Most store owners start the day the same way: open the admin, look at the sales number, scroll the orders, wonder vaguely about stock, and get pulled into email before finishing. Things get missed. An order sits unfulfilled for four days. A bestseller runs out on a Thursday.
A fixed routine fixes that, and AI makes it short. Six questions, the same ones each morning, asked of an assistant that can read your live store data. It takes about ten minutes, and all of it is read-only, so nothing can break.
Set up once
You need an assistant that can see your store. Either works:
- Sidekick, inside your Shopify admin. No setup.
- Claude or ChatGPT, connected through Shopify's official integration. Read-only access is all this routine needs. Setup guides: Claude and ChatGPT.
If you use Claude or ChatGPT, begin each prompt with "Using Shopify" so the assistant uses the integration. In Sidekick you can skip that.
The six questions
1. How was yesterday?
Show total sales, number of orders, and average order value for yesterday, and the same for the same day last week.
One day means little by itself. The comparison is what tells you whether yesterday was normal. Same weekday last week is a fairer yardstick than the day before, because Tuesdays and Saturdays behave differently.
Look for: a number that's well off. Up is worth understanding as much as down.
2. What's stuck?
Which of my recent orders are still unfulfilled, and when was each one placed?
Look for: anything older than your normal turnaround. One old unfulfilled order is usually a problem with that order: an address issue, an out-of-stock item, something that fell between two people.
3. Anything unusual?
From my most recent orders, flag any with an unusually large quantity of one item, or a much higher total than usual.
Look for: the order that needs a human look before it ships. A large quantity might be a wholesale buyer you'd want to email, or a mistake, or fraud.
A connected Claude or ChatGPT lists up to 50 recent orders at a time. On a busy store, ask about "orders since yesterday morning" in smaller windows.
4. Am I about to run out of anything?
Show current inventory by variant for these products: [your top 10 sellers].
Keep your top sellers as a saved list and paste it in. This is the question most likely to save you money. Finding out on Monday that a size is down to four units gives you time to reorder. Finding out when it's sold out doesn't.
Look for: any variant below roughly a week of sales.
5. What moved?
Show my top 10 products by units sold yesterday.
Look for: surprises. A product that isn't normally on the list is telling you something: a mention somewhere, a search trend, a competitor out of stock. Worth five minutes to work out why.
6. Sum it up
Summarize all of that in five lines, and tell me the one thing I should deal with first.
The assistant has the whole conversation in view, so it can rank. You'll sometimes disagree with its pick, and that's fine. It's a first draft of your priorities, written in ten seconds.
Make it faster
Save the prompts. In Sidekick, save your refined prompts as skills and re-run them with a slash shortcut, as covered in Sidekick prompts that save time. In Claude or ChatGPT, keep them in a note or in saved instructions.
Fill in the specifics once. Your top-seller list, your normal turnaround, your threshold for a "large" order. A prompt with real numbers gives a usable answer the first time.
Ask one thing at a time. It's tempting to paste all six as a single prompt. Shopify's documentation says analytics works best with one short, specific question, and layered requests are where answers go wrong. Six quick questions beat one tangled one.
Turn the repeatable parts into alerts
A morning check catches things once a day. Some things shouldn't wait until morning.
If question 4 keeps being the important one, stop asking it. Have Sidekick draft a Shopify Flow that emails you the moment a top seller drops below your threshold. Same for question 2: a Flow can flag any order unfulfilled after three days.
The routine then gets shorter, and it's left with the questions that need judgment: was yesterday normal, and is anything odd. We walk through this in building Flow automations with AI.
Trust, then verify
For the first week, spot-check. Compare question 1 against your analytics dashboard, and question 4 against the product's inventory in the admin. If the numbers agree, you've confirmed your phrasing works. If they don't, the cause is nearly always a different metric, a different date range, or a different location. We cover each in asking AI about your Shopify sales and checking inventory with AI.
After that, verify only when a number surprises you.
A weekly add-on
Once a week, add three slower questions:
- "Show total sales by week for the last eight weeks."
- "Which products had the most sales this week compared with last week?"
- "Which discount codes were used this week, and how many orders used each?"
Daily numbers are noisy. The weekly view is where you see a real trend.
One check your store data can't answer
Everything above looks at what happened in your store. It can't show you the sale that never arrived: the shopper who asked an AI assistant for a product like yours and was pointed somewhere else. That doesn't appear in orders or analytics.
It's worth looking at now and then, the way you'd check your search rankings. The free AI readiness checker shows what AI assistants can read from your storefront, and takes about a minute.
The routine on one page
- Yesterday versus the same day last week
- Unfulfilled orders and their age
- Unusual orders
- Stock on your top sellers
- What sold
- Five-line summary and the first thing to do
For more prompts to adapt, see 25 prompts for a connected Claude or ChatGPT.
Comments
Every comment here comes from a verified email. Write yours, confirm from your inbox, and it's live.
Loading comments…