In 2005 I launched MarketOnline. Back then, to a lot of people an “online store” sounded like something halfway between science fiction and a scam. Friends asked me, with the best intentions, who on earth would buy something they couldn’t touch, from a company they had never seen. It was a fair question. The answer took us years.
More than twenty years have passed since. I went through MarketOnline, Mediadot, Azerty, Floria, then Retargeting, where I worked every day with online stores that wanted to sell more. Looking at e-commerce in 2026, I only half recognise it. So I tried to write down what has really changed and, more importantly, what has stayed exactly the same.
Trust: from fear of cards to one-click payment
In 2005 the number one problem wasn’t price, it was fear. People didn’t want to type their card details into a website. Many didn’t use their card for anything other than withdrawing their salary. So cash on delivery was the rule, not the exception: the customer saw the parcel, paid the courier and only then felt safe. For the store, that meant money stuck in transit for days and parcels refused at the door.
Today paying by card is routine, digital wallets on the phone have made it almost invisible, and younger shoppers don’t even think about it. And yet in Romania cash on delivery is far from dead. It still accounts for a significant share of orders, especially outside the big cities. The lesson: don’t force customers to trust you on your terms. Give them the option they feel safe with.
Traffic: from web directories to AI-generated answers
Our first customers came from website directories, price comparison sites and a far simpler Google than today’s. A good title and a few links were enough to rank. Back in 2009 I was writing about things like the search algorithm in online stores, and what struck me then was how cheaply you could bring in a visitor.
Today traffic is fragmented: Google, social networks, marketplaces, short video, influencers and, for the past year or two, AI-generated answers that sometimes recommend a product without the person ever reaching your site. Attention is more expensive, and relying on a single source is more dangerous than ever. One algorithm change overnight can cut your sales in half.
Logistics: from “next week” to “in a few hours”
In the MarketOnline days, couriers were slow, expensive and few. Delivery in two or three days was an achievement, and customers accepted it because they didn’t know any better. Today next-day delivery is standard, parcel lockers and pickup points have changed shopping habits, and quick commerce promises products in under an hour. I wrote recently about quick commerce in Romania: it’s spectacular, but its economics are fragile without volume and density.
Platforms: from building everything to a monthly subscription
In 2005 an online store meant code written from scratch, your own server, hand-built integrations and a developer you called in the middle of the night. Today you can open a store in a few days on a SaaS platform, with payments, invoicing and shipping plugged in from a menu. Or you can sell directly on a marketplace without a site of your own.
The barrier to entry has dropped enormously, which is wonderful. But it also means the difference no longer comes from technology. Everyone has the same shopping cart. The difference comes from the product, the brand and the relationship with the customer.
Data: from retargeting to privacy
Between 2013 and 2018, at Retargeting, I lived through the golden age of data: you could follow a visitor, show them the product they had left in their cart and send them the right message at the right moment. It worked brilliantly and, to be honest, sometimes it was too much.
Then came GDPR, browser restrictions and the debate about third-party cookies going away. Today the data that matters is the data customers give you willingly: their email, their account, their order history, their preferences. Stores that built their own customer base sleep better than those that rented everything from the ad platforms.
Mobile and customer expectations
We started on desktop, with sites designed for big monitors. Today most visits come from phones, and a checkout that doesn’t work smoothly on mobile is simply money lost. Mobile-first is no longer a trend, it’s basic hygiene.
Customers have changed too. They expect fast delivery, easy and often free returns, genuine reviews from other buyers and an answer within minutes, not days. What was a “wow” in 2005 is today’s bare minimum.
AI: from product descriptions to agents that shop
Artificial intelligence has already slipped into online stores through the back door: product descriptions written in seconds, assistants answering questions at any hour, personalised recommendations that actually make sense. What we dreamed of doing with personalisation ten years ago now takes a fraction of the effort.
The next step is more interesting: AI agents shopping on the customer’s behalf. Someone says “find me a good pair of running shoes under a certain budget that will arrive by Friday”, and the agent compares, chooses and pays. If that becomes common, a store won’t be selling only to people but also to an algorithm that reads specifications, prices, stock levels and return policies. Clean, accurate product data will matter more than beautiful photos. It is one of my predictions for 2026.
What hasn’t changed at all
Take the technology out of the equation and e-commerce in 2026 looks strikingly like it did in 2005. The same things kill or save a store:
- Margin. If you sell without margin in order to grow, all you grow is the loss.
- Cash flow. Stock, ads and payment terms can bury you even with strong sales.
- Customer service. Someone who gets quick help when something goes wrong becomes a loyal customer.
- Returning customers. The second-time buyer is the one who makes the profit.
- An honest product. You can fool a customer once. Reviews take care of the rest.
What I would tell someone opening an online store in 2026
I wrote at more length about starting an online store from scratch, but in short: pick a niche you understand, not one that looks fashionable. Start on an existing platform rather than building technology. Try a marketplace for volume, but build your own customer list from day one. Calculate your margin after returns, shipping and advertising, not before. Offer cash on delivery too, if you sell in Romania. Use AI for repetitive work, not for the relationship with your customers. And don’t put all your eggs in a single traffic channel.
E-commerce has become easier to start and harder to win. In 2005 we were fighting people’s fear of buying online. In 2026 we’re fighting for their attention. What do you think will change by 2030, and what will stay the same?