By: Eva Keller
A customer walks into a store and spends $45. Ask most owners what she’s worth, and they’ll tell you $45, because that’s what the receipt says. According to Tim Shea, founder and CEO of the Los Angeles-based data science agency Latticework Insights, that answer is the most expensive mistake in retail, and it’s one almost every store owner makes without realizing it.
A receipt shows what a customer spent today. It says nothing about what she’s worth. Tim has spent 25 years building data infrastructure for brands including the NBA, UFC, Sweetgreen, and Reddit, and he says the gap between those two numbers, spend-to-date versus true value, is where most retailers leave money on the table.
“Your LTV is not $100 just because a dashboard says so,” Tim says. “You don’t have one type of customer. You have many, and most businesses are averaging them together, which erases the exact information that would tell you where to spend.”
The industry habit is to treat every dollar of revenue the same. Tim argues that’s backward. The number that matters isn’t what a customer spent once. It is what she’s likely to spend across her relationship with the business, based on how customers like her have actually behaved.
That distinction sounds academic until you see the math. Tim says a common figure he sees across e-commerce and DTC brands is that roughly 80 percent of customers never make a second purchase. The 20 percent who do are carrying the business, and inside that group, the timing of the second purchase predicts almost everything that follows. Customers who buy again within 30 days are far more likely to reach a third, fourth, and fifth purchase than customers who wait 60 or 120 days for the next sale.
“We’ll look at a company’s data and find that half their revenue comes from customers who bought three times or more,” Tim says. “Get someone to a second or third purchase, and a meaningful share of them go on to buy ten things. That’s not a rounding error. That’s the business.”
This is why Tim built what he calls the SMART framework: speed, margin, attribution, retention, and tiers, as the starting exercise for every client. Speed asks how quickly a customer buys again. Margin asks what it actually costs to acquire and serve her, not just what she paid. Attribution asks where she came from. Retention asks what it takes to get her to buy again. Tiers asks us to accept that we do not have only one type of customer, because a $45 customer who needs a discount to buy and a $450 customer who buys every few months should never be averaged into a single number.
The obvious objection is that this kind of analysis used to be expensive. A retailer who wanted real cohort-level LTV data faced a six-to-twelve-month build, a data warehouse, a pipeline, a hired data science team, a business intelligence layer, often running well into six figures before a single insight came out the other end. For a small or mid-sized retailer, that cost alone was reason enough to keep guessing.
Tim says that barrier is the thing that’s actually changed, not the underlying math. AI has collapsed most of the stack that used to require a dedicated technical team to build. A retailer doesn’t need to buy a data warehouse before asking a useful question about customer behavior anymore. She needs someone who already knows which questions separate a $45 customer from a $450 one, and the tools to answer them quickly.
“The math hasn’t gotten easier,” Tim says. “The cost of getting to the math has.”
None of this requires a data science degree to start. Tim’s advice for a retailer this week is narrow and specific. Pull last year’s purchase data and look at one thing: how long it took repeat customers to buy the second time. Customers who came back fast are worth studying, not just celebrating. They’re the pattern the rest of the business should be built around, and finding them no longer takes six months or six figures. It takes asking the right question of the data that’s already sitting there. It’s the same starting point Latticework describes in its breakdown of how elite brands use LTV:CAC analytics to succeed.
About Tim
Tim Shea is the Founder & CEO of Latticework Insights, working at the intersection of Data Science & Retail for 25 years. Tim works extensively with technologies such as Snowflake, Fivetran, and LangChain, as well as LTV:CAC, Retention, Financial Forecasting, and Growth Analytics. Tim’s clients include Sweetgreen, NBA, UFC, Converse, Princess Cruises, Reddit, Quora, and Salesforce.
