The future of eBay commerce

I have been building software for eBay sellers since 2006. In that time I have watched the platform reinvent itself four or five times, and watched a lot of good businesses get caught out by changes nobody warned them about.

So before this series starts properly, here is what I actually think. Some of it will annoy people. Some of it will turn out to be wrong, and I will leave it up anyway, because a prediction you quietly delete was never a prediction.

What eBay gets right, and nobody says

It is fashionable among sellers to say eBay has lost the plot. I do not think that is true, and I would rather be accurate than popular.

Authentication is the best thing eBay has built in ten years. We are about to be buried in listings written by machines and photographed by nobody. A marketplace where somebody physically opens the box and checks the watch is holding something the others cannot copy in a hurry.

Category depth is the other one. eBay still sells the weird stuff — the vehicle part with a fitment table, the obsolete industrial fitting, the thing with a market of four hundred people worldwide. Every marketplace that has tried to tidy that away has ended up smaller. And eBay has not thrown business sellers overboard to chase casual ones, which was a choice, and plenty of platforms chose the other way.

What it is getting wrong

The listing form has become a tax on being professional. Every release adds fields. Nothing is ever taken away. It got worse a little at a time, which is exactly why nobody treated it as an emergency, and at two hundred items a month it is an emergency.

Then there is relisting. Everyone knows stale listings do worse. There is still no way to schedule an automatic relist, so every serious seller either does it by hand every month or pays somebody like me to do it. That should just be in the platform. It is not a hard feature. It has been missing for a decade.

And the redesigns keep landing without warning. Seller Hub has moved things professionals click forty times a day, with no notice and no way back. If thousands of businesses run their operation on your screen, you owe them a heads-up before you move the furniture.

The bit nobody at eBay will ever put in an email

Every year, a little less of your sales arrive without a fee attached to them. Nobody announces that. There is no blog post. It happens through a hundred small changes — how ads get credited, what gets promoted where, which listings surface without paying.

I think that continues, and within two years running promoted listings stops being a growth tactic and becomes the cost of being visible at all. Not because anyone at eBay is twirling a moustache — it is what happens to every marketplace once advertising revenue gets big enough to have its own targets.

If you want one number to watch this year, it is the share of your sales that carried no ad fee. Most sellers have never calculated it. It is the most useful figure in the business and it takes an afternoon.

AI, minus the noise

Every marketplace has an AI listing feature now, and almost all the talk about it is wrong in the same way. It is sold as “the AI writes your listing so you do not have to”. That is the least interesting thing it does.

Writing the words was never the hard bit. Checking them is. One AI-written title is fine. Six hundred that nobody read is a catalogue full of confident mistakes, and confident mistakes are worse than obvious ones, because nobody spots them until a buyer does.

My honest guess is that most of the AI seller tools launching this year will not exist in two — not because AI is a fad, but because they are features being sold as products. Anything that only generates text will be free and built in within eighteen months. The sellers who win will not be the ones generating the most. They will be the ones who put a checking step in the middle.

The thing eBay should be doing and is not

I will be direct about my interest before I start: I own a company that does this for a living. Weigh what follows accordingly. I still think it is the most obvious unclaimed win on the platform.

Optimisation is not a finishing touch. It is the foundation. A listing with complete, accurate, structured data — every item specific filled, every attribute right, everything a buyer might filter on actually present — is what every other part of selling on eBay is built on top of. Search visibility sits on it. Filters sit on it. Whether Google shows your listing at all sits on it. Whether the buyer gets the right item first time sits on it. Get the data right and everything downstream works. Get it wrong and nothing downstream can rescue it, however much you spend on ads.

eBay knows this. It publishes guidance about it. It built a report to measure it. And then it leaves the actual work to sellers who are already too busy — people running dispatch, handling messages, sourcing stock and absorbing whatever changed that week. Telling a seller their data is incomplete is not the same as helping them fix it. Most of them already know. They do not have the hours.

So the standard falls short. Not for a few sellers — across whole categories, at scale, permanently. Listings go up missing the fields buyers filter on. Buyers do not find them. Sellers conclude eBay does not work as well as it used to. Costs rise, sales soften, and another cohort quietly drifts off. That churn is not mysterious. It is what happens when you build the whole system on data quality and then make data quality somebody else’s problem.

Here is what I would do if it were mine. Build optimisation into the act of listing itself. Not a report you look at afterwards. Not an export, a spreadsheet and a re-upload. Inside the flow, at the point of creation and across everything already live, rolled out globally rather than piloted in one market for two years.

The result would be near enough immediate. No more missing attributes. Complete, compliant listings as the default rather than the exception. Search visibility, impressions and views up across the board, because the data is finally there to match against. Traffic arriving from outside eBay, because the listings are legible to machines that read fields rather than prose. More transactions at close to zero incremental cost, because none of this is new inventory or new advertising. It is the same catalogue, described properly.

Buyers get the item they wanted, first time. Sellers get sales they were already entitled to. And eBay gets what it has wanted for fifteen years: inventory genuinely comparable to Amazon’s on quality of description, across a catalogue Amazon cannot match on range. I think the effect on GMV would be measured in billions. I would take that bet publicly.

They could. They won’t.

eBay could build this into the listing flow. They have the search data, the category data and the engineers. But they won’t.

They could partner with any of the handful of companies who have spent a decade solving exactly this problem. But they won’t.

They could buy one outright — something tested, working, already running at scale — and switch it on globally inside a year. But they won’t.

I want to be careful here, because this is not laziness and it is certainly not stupidity. eBay is an oil tanker. You plan the turn two miles before you make it, and the decision never rests with one person. By the time a change like this has been proposed, scoped, argued over, resourced, committee’d, piloted in one market and reviewed, the thing it was built to fix has moved.

Even if they started today, they would arrive late.

Which leaves the unglamorous option: buy something already built, already proven, already running. Skip the two miles. But they won’t.

If I am honest with you, I do not think they could, even if someone senior wanted it badly enough. There is too much process and not enough people who believe it matters. A change of this size needs one person willing to be completely unreasonable about it for eighteen months — and large organisations are extremely good at removing exactly those people.

This is what happens when a business loses its entrepreneur. The one who could cut through the process and force a decision, not because they had consensus but because they refused to accept the alternative. What replaces them is a committee. Committees are excellent at avoiding mistakes and hopeless at making decisions. The company starts to behave like a political party — everybody with a stake, nobody with a mandate, every idea negotiated down until it is the version nobody objects to.

You cannot run a business by committee. You can only maintain one.

I will keep saying it anyway, because one day somebody over there will have both the mandate and the appetite, and I would like it to be on record. The offer stands.

The one that will get me in trouble

Half of what my industry has sold as “eBay listing design” for twenty years deserves to disappear, and I include some of our own early work in that. Banners. Decorative frames. Colour schemes that matched a logo and did nothing else. That was never design. It was wallpaper, and a marketplace that increasingly renders listings its own way is right to be squeezing it out.

But that is not what design means, and my industry has let the word get away from it.

Design is whether a buyer looking at one product can see the other fourteen you sell that fit their car. It is whether your categories are generated from your actual live catalogue and update themselves, or whether somebody typed them in once in 2019 and they have been quietly wrong ever since. It is whether the size guide, the compatibility table and the returns terms sit where a buyer looks for them, or three paragraphs down. It is cross-sell that works, navigation that reflects what you genuinely have in stock, and product information laid out in a structure a person can scan in four seconds.

That is worth more every year, not less. Because buyers are the only reason any of this exists. Not the algorithm, not the fee structure, not the software industry I belong to. A buyer who finds the right item quickly, understands exactly what they are buying, and notices the two other things they needed — that is the whole game. Everything else on this page is downstream of it.

So the prediction is not that design is dying. It is that the decorative half is finished and the functional half is about to matter more than it ever has — and that most sellers, and a fair number of my competitors, still cannot tell the two apart.

The next twelve months

AI listing tools stop being a selling point. Everyone will have one. Nobody will win by having one.

Item specifics stop being a project and become a job. Required fields keep spreading, category by category. Anyone treating it as a one-off tidy-up will be back where they started inside six months.

What happens after the sale is where your money goes now. Returns rules, feedback rules and dispute handling have all moved against the seller in the last two years. Most operators are still squeezing the buy price while losing more out the back than they gained at the front.

The next two years

Buyers stop searching and start asking. When someone’s assistant is choosing between three products for them, it does not read your description. It reads the fields — brand, size, compatibility, condition. Every hour spent on prose and no hours spent on specifics is an hour spent preparing for the last decade.

Running one channel becomes the risky option. Not because eBay declines. Because the cost of adding a second has collapsed and the cost of losing your only one has not.

What I would do this year, if I were you

Get your products written down as data, not as sentences. Brand, size, colour, fitment, condition — in fields, in one place you control. If your catalogue only exists as paragraphs inside listings, none of it is available to you.

Work out how long stock sits before it goes live. Almost nobody has the number. It is usually the biggest pile of money in the business that nobody is looking at.

Ask what your export actually contains — before you need it. When inkFrog closed, sellers found the export carried listings but not templates, profiles, rules, drafts or scheduling. The thing that made it too much hassle to leave is the same thing that traps you when leaving stops being optional. Go and check yours this week.

What I am going to do about it

This series. Twenty-five numbered pieces on the problems serious sellers actually have, in the order a business hits them. Every one opens on a problem. Where something I built happens to be the answer, I will say so and you can weigh it accordingly. Where the answer is changing how you work, or using eBay’s own free tools, I will say that instead. And where I do not know, I will say that too, because a series that is right about everything is not being honest about anything.

Twenty years has taught me one thing above all the rest. Anyone can build a business in a good market. The test is what you do when the market changes without warning you. It is about to change again.