Your store platform does the selling. The work around it lands on the same few people: checking how yesterday went, spotting what's about to sell out, typing up the wholesale order that arrived by email, answering "where's my order?", and making things right with the customer who wasn't happy. None of it is hard, and all of it comes back every day.
Here are the workflows we'd set up first for a small online store, each a ready-made template in Lodol's library, plus how to handle returns with a person deciding every refund. They share one rule: when something is unclear, a person hears about it rather than the workflow guessing.
1. Start every day knowing how yesterday went
The chore: clicking through admin screens each morning to piece together sales and stock.
The workflow: at 7:00 every morning, Start every day with your Shopify sales and stock emails you one brief. It leads with yesterday's orders, sales and average order, each compared with the same weekday a week before, then lists the five products that sold the most units and every tracked item at or below your low-stock level, sold-out items first, so you hear about a sell-out before a customer does.
What it won't do: mistake a problem for a quiet day. Yesterday runs midnight to midnight in your store's own time zone, and test, canceled, voided and fully refunded orders are left out. A day with no orders still gets its email, saying so. If Shopify can't be read, nothing is sent and the run is marked as failed.
2. Hear about low stock and expiry dates in time
The chore: keeping an eye on stock that lives in a spreadsheet, such as goods with a use-by date or supplies you reorder by hand.
The workflow: at 9:00 every morning, Daily inventory monitor for low stock and expiry checks each item on your inventory tab two ways: its quantity against its reorder level, and its expiry date against today. You get one email, only on mornings when something needs attention, with a reorder list, grouped by supplier when your sheet names one and with how many to order when it says, then what has expired or expires soon. Each alert is noted in the row's Alert Status cell, so the next email separates new from outstanding.
What it won't do: skip a row it can't read. A row whose quantity or date makes no sense is listed under "check these rows". Only the Alert Status cell is ever written, a status you typed by hand is left alone, and an empty tab fails the run rather than passing for a morning with nothing to reorder.
3. Turn order emails into rows
The chore: wholesale and trade customers who order by email, and someone retyping each line.
The workflow: every hour, Turn order emails into spreadsheet rows reads the order emails matching your Gmail search, has AI pull out who is ordering and every item and quantity, and adds one row per item to your order sheet, with the customer, the order reference, the requested delivery date and the email it came from.
What it won't do: guess at an order. If an email isn't clearly an order, or any item's name or quantity is missing or unclear, nothing from it is written and the email is labeled "Order needs review" for a person. Handled emails are labeled "Order logged", so no order is recorded twice.
4. Answer customer emails faster, with a person on every reply
The chore: the same questions about delivery and returns, answered one at a time.
The workflow: Draft a reply to every new email in your voice writes a reply to each new email in Gmail and saves it as a draft under the message it answers, for you to read, change and send. Draft replies to customer emails in Outlook does the same in Outlook. If you'd rather approve each reply in one place, Review and send replies to support enquiries shows you each reply next to what the customer wrote, and sends it only when you approve.
What it won't do: make promises. Where a reply needs a fact the AI doesn't have, such as a delivery date or a price, it leaves a gap in square brackets for you to fill, and the Outlook version leaves refunds, discounts and dates for you to decide. Automatic replies and bulk mail get no draft at all. A good first change: add a step that looks the parcel up in AfterShip or Shippo before the reply is written, so "where's my order?" gets its real status instead of a blank.
5. Thank the happy customers, win back the unhappy ones
The workflow: once a day, Thank happy customers and win back unhappy ones with a coupon reads your sheet of purchases and feedback, has AI judge whether each customer's feedback is positive, neutral or negative, and emails each one from your Gmail in the tone you choose. Negative feedback gets an apology and your coupon code; everyone else gets a thank-you that answers what they said.
What it won't do: email anyone twice about the same purchase, or offer a coupon on the thank-you path. Every email is logged on a tab that's checked before each send, and each run sends a capped number. It sends on its own once turned on, so to read each email first, add a Request approval step before the send.
6. Returns, with a person deciding every refund
There's no returns template yet, because every store's policy is different, but it's a short workflow to build, and the decision stays with a person:
- Collect the request. A new email to your returns address starts the run, or a new submission to a returns form in Fillout.
- Look up the order. A Shopify step fetches the order: what was bought, when and for how much.
- Ask before any money moves. A Request approval step asks the person who owns returns, for example, "Refund $64.00 to Dana Reyes for order #1043? Reason given: arrived damaged." They answer from the email or from Lodol, and can be reminded if they don't.
- Act on the answer. If they approve, refund the payment in Stripe if that's how you take payments, or buy a label with Shippo from the customer's address to yours, and email the customer. If they reject it, email the reason instead.
Every decision is recorded with who made it and when, so "who approved that refund?" has an answer in the run. Or describe it to Claude or ChatGPT and check the first version it builds.
Listings and prices
Draft SEO meta titles and descriptions for Wix products writes a search title and description for each product into a sheet, and is told never to invent price, stock, delivery or discount claims. Nothing is written back to your store; you paste in the ones you like. Turn a shopping search into a price sheet runs one shopping search a week for a product you name and logs each result's price, seller and rating, building a price history of what others charge.
What it costs
Lodol charges per run, however many steps a run takes, and nothing per order, product or seat. For a store whose inbox gets about 20 customer emails a day, the core five use:
| Workflow | When it runs | Runs a month |
|---|---|---|
| Morning store brief | Every day | 30 |
| Inventory monitor | Every day | 30 |
| Order emails to rows | Every hour | 720 |
| Reply drafts in Gmail | Once per new email | About 600 |
| Thank-you and win-back emails | Every day | 30 |
| All five | About 1,410 |
That fits Builder's smallest tier, 2,500 runs for $18 a month. The reply drafts run once for every email that lands, including ones they skip, so point them at the address customers write to. The AI steps run on your own Anthropic or OpenAI account, which bills you at its own rates; Lodol adds nothing.
Start with tomorrow morning.
Add Start every day with your Shopify sales and stock to your workspace and read your first brief over coffee. Or browse all of the e-commerce workflows, and see how Lodol fits an operations stack on the logistics page.