Starting an online shop: what actually matters in the first 90 days
Most new shops spend their first three months on the wrong things. Here is the order of work that gets you to a first repeatable sale instead of a beautiful, empty storefront.
Photo: free stock photography (Unsplash licence) — see imprint
The first 90 days decide almost nothing about design
Every founder we meet in month one wants to talk about the logo, the hero image and the colour of the buy button. Almost none of them want to talk about who is going to pack the parcel, what happens when a customer returns it, or which of their 400 products actually has a description worth reading. Three months later the shop looks superb and has eleven orders, nine of which came from friends.
This is not a design problem. Design matters — but it multiplies whatever is already there. A great storefront on top of unclear products, unclear delivery and unclear pricing multiplies zero. The first 90 days are for finding out whether anybody wants the thing at a price that leaves you money, and everything else is decoration on that question.
Days 1–30: sell something badly
The goal of month one is a single real transaction from a stranger. Not a pretty shop — a transaction. Take your ten best products, write honest descriptions, set a price you can defend, connect one payment method and one shipping option, and open. It will be ugly. Ugly and live beats beautiful and theoretical, because only a live shop tells you what you got wrong.
Ten products is not a limitation, it is a filter. If you cannot write a paragraph about why someone should buy a product, that product is not ready to sell — and a catalogue of 400 items you cannot describe is 400 pages Google will rank for nothing. Start narrow, learn what sells, widen from evidence.
Days 31–60: fix the leaks you can now see
With real traffic you finally have real data, and the data is almost always rude. People add to cart and vanish at the shipping cost. They read the product page for forty seconds and never scroll to the buy button. They search for a term your catalogue does not use. None of this was visible in the mockup, and all of it is fixable in an afternoon once you know.
The single most common leak in a new shop is a surprise. A shipping cost that appears at step three, a delivery time that turns out to be 'three weeks', a payment method your customers do not have. Surprises kill carts far more reliably than an unfashionable font. Put every cost and every date on the product page, before the cart, and watch the abandonment rate drop without touching the design.
Days 61–90: make one channel work before adding a second
By month three the temptation is to be everywhere: Instagram, Amazon, Google Ads, a newsletter, a blog, a marketplace. Everywhere is nowhere on a founder's calendar. Pick the one channel where your buyers already are, and work it until it either produces repeatable orders or clearly cannot. One channel that works funds all the others; five channels at 20% effort fund nothing.
This is also the moment to be honest about the numbers. If a sale costs you 18 euro to acquire and leaves 12 euro of margin, no amount of traffic fixes that — you have a pricing or a product problem wearing a marketing costume. Work out your real margin per order, including packaging, returns and the payment fee, before you spend a cent on ads.
What to postpone without guilt
Some things feel urgent and are not. A custom theme, a loyalty programme, a mobile app, a headless architecture, a full ERP integration — every one of these is a real project with a real bill, and every one of them is cheaper and smarter to do once you know what sells. Nothing here is wrong forever; it is wrong now.
- Custom theme — a clean standard theme sells fine for the first year.
- ERP integration — do it when manual entry actually hurts, not before.
- Headless / app — needs a proven catalogue and real traffic to pay off.
- Internationalisation — one country, one tax model, one shipping story first.
| Phase | The only goal | What to ignore |
|---|---|---|
| Days 1–30 | One order from a stranger | Design, catalogue size, channels |
| Days 31–60 | Remove surprises from the funnel | Ads, marketplaces, redesigns |
| Days 61–90 | One channel producing repeat orders | Everything that is not that channel |
| Month 4+ | Scale what is proven | Nothing — now you may build |
- Go live ugly in month one — only a live shop tells you what is wrong.
- Surprises in the funnel cost more orders than any design decision.
- Know your true margin per order before you buy a single click.
- One working channel beats five half-worked ones, every time.
Frequently asked questions
Ten, if you can describe all ten well. The constraint is not the shop system — it is how many products you can honestly explain. A catalogue you cannot describe is a catalogue that will not rank and will not convert, and it is far easier to add products later than to fix 400 thin pages.
In the first 90 days, almost any established system will do — the differences that matter to you now are tiny. Choose based on where you expect to be in year two: heavy B2B pricing rules, ERP integration and custom logic point to Shopware, while a simple B2C catalogue is well served by anything hosted. Do not let the system choice delay going live by two months.
Only a small test budget, and only once you know your margin per order. Ads are an amplifier: they make a working funnel bigger and a broken funnel expensive. If you do not yet know what a customer costs and what they leave behind, you are not buying traffic — you are buying a lesson at full price.
When a standard theme is demonstrably holding you back — not when you are bored of it. That usually means you have traffic, you have conversion data, and you can name the specific thing the theme prevents. Before that point a custom theme is an expensive way to feel productive.
We do this for a living — Shopware, Node.js, React, ERP integration and automation for B2B.
Talk to an engineer