What Dead Economists Can Teach Us About the Madness of Modern Ecommerce

I've been thinking about Adam Smith lately, which is either a sign of encroaching middle age or evidence that ecommerce has finally broken my brain. Probably both.

Smith wrote about pin factories in 1776, describing how the division of labor could transform ten workers from producing perhaps twenty pins a day collectively to producing upwards of forty-eight thousand. Revolutionary stuff. The kind of efficiency gain that makes venture capitalists weep with joy even two and a half centuries later.

But here's what Smith understood that we seem to have forgotten: he was describing a production system where the workers could still see the pins. They understood what they were making. The feedback loop between action and outcome was immediate, tangible, measured in hours rather than quarters.

Modern ecommerce has taken Smith's logic and extrapolated it into something he wouldn't recognize.

We've divided labor so minutely that no one can see the whole anymore. The person optimizing checkout flow has never met the person managing inventory. The team building recommendation algorithms doesn't talk to the people handling returns. We've achieved spectacular efficiency in individual components while creating systems that make no sense as a whole.

The Measurement Problem

Friedrich Hayek spent considerable energy explaining why centralized economic planning fails. The knowledge problem, he called it. The information needed to make good decisions is distributed throughout the system, held by thousands of individuals, constantly changing. No central planner can collect and process it all quickly enough to make optimal decisions.

Ecommerce platforms have attempted to solve this through data. Collect everything, measure everything, let the algorithms decide. It's central planning through technology, and it suffers from exactly the problems Hayek identified, just faster.

I've seen companies with dashboards displaying hundreds of metrics, updated in real-time, and somehow no one can answer the simple question: are we making it easier or harder for customers to buy from us this month compared to last month? The data exists, somewhere, distributed across seventeen different tools that don't quite talk to each other properly. The knowledge problem hasn't been solved by big data. It's been obscured by it. We're collecting measurements at a volume that makes meaningful synthesis impossible, then making decisions based on whichever metric happens to be declining most dramatically this week.

The Tyranny of Velocity

Every ecommerce business I encounter is optimizing for speed. Ship faster. Update inventory more frequently. Push changes to production multiple times per day. Test, iterate, deploy. The velocity is intoxicating. It feels like progress.

But speed without direction is just expensive motion. I watch companies implement same-day delivery without first ensuring that next-day delivery actually works reliably. I see businesses adding new payment methods before they've sorted out why three percent of transactions are failing mysteriously. The assumption is that more is better, faster is superior, and any pause for calibration is weakness.

We're so busy optimizing for speed that we've created systems where the majority of people who start to buy something give up partway through. That's not an optimization problem. That's a fundamental design failure.

The paradox is that speed makes us slower. When you're moving fast enough that you can't see what you're breaking, you end up spending more time fixing things than if you'd built them properly in the first place. Returns, customer service inquiries, failed transactions, abandoned carts. These are all symptoms of systems optimized for velocity rather than function.

Disconnected by Design

The average enterprise ecommerce operation runs on somewhere between fifteen and thirty different systems. Payment processing. Inventory management. Order management. Warehouse management. Customer relationship management. Email service providers. Analytics platforms. A/B testing tools. Personalization engines. Each one best-in-class for its specific function. Each one optimized independently.

None of them talk to each other properly.

I mean, they technically integrate. APIs exist. Data flows from one system to another with only occasional catastrophic failures. But the people using these systems don't talk to each other. The teams are separate. The metrics are separate. The incentives are separate.

Apparently companies waste up to 25 percent of their technology budgets on redundant or poorly integrated systems. That's not a rounding error. That's a quarter of your technology investment producing friction instead of function.

Smith's pin factory worked because everyone could see the pins being made. Modern ecommerce is more like eighteen separate factories, each making a component, none of them entirely certain what the final product is supposed to look like.

What Gets Measured Gets Gamed

Here's what happens when you optimize systems in isolation: each team hits their numbers while the overall experience degrades. The merchandising team optimizes for average order value, so they implement minimum order thresholds for free shipping. The logistics team optimizes for cost per shipment, so they slow down delivery times. The customer service team optimizes for handle time, so they resolve tickets quickly without actually solving problems. The product team optimizes for conversion rate, so they make the checkout process more aggressive.

Each metric improves. Each team celebrates. Customer satisfaction declines.

Hayek understood this intimately. His work on spontaneous order argued that the most effective systems emerge from the distributed knowledge of many individuals making local decisions. But that only works when the people making decisions can see the consequences of those decisions.

In modern ecommerce, the distance between decision and consequence has become so vast that the feedback loop is essentially broken. By the time you realize that the new checkout flow is causing problems, you've already moved on to optimizing something else.

The Correction

I'm not arguing for a return to some imagined simpler time. Scale matters. Efficiency matters. The ability to serve millions of customers requires systems that can operate without constant human intervention. But efficiency in service of what, exactly? More transactions? Higher valuations? Or actually serving customers well?

What would ecommerce look like if we measured success not by how quickly we could get a package to a customer, but by how often we got them exactly what they actually needed? If we valued the reduction of returns as highly as the increase of orders? If we designed systems where the people making decisions could see the consequences of those decisions?

The economists can't tell us how to build better ecommerce systems. They died long before anyone imagined such things. But they can remind us that efficiency is not the same as effectiveness. That speed is not the same as direction. That you can optimize yourself right into dysfunction if you're not careful about what you're optimizing for.

Smith's insight about the division of labor was profound. But he was describing a system where the division served a clear, visible purpose. Modern ecommerce has divided everything so thoroughly that we've lost sight of the purpose entirely.

Somewhere, I imagine Adam Smith looking at a modern fulfillment center and nodding approvingly at the division of labor, right up until he asks what exactly we think we're accomplishing with all this efficiency. And then the nodding stops.

The question isn't whether we can move faster. We've proven we can. The question is whether we should. And whether, in our rush to optimize everything, we've forgotten to ask what optimization is actually for.