By: Eva Keller
Every retail brand has one. Most have thirty. A dashboard full of lines and percentages that someone checks after a bad week, ignores during a good one, and argues about in the quarterly review when two people pull the same metric and get different numbers.
Tim Shea has spent 25 years watching this happen at companies ranging from Pepsi and the NBA to direct-to-consumer brands burning through their first round of venture money. His diagnosis is consistent: the dashboard is not the problem. The dashboard is what happens after you fail to solve the problem.
“Everyone says, okay, we’re willing to invest in a Tableau dashboard,” says Shea, founder and CEO of Latticework Insights. “But can everyone agree on what metrics matter for the company? That turns out to be the hard part.”
The hard part is also the part no vendor sells you.
Most retailers approach data as an infrastructure challenge. They consolidate platforms, migrate warehouses, and build visualizations. They check the boxes. What they rarely do is answer the prior question: what are we actually trying to move? Tim finds that most leadership teams cannot reach consensus on something as fundamental as their own customer lifetime value. Not because they lack data, but because LTV isn’t a number, it’s a story about which customers you are counting, across what time horizon, net of which costs. That story has as many versions as it has stakeholders.
The result is a room full of smart people staring at the same dashboard and disagreeing about what day it is.
This is what Tim calls the data culture problem, and in his estimation, it is the most expensive line item most retailers never see. A company can run thirty simultaneous A/B tests and still make no decisions if the team cannot agree on which metrics constitute success. The modern data stack, the warehouses, the pipelines, the visualization layer, is only as useful as the organizational consensus underneath it. Without that consensus, another dashboard is just another surface for the disagreement to play out on.
The misread goes deeper when AI enters the conversation. Retailers are being told, credibly and loudly, that agentic commerce and LLM-powered analytics will resolve these ambiguities for them. Tim is skeptical, not of the technology, but of the sequencing. “People say, what should our AI strategy be? And I’m like: do you have your data stuck in twenty different platforms? Because you need to solve that first.” AI applied to fragmented, unaligned data does not produce insight; it produces faster noise.
Where Tim does see genuine AI value in retail is narrower and more concrete than the headlines suggest. AI-generated code, reviewed and tested by experienced engineers, can produce custom dashboards that are deterministic, stakeholder-specific, and far more credible than anything built in a generic BI tool. Agentic workflows can accelerate pattern recognition and exploratory analysis, surfacing thirty potential interventions in an hour rather than a quarter. But these are multipliers.
What they multiply matters enormously. A brand’s highest-value customers carry disproportionate weight, and focusing retention efforts on that group is where the real leverage sits. That kind of leverage requires knowing which customers to target and having an organization aligned around pursuing that specific number, not a suite of AI tools and a crowded analytics portal.
The retailers Tim describes as elite are not the ones with the most sophisticated data infrastructure. They are the ones running A/B tests, actually running them, not planning to, and acting on the results. That capability requires speed, margin clarity, attribution discipline, and a shared language about what the business is optimizing for. It requires, in short, what the dashboard was supposed to represent but almost never does: a common understanding of what success looks like.
The fix is not a better visualization. It is an agreement about what you are trying to see.
