A Week in a Small Online Shop, With AI Doing the First Draft

Running a small online shop means doing five jobs before lunch: accountant, customer service, copywriter, stock planner and parcel packer. Most owners have already tried ChatGPT or Claude for a product description. Far fewer have changed how the week actually runs.

That's normal. In a Goldman Sachs survey of small business owners, only 14% said they had built AI into how the business actually runs. And owners are cautious for good reason: in Bluevine's survey, data safety was their top concern about AI, at 33%, up from 23% a year earlier.

So here's one ordinary week in a small shop, with the places AI can take the first draft and the places the owner stays in charge. It's an illustration of the kind of work we mean, not a client story.

Monday: the numbers

Before: five tabs open. Shop dashboard, two ad accounts, maybe a marketplace, then a spreadsheet to copy it all into.

With AI: you ask the assistant to read those pages and put last week's sales, ad spend and orders into one line of your tracker. We do exactly this for our own projects.

Your part: for the first few weeks, check its totals against the dashboards. Once they match week after week, a quick glance is enough.

Every day: the inbox

Before: the same questions, typed fresh each time. "Where's my order?" "Does it come in blue?" "How do I send it back?"

With AI: each email is sorted by type and a reply is drafted in your tone, using your own delivery and returns wording.

Your part: read and send. Complaints, refunds and anything unusual get your own reply, written by you.

Wednesday: a new product goes up

Before: the supplier's sheet, rewritten line by line into a title, a description, sizes and care notes.

With AI: give it the supplier details, your own notes and two descriptions you like, and it drafts the listing in the same voice.

Your part: check every fact. Materials, sizes, what's in the box. A description that promises too much comes back as a return.

Friday: what to reorder

Before: scrolling the stock list, guessing.

With AI: from your stock export, a short draft list: what's running low, what sold faster than usual, what hasn't moved in two months, with a reason on each line.

Your part: the decision and the money. The list makes sure nothing runs out by surprise; what you order is up to you.

Once a month: what customers are telling you

Before: reviews and return reasons pile up unread.

With AI: they're grouped into themes, with the three most common at the top: "runs small", "colour looks different from the photo", "arrived later than promised".

Your part: fix the one that's costing you most. Often it's a photo or a size chart, the cheapest improvement a shop can make.

Lines the owner doesn't hand over

• Prices and discounts. The AI can show you the numbers. What you charge is your call.
• Unchecked messages to customers. Drafts, yes; automatic sending, not yet. If you ever add a customer-facing assistant, it must say it's an AI.
• Card and payment details. Never pasted into any AI tool.

Try it for two weeks

You don't need new software to test this. A business plan of Claude or ChatGPT in the shop's own account is enough to start.

• Week 1: pick one day from the week above, usually Monday numbers or the inbox. Do it with AI, but keep doing it the old way too.
• Week 2: compare. Did the drafts need heavy fixing? Did the totals match? How much time did it actually take?
• Then decide: keep it, adjust it, or drop it, before adding a second day.

If you want a hand

Our AI Check-up is a 60-minute call about your shop, ending with a one-page plan of which parts of your week to hand over first. If you book a build within 30 days, the fee comes off the price, and everything we set up stays in your own accounts.