The webshop is not the hard part. The hard parts are your product data, your picking, and your cold chain. Selling groceries online doesn't add a sales channel — it changes how your shop works. This guide covers what actually lands on you, in the order it lands.

Is it worth it?

The honest answer: it depends whether you treat it as extra revenue or as a second operation. An online channel that runs alongside everything else, owned by nobody, costs money and creates friction. An online channel with clear processes is a second business using the same stock and the same floor space.

Two numbers for orientation. The average online grocery basket is considerably larger than the in-store one — customers who plan and have it delivered shop for several days at once. At the same time the margin per order is lower, because picking and delivery consume labour that in-store the customer provides for free. The channel works when the bigger basket more than covers that labour. Which is exactly why picking efficiency decides profitability, not shop design.

One concrete reference point: Frox, a regional grocery business in Bolzano, did around €550,000 in online revenue in 2025 — with a single employee. That isn't evidence it's easy. It's evidence of how much rides on getting the processes right.

The four routes online

1. A marketplace

Fast to start, no technology of your own. In exchange the customer relationship isn't yours, commission runs into double digits, and you sit directly beside your competitors. A sensible test channel for specialty goods; rarely viable for a full range.

2. A general-purpose shop platform

Cheap to start and perfectly adequate for many sectors. Groceries are not many sectors: variable weights, deposits, per-item VAT rates, best-before dates, allergens, delivery windows and cold chain are requirements a generic platform has to cover with plugins. It works for a while, then becomes a maintenance problem.

3. Custom development

Full control, and entirely your cost — the initial build, then maintenance, security updates and further development forever. It makes sense at a size where your processes are themselves the competitive advantage.

4. A specialist platform

Software built for grocery, which already knows about all of the above. You pay a recurring fee instead of an upfront investment, and you give up some freedom at the edges. In return you skip the part where most projects fail: rebuilding things that ought to be a given.

Which route is right depends mostly on range breadth and order volume. We've broken the money down separately in What does a grocery e-commerce platform cost?

Product data is the real work

This is the part almost everyone underestimates. In store you need an article number, a price and a category. To sell online you need, per item, at least: a usable photo, a name that makes sense without the shelf in front of you, ingredients and allergens, nutritional values, origin, pack size and unit price.

Across 3,000 items, done by hand, that's a project measured in months. Which is why two questions matter more than any design decision:

  • Does the data arrive automatically from your ERP or your wholesaler?
  • What happens when a supplier changes a recipe — do you update it manually?

If your platform connects to the systems you already run, most of this work disappears. If it doesn't, the work is permanent. More on that under Integrations.

Picking: where it tips

This is where the channel makes or loses money. A 40-line order picked by someone walking the shop with a printed sheet easily takes half an hour. The same order, route-optimised and worked through a handheld scanner, takes a fraction of that.

What matters:

  • Route optimisation. The pick list follows shelf order, not the order the customer added things to the basket.
  • Batch picking. Picking five orders in one pass walks the shop once instead of five times.
  • Variable weights. 300g of cheese is never exactly 300g. The mistake is to weigh during picking — weighing is the slowest single action in the whole flow. Weigh and label at the counter beforehand, with the weight encoded in the barcode; picking then means scanning and nothing else.
  • Substitutions. What happens when something is out of stock? Who decides, and does the customer find out before delivery?

For scale: on a platform built for this flow, around 60 seconds per order is achievable. The gap between half an hour and one minute isn't convenience — it's the gap between a loss-making channel and a profitable one.

And payment, with variable weights?

If the final total is only known after picking, no money can move at checkout. The standard approach is a pre-authorisation for the estimated total plus a buffer: the card limit is reserved, and the actual amount — always lower — is captured after picking. The alternative, charging the estimate and refunding the difference, works but produces a refund on nearly every order, with fees and bookkeeping attached.

There's a simpler route where it's available: with invoice payment or SEPA direct debit the amount isn't fixed until after picking anyway. The question never arises — one reason those methods suit grocery particularly well, and why they're the norm in German-speaking markets.

Delivery or collection?

Start with collection. Click & collect needs no vehicle, no route planning and no cold chain on the road — and it still answers the only question that matters: will your customers order online at all? What that looks like in practice is in Introducing click & collect in grocery retail.

Delivery is the more expensive and more complex step: time windows, route planning, cost per drop, refrigeration in the van, and what happens when nobody is home. Whether and when it pays is covered in Running your own grocery delivery: does it pay?

Where to start

  1. Narrow the range. Not 3,000 items online — the 300 that make up 80% of orders.
  2. Build clean data for those 300. Then you'll know how much work the rest will be.
  3. Start with collection. One time window a day is enough at the beginning.
  4. Measure picking. Minutes per order is your most important number, not shop visitors.
  5. Only then expand. Range, time windows and delivery, in that order.

Frequently asked questions

How long does it take to get a grocery webshop running?
Technology is rarely the bottleneck — with a specialist platform, a few weeks is realistic. The time goes into product data and into bedding in the picking process. Starting with a narrow range and collection only is considerably faster than a full launch with delivery.
Does selling food online require extra product information?
Yes. In distance selling, mandatory information such as ingredients, allergens, nutritional values and unit price must be available before the order is completed, not only on the packaging at delivery. It's one of the reasons product data maintenance is the most demanding part of the project.
Is an own shop worth it alongside marketplaces and delivery apps?
On a marketplace you pay commission and stay interchangeable, because you sit beside everyone else. Your own channel takes more work to build but gives you the customer relationship, the data and control over pricing. Many retailers run both: the marketplace for reach, their own shop for regulars.
At what size does this start to pay?
Order density matters more than shop size. Twenty orders a day concentrated in one catchment area can be picked and delivered efficiently; twenty orders spread across a region cannot. Before deciding on range, work out how tight your delivery area is.