What her team built over three years at one of America’s oldest financial institutions is a case study in sequencing: what organizational capability to build before the technology question comes up.

Three Years Before the AI Conversation

Hu’s team spent roughly three years building capability before AI deployment became a serious conversation. The investment went into areas that don’t generate conference keynotes: data infrastructure that connected systems operating in silos, in-house data science replacing agency and vendor black-box models, media mix modeling built internally rather than licensed from a platform.

The measurement piece was the linchpin. On the Building Better CMOs podcast, Hu described her goal as transforming marketing from a cost center into a demonstrable profit driver (2. Building Better CMOs, 2026). That framing shaped every hiring decision and infrastructure investment her team made.

In practice, that meant her team could walk into the CFO’s office with models they’d built themselves, showing what marketing spending produced. In-house models calibrated to New York Life’s sales cycles, distribution channels, and customer segments. The finance team trusted the numbers because the marketing team had earned that trust over years of validated measurement.

Insurance presents a harder attribution challenge than most industries. Sales cycles stretch across months or years. Customer journeys cross agent interactions, digital touchpoints, and brand awareness that’s hard to isolate. Building models for these realities required domain expertise no vendor platform supplies out of the box.

Data literacy extended across the marketing organization, not confined to a specialized analytics group. Measurement became a cultural expectation. When someone proposed a campaign, the conversation included how they’d prove it worked. That discipline reshaped how the entire function operated.

What Changed When AI Arrived

When New York Life deployed AI capabilities, the technology landed on prepared ground. The data infrastructure was already connected. The analytics team already knew how to evaluate outcomes. Leadership already trusted marketing’s numbers because those numbers had been validated for years.

AI accelerated what the organization already knew how to do. Models got faster. Customer insights became more granular. Optimization cycles shortened. The acceleration was productive because it amplified real capability rather than papering over gaps.

The foundation also positioned the team to move on emerging challenges that less prepared organizations haven’t touched. Hu described her team’s work on answer engine optimization and generative search. Their existing measurement infrastructure gave them an edge: they could test, measure, and iterate on new channels instead of waiting for industry consensus.

McKinsey’s State of AI research confirms this pattern isn’t unique to New York Life. Organizations that qualify as AI high performers are nearly three times more likely to have redesigned workflows and processes around AI rather than layering it onto existing operations (3. McKinsey, 2025). Only about 6% of organizations meet that bar. The remaining 94% are trying to extract value from AI without having built the organizational muscle to use it.

The same dynamic plays out in marketing broadly. The Content Marketing Institute’s 2026 B2B research found that team skills and content quality ranked as the top drivers of marketing effectiveness, above technology investment (4. Content Marketing Institute, 2026). Capability drives results. Technology amplifies them.

The Uncomfortable Math

Hu’s approach produced something most CMOs can’t demonstrate: concrete, finance-validated evidence that marketing investment drives business outcomes. She can sit across from the CFO with proof. That’s a competitive advantage most marketing leaders would trade half their stack for.

But the approach took three years. Three years of data infrastructure that didn’t produce AI headlines. Three years of hiring and developing analytical talent. Three years of building measurement credibility one quarter at a time. That timeline doesn’t survive a quarterly board review if the only metric is visible AI deployment.

Most organizations want AI results without that investment. Vendors encourage the expectation. The 80% failure rate tells you how that works out.

If your organization is evaluating AI for marketing, the first question is whether you’ve done the prerequisite work. Can your team measure what a new platform produces? Is your data clean and connected enough to feed it? Does your organization have the analytical talent to interpret what comes back and act on it? Those are capability questions. The answers determine whether AI investment produces returns or produces spend.

New York Life didn’t deploy AI and get lucky. They spent three years making sure luck wasn’t required.