When Amy Avery walked into her Bank of America interview, she was not sold by a title, a compensation package, or a corporate mission statement. She was sold by a number: 67 million. That was the count of clients the bank was actively interfacing with at the time — a figure that landed on Avery with the weight of pure analytical possibility. "Gosh, that's so much information," she recalls thinking. "Think about what you could do with it." Years later, as Managing Director of Analytics, Modeling and Insights, Avery is doing exactly that — and the architecture she and her colleagues have built around that data mountain is becoming one of the more instructive case studies in how legacy financial institutions are attempting to operationalize artificial intelligence at genuine scale.
The framing Avery uses to describe her internal working relationships is deceptively simple but analytically precise. In the partnership model she operates within, one side owns "the what" — the domain knowledge, the business questions, the strategic priorities embedded in the data — and the other side owns "the how" — the methodology, the modeling frameworks, the technical execution that transforms raw information into actionable intelligence. It is a division of labor that sounds obvious until you consider how frequently it collapses in practice inside large financial institutions, where data scientists and business leaders routinely talk past each other in expensive and time-consuming ways.
That 67 million client figure is not merely a recruitment anecdote. It represents one of the largest proprietary behavioral data sets in global retail and commercial banking — a corpus that spans checking and savings behavior, credit utilization, investment patterns, mortgage lifecycles, and the millions of daily transactional signals that accumulate across a client base of that magnitude. For an analytics leader, it is both an extraordinary resource and an extraordinary responsibility. The scale creates competitive advantage, but it also demands disciplined governance, rigorous modeling standards, and a clear organizational philosophy about what questions are worth asking in the first place.
This is precisely where the "what versus how" framework earns its keep. In many financial institutions, artificial intelligence and machine learning initiatives stall not because the technical capability is absent but because the business definition of the problem is underspecified. Data teams build models in search of a question. Avery's approach — rooting the analytics function in domain-defined priorities before methodological choices are made — inverts that failure mode. The business side arrives with the "what," and the analytics and modeling function deploys the "how" in service of that pre-defined objective. The result, in principle, is a tighter feedback loop between strategic intent and quantitative output.
Bank of America's investment in data and artificial intelligence infrastructure has been an ongoing institutional priority, and Avery's role sits at the center of that effort. The bank's client scale — now extending well beyond the 67 million figure that first caught Avery's attention — means that even incremental improvements in the precision of analytics models can translate into meaningful outcomes across credit decisioning, customer retention, product personalization, and risk management. At that volume, a fractional gain in model accuracy is not an academic distinction; it is a business result measured in hundreds of millions of dollars and millions of client interactions.
The broader industry context makes this kind of internal clarity especially valuable. Across the banking sector, the competitive pressure to deploy artificial intelligence has intensified dramatically. Institutions from JPMorgan to Revolut have made public commitments to AI-first operations, and the talent competition for analytically sophisticated leaders has grown correspondingly fierce. Avery's own career trajectory — drawn into banking by the sheer informational gravity of a 67-million-client data set — reflects a broader shift in how top analytical talent now evaluates financial institutions: not by prestige alone, but by the quality and scale of the data problems on offer.
What This Means for the Industry
The significance of Avery's account extends beyond Bank of America's internal organizational design. It illustrates a maturation in how major financial institutions are thinking about the relationship between data infrastructure and strategic execution. The era of deploying AI as a marketing claim — bolting machine learning onto legacy workflows without rethinking the underlying problem architecture — is giving way to something more structurally serious: clearly delineated roles, domain-first problem definition, and analytics functions that are genuinely embedded in business decision-making rather than siloed alongside it.
For the 67 million clients whose financial behavior underpins this entire enterprise, the stakes are concrete. Better-calibrated models mean more appropriate credit offers, more relevant financial guidance, and risk frameworks that are responsive rather than reactive. The ambition Avery brought to her interview — the instinctive recognition of what could be done with information at that scale — is, in effect, the animating idea behind Bank of America's broader AI strategy. Whether the institution can fully deliver on that ambition across the complexity of a global balance sheet remains the open question. But the analytical leadership architecture being described suggests the bank is, at minimum, asking the right questions in the right order.
Written by the editorial team — independent journalism powered by Codego Press.