Choose one platform, photograph your 30 to 80 best-selling products in a single session, record the facts for each in a spreadsheet, and have an AI assistant such as ChatGPT or Claude turn every row into a title, description and tags that you then check. That gets real listings live over one weekend; the full range can follow.
By Sunday evening you should have draft products imported, photographed, priced and checked, with payments and collection or delivery tested. The main caveat: AI writes only as well as the facts you give it. It cannot see your shelves, it does not know your stock, and it will happily invent a material or a page count if the spreadsheet leaves a gap.
Three decisions to make before Friday night
Weekends stall when these are left open, so settle them in advance.
Which platform
| Your situation | Sensible first choice | What it costs to start (list price, USD) |
|---|---|---|
| You already take card payments at the till with Square | Square Online, which shares your item library | A free plan exists; you pay card processing fees per sale |
| You want your own shop that can grow, with built-in AI writing and photo tools | Shopify | Basic plan about $39 a month, or $29 a month billed yearly; check the pricing page for your country and any trial offer |
| You sell handmade, vintage or craft items and want existing buyers | A marketplace such as Etsy, alongside or instead of your own shop | Listing and transaction fees rather than a monthly plan |
The rest of this tutorial uses Shopify for the step-by-step detail, because its CSV import and free Shopify Magic tools suit a weekend push. The spreadsheet and prompt work the same way anywhere.
Which products go first
Pick items you can reorder, that have steady stock, and that people already ask for. Skip one-offs, anything fragile you have not worked out how to post, and products whose prices change weekly. Thirty good listings beat 300 half-finished ones.
Collection, delivery or both
Offering local collection only for the first fortnight removes the hardest part of the weekend: postage rates, packaging and courier accounts. You can add delivery once orders prove there is demand.
Friday evening, two hours: the spreadsheet the AI writes from
This is the step that decides whether the AI output is usable. One row per product, and these columns:
- SKU: your own short code, also used as the photo file name.
- Name as you say it in the shop, not a marketing title.
- Maker or brand, product type, material or format.
- Size, dimensions and weight: the facts shoppers cannot check by picking the item up.
- Colour and variants, with one variant per line if the price differs.
- Price and current stock count.
- Three questions customers ask about it at the counter. These become the most useful lines of the description.
- Who buys it and why: "bought as a leaving gift", "regulars who collect the series".
The fastest way to fill it is to walk the shop with your phone recording voice notes ("SKU 014, linen notebook, A5, 192 pages, dot grid, sage or rust, 14 dollars, people ask whether it lies flat"), then paste the transcript into your AI assistant:
Below is a transcript of me walking round my shop describing products.
Turn it into a table with these columns: SKU | Name | Maker | Type |
Material/format | Size/dimensions | Weight | Colours/variants | Price |
Stock | Customer questions | Who buys it.
Use only what I said. If a column is missing for a product, write CHECK.
Do not guess dimensions, materials or prices.
[paste transcript]Expect about a third of cells to come back as CHECK. That list is your Saturday-morning measuring and weighing job, which is exactly why it is worth doing on Friday. A filled-in row looks like this (illustrative):
| Column | Row for SKU 014 |
|---|---|
| Name | Linen notebook |
| Maker / type / format | Small stationery maker / notebook / linen cover, dot-grid pages |
| Size and weight | A5, 192 pages, CHECK weight |
| Variants | Sage, rust (same price) |
| Price and stock | $14; 9 sage, 4 rust |
| Customer questions | Does it lie flat? Will fountain pen ink bleed through? CHECK both |
| Who buys it | Journal keepers; bought with a pen as a gift |
Notice that the customer questions are recorded as questions, with CHECK against the answers. You test the notebook with a fountain pen on Saturday; the AI must never answer them for you.
Before closing the laptop on Friday, ask the assistant to list every CHECK cell as a to-do list grouped by the tool you'll need. An illustrative extract: "Kitchen scales: SKU 014 weight, SKU 027 mug weight, SKU 033 candle weight. Tape measure: SKU 031 tote bag dimensions, SKU 040 print size. Test: SKU 014 lies flat, SKU 014 fountain pen show-through, SKU 027 dishwasher safe (check the maker's label). Ask supplier: SKU 033 burn time." Grouping by tool means Saturday's measuring takes one pass round the shop instead of three, and the "ask supplier" group can go out by email first thing so replies arrive before you write.
Saturday morning, three to four hours: one photo session
Set up once and do not move anything: a table by a large window, a sheet of white card curved up behind the product, the phone on a small stand, and the same angle for every item. Shoot three images per product: the front, a close detail (texture, stitching, a sample page), and one that shows scale, such as the item in a hand. Name each file with the SKU so matching photos to rows later is mechanical.
AI clean-up comes next. Shopify's media editor, part of Shopify Magic and free on paid plans, can replace a messy background with a solid one or a transparent one, and it can extend an image that is cropped too tightly. Dedicated tools such as Photoroom do the same in batches, which matters once you pass 50 or 60 products; its batch mode is on the paid plans. Our tutorial on creating product photos with AI on a small budget goes further into lighting and tool choice.
One rule for the whole weekend: AI may change the background, never the product. If the clean-up shifts a colour, softens a texture or removes a scuff a buyer should know about, go back to the original photo.
Saturday afternoon, three hours: turn rows into listings
Paste ten rows at a time into ChatGPT, Claude or Gemini with this prompt. Batches larger than ten tend to drift: descriptions get vaguer and details start migrating between products.
You are writing product listings for a small independent shop.
Voice: warm, plain, specific. British English. No superlatives,
no "perfect gift", no claims I have not given you.
For each row below, write:
1. Title (under 70 characters): product type first, then the
detail that tells it apart (maker, colour, size).
2. Description (80-150 words): first sentence says what it is and
who it suits; then answer the customer questions in the row;
end with size, material and care facts as a short list.
3. Tags: 5-8 lowercase words or phrases a shopper might search.
4. SEO title (under 60 characters) and SEO description
(under 155 characters).
5. Image alt text describing the main photo in one sentence.
Use ONLY facts in the row. If you need a fact that is missing,
write [CHECK: what is missing] instead of guessing.
Return the results as a table with one row per SKU.
[paste 10 rows]Here is an illustrative extract of what comes back for SKU 014, and what to fix before it goes anywhere near the shop:
Title: Linen A5 Notebook, Dot Grid, 192 Pages - Sage or Rust
Description: A linen-covered A5 notebook with 192 dot-grid pages,
for journal keepers and anyone who likes a notebook that lies
completely flat. The thick, bleed-proof paper takes fountain pen
ink without show-through...
Tags: linen notebook, dot grid journal, a5 notebook, bullet journal,
stationery gift
SEO title: Linen A5 Dot Grid Notebook | Sage or Rust (41 characters)Two problems, both classic. "Lies completely flat" and "bleed-proof paper" answer the customer questions in the row, which said CHECK; the AI filled the gap with the answer people want to hear. If your Saturday test showed slight show-through with a wet nib, the honest line is "light show-through with very wet fountain pen inks". The title and tags are fine. That is why the prompt asks for [CHECK] markers, and why you read every output, not a sample.
The same example shows the difference a spreadsheet row makes to a title. Before, the till called it "Notebook - linen"; after, the title leads with what a shopper types ("linen A5 notebook") and adds the detail that separates it from the next notebook on the page (dot grid, page count, colours).
Alt text needs the same attention, because it's what screen readers read out and what search engines use to understand the photo. A typical first draft says "Image of a notebook". A useful one says "Sage linen A5 notebook lying open to show dot-grid pages, pen resting on top". The second tells a shopper who can't see the photo what they'd see, and it matches the words people search. If the model's alt text starts with "Image of" or "Picture of", add a line to the prompt banning both.
If you are on Shopify and prefer to stay inside the admin, the "Generate text" option in the product description field does a similar job one product at a time, free on every plan. It is slower for 60 products, but handy for the odd item added later.
A special warning if you sell books, or anything with a famous name
Do not let AI write synopses or specifications from memory. Ask it about a novel and it will produce a confident blurb that may contain a character or ending that is not in the book. Use the publisher's description or the manufacturer's specification, and let AI write only your own line: why you stock it, who in the shop loves it, whether it is signed.
Sunday morning: import as drafts, then check every listing
Shopify's product CSV template accepts the columns you now have. The ones that matter for a first import are Title, URL handle, Description, Vendor, Type, Tags, Status, SKU, Price, Inventory quantity, Image alt text, SEO title and SEO description. Set Status to draft for every row so nothing goes live until it has been checked. Images can be imported from a Product image URL column only if the photos already sit at public web addresses, so for most shops it is simpler to import the text, then drag each product's photos in by hand. With SKU file names that takes a minute or two per product.
Then check each draft against this list. Budget two to three minutes a product.
| Check | How | What usually turns up |
|---|---|---|
| Every [CHECK] resolved | Search the admin for "CHECK" | Two or three per batch that slipped through |
| Facts match the item | Hold the product while reading the description | A material or size copied from a neighbouring row |
| Price and stock right | Compare with the till system | Old prices from a price list |
| Photos match the listing | Look at the first image in the product list view | Swapped photos between similar SKUs |
| Descriptions do not all sound alike | Read five in a row aloud | The same opening phrase on every product |
| Mobile view readable | Preview on your own phone | Long first paragraphs that push the price off-screen |
A realistic mistake from this stage: the URL handle column was left empty for two notebooks that share a name, so the import created them as one product with the second overwriting the first. Give every row a unique handle (the SKU works) before importing, and count the products in the admin afterwards against the rows in your sheet.
Sunday afternoon: the jobs that turn listings into a shop
- Payments and one test order, refunded afterwards, so you see exactly what a customer receives.
- Collection or delivery settings, with the collection times written as you would say them.
- Shop policies: AI can draft returns, delivery and privacy pages from your current practice, but consumer rules differ by country, so check them against official guidance or with an adviser.
- Stock link to the till, if your platform supports it, or a daily five-minute stock check if not.
- Tell existing customers: a sign by the till, a post, an email to anyone who has agreed to hear from you.
An independent bookshop's weekend in numbers
The figures here are invented but realistic. A two-person independent bookshop puts 64 products online: 40 stationery and gift lines plus 24 signed or staff-pick books, with local collection only.
| Stage | Time | Notes |
|---|---|---|
| Friday spreadsheet from voice notes | 2 hours | 21 CHECK cells to measure |
| Saturday photos and clean-up | 4 hours | 192 images, backgrounds replaced in batches |
| Saturday AI drafting, seven batches | 2.5 hours | Book blurbs taken from publishers, not AI |
| Sunday import and checks | 3.5 hours | About 3 minutes per product |
| Sunday payments, policies, test order | 2 hours | Policies drafted by AI, then edited |
That is 14 hours across two people. Writing 64 descriptions by hand at 15 minutes each would have taken 16 hours for the words alone. First-month cost: Shopify Basic at $39 (or $29 billed yearly), plus ChatGPT Plus at $20 if the free tier's limits get in the way. That is roughly $50 to $60, before card fees. Our breakdown of what AI costs a small retail shop in its first year covers the running costs after launch.
Where weekend listings usually go wrong
- Invented details. The prompt's "use only facts in the row" rule reduces this but does not remove it. The Sunday check exists for this reason.
- Sixty descriptions with one voice tic. AI repeats openings ("Crafted with care…"). Ban the phrase in the prompt and redo the batch.
- Overselling. An item sells in the shop on Saturday and online on Sunday. Link stock to the till or list fewer units online than you hold.
- Photos that flatter. Colour shifts after clean-up are the most common cause of "not as described" returns.
- Trust-damaging shortcuts such as fake scarcity or generic reviews. The tutorial on online shop AI mistakes that hurt trust lists the ones to avoid.
How to tell after a fortnight whether the listings work
Look at three things per product in your platform's analytics: views, add-to-cart rate and orders. A product with views but no add-to-carts usually has a price, photo or missing-fact problem. Every question a customer emails about a listing is a fact the description should have contained; add it to the spreadsheet row and regenerate. When you are ready to improve the copy itself, move on to writing Shopify product descriptions with AI that convert.
The same weekend for a florist's made-to-order range
A florist's range is mostly made to order, so the spreadsheet has fewer rows and different columns: a dozen bouquet styles at three sizes each, with columns for stem count, typical flowers, colour palette, vase included or not, and delivery cut-off time. The photos show one real example of each bouquet with a line that the AI must include on every listing: "Flowers vary with the season; we match the colours and style shown." The AI is useful for writing twelve distinct descriptions from similar ingredients, which is exactly where hand-written listings start to sound the same. The Sunday check focuses on delivery cut-offs and prices by size, the two facts that cause the most complaints.
If going fully online still feels like too much, AI can help a shop without a web store at all, through listings on marketplaces, email and social posts; see using AI without an online store.
Questions shop owners ask before their first online weekend
Can I do the whole weekend from my phone?
Photos, yes. The writing stage is much easier on a laptop, because you are pasting batches of spreadsheet rows into an AI assistant and copying the results back. Shopify's own description generator in the product editor is also desktop only, according to Shopify's help pages. A phone works for checking listings on Sunday, which is useful, because most shoppers will see them on a phone.
Should I use AI-generated lifestyle backgrounds behind my products?
For a first launch, keep the main image a real photo of the real item on a plain background. Generated scenes can come later for one secondary image, as long as the product itself is not altered. Shopify notes that images made with its media generation carry an invisible watermark. If a buyer could be misled about size, colour or condition, do not use it.
Do I need new stock software before going online?
Not for the first weekend if you list only items with steady stock and check counts daily. If you already use Shopify POS or Square at the till, their online stores can share one stock count with the shop, which removes most overselling. If you sell on several channels, sort out stock syncing before you add more products.
Further reads
- Shopify Magic vs ChatGPT for Product Descriptions — Decide whether Shopify's free writer is enough before paying for ChatGPT.
- How to Use Shopify Sidekick to Run Your Store Faster — Use Shopify's assistant for the admin jobs after launch.
- Your First 30 Days of AI Customer Support for an Online Shop — What to set up when customer questions start arriving.
- How to Get Your Products Recommended in ChatGPT Shopping — How new listings can surface when shoppers ask ChatGPT.
- How Much Does It Cost to Add AI to a Shopify Store? — A fuller costing once the shop is running.
- Can a Small Online Shop Use AI to Compete With Bigger Brands? — Where a small online shop can realistically out-serve bigger ones.
- Best AI Inventory Tools for Small Retailers Compared — AI inventory tools for small shops compared: Prediko, Inventory Planner, Square Plus and Shopify Sidekick, plus what to do for weighed and perishable stock.
- How to Write eBay Listings With AI That Sell Faster — Sold-price research, a title formula, item specifics and flaw-honest descriptions for eBay listings drafted with AI, plus the policy traps to avoid.
- How to Write Product Descriptions With AI That Actually Sell — The fact sheet, buyer questions and prompt structure that turn AI product descriptions from adjectives into answers, with before-and-after examples.
- AI Alt Text: How to Generate and Check Image Descriptions — Generate alt text with AI for product photos, banners and charts, then check every description with a five-point test before it goes live.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Shopify pricing page and Shopify help pages (product CSV import, media generation, Shopify Magic text generation); Square Online plans page; OpenAI and Anthropic plan pages as summarised in the site fact sheet.