Back to Insights
·5 min read·Adam Roozen

Residue

Nearly 30 years of digital commerce waves, and the layer they all settled into

Let's look at nearly 30 years of digital commerce waves together. Early e-commerce. The mobile revolution. Social commerce experiments. And now the AI wave. I've been building and advising in this space since before most people had broadband internet. So the watching part of this is firsthand. The interpreting part is where I'd hedge.

Notice what each wave shared. It showed up with roughly the same announcement (everything is different now) and it left the channels relabeled. E-commerce, mobile, social, marketplace, DTC. The labels kept changing, but the transaction, the cart, and the message kept happening underneath every one of them.

Let's look at how that works. Then let's take it out of commerce entirely.

The waves

The wave is all anybody talks about, and to be fair, waves are real. Mobile put the store in everyone's pocket and behavior genuinely changed. Social moved the conversation into a different room. Now AI is here, and honestly, it's early. Maybe VERY early.

But once you've watched a few waves come through, you stop watching the wave itself. I start watching for the same categories that keep settling under each new label.

A working definition

I want to give that something a name, because names make patterns portable, and portable is the point of this whole exercise. Here's my working definition:

"Residue is what a wave leaves on the floor after it passes."

Residue is whatever settles while everyone else is still watching the surface. In digital commerce, as far as I can tell, the residue comes in three kinds:

(a) transactions
(b) abandoned carts
(c) customer interactions

Every transaction, every abandoned cart, every customer interaction. All data. All sinking to the same floor, wave after wave. Or at least that's how it looks from where I've been sitting.

One shop, walked slowly

All of digital commerce is too big for a specimen, so let's shrink it to one shop. Marisol sells coffee gear from her own site – she's made up, but the pattern isn't. Let's walk her through the waves and count what settles.

First wave. A customer finds Marisol from a desktop computer, probably at home, probably patient. She buys a kettle: that's a transaction, residue type (a). A second shopper loads a cart, gets distracted, closes the tab: abandoned cart, type (b). A third emails to ask whether the grinder handles oily beans: customer interaction, type (c). One wave, three kinds of settling, and Marisol gets to keep all of it.

Now mobile rolls in. What actually changed? Mostly the screen, and the patience. The kettle still sells, just from a couch instead of a desk. The cart still gets abandoned, maybe faster this time, mid-errand. The question still arrives, thumb-typed now. New channel label, same three settlers.

Then the social commerce experiments moved the conversation into yet another room, and marketplace and DTC stacked more labels onto the channels. I'm not saying nothing changed up top. Plenty did. I'm saying the floor kept collecting the same categories the whole way through.

Now the AI wave, which is the one we're actually standing in. It looks like a shopper might soon send an assistant out to compare kettles and check out without ever visiting Marisol's site at all. Maybe that's already happening somewhere. If it is, a transaction still settles for her. But when the customer never shows up and never speaks, who collects the interaction? I don't know, tbh. There might even be a fourth kind of residue out there that nobody's named yet. It's early.

(I've misread waves before, fair warning. The ones I felt most certain about sometimes left the least behind.)

Other floors

Alright. Let's pull the pattern off digital commerce entirely and set it down somewhere else. A couple places come to mind.

Here's a smaller example, and I'll invent the person the same way I invented Marisol: Dr. Ellis, a doctor a few decades into practice. Every patient arrives like a new wave. New face, new story, new worry. Walk one patient through her office and watch what settles. The patient describes her symptoms, and the symptoms settle. Dr. Ellis picks a treatment, it works or it doesn't, and the response settles. The patient recovers, or she comes back, and the outcome settles. Symptoms, responses, outcomes. That's Dr. Ellis's version of (a), (b), and (c). A younger doctor reads a chart with mostly training behind it. Dr. Ellis reads the same chart with training plus decades of settled cases. I'd expect those two readings to differ, sometimes by a lot.

Or a coach. Call him Coach Reid. Every season delivers a new roster: new wave, new personalities, new strengths. Walk one of his games through the same counting. Possessions pile up, one after another. Somebody blows the same defensive assignment twice, and that settles. The film gets stored, and film from a rough night early in the season is still doing teaching work weeks later. Possessions, missed assignments, game film – Coach Reid's residue. The floor gets a little thicker every season. When he watches September film in November, he's reading the missed assignments alongside the outcomes. I think that is the floor doing the coaching, not the current roster.

Consider this: weather versus climate. Day to day, everything looks different. A storm, a headline, an oddly warm week. The settled layer is climate, and it's the one that tells you what actually changed. Climate is the residue of daily weather, collected over decades.

The open question

Here's the part I keep circling. I think the reason Marisol's floor means something to her is that she was there when each abandoned cart or question arrived, so she knows which transactions came from which interactions. Say she emailed me her complete history tomorrow: every transaction, every abandoned cart, every message. Reading her export might show me the transactions, but I probably wouldn't know which customer questions preceded them, because I wasn't in the room when the questions arrived.

And there's one thing I genuinely don't have an answer for. This AI wave might be the first where some of the watching gets handed to machines that never get tired. Collecting the data, sure, that part looks easy. Whether a machine can sit with a floor long enough for the residue to mean anything, the way it seems to mean something to Marisol or Dr. Ellis – I'm not so sure. I don't have that one worked out.

Written by

Adam Roozen

Strategic Advisor. AI Strategy, Digital Commerce, Technology Transformation

Nearly 30 years of operating experience · Walmart · Sam's Club · Echidna

Work with Adam