I have seen an operator in home goods who runs five stores.
Last month I talked with her. She said two completely different things happen in a day. One is deciding what to push today, how to adjust the price, and how to schedule the promotion. The other is getting those decisions into five back ends.
The first is her work. The second, in her own words: she could do it with her eyes closed, but she cannot do five stores with her eyes closed.
What follows takes apart the second thing. Which stretches of a day are repeated actions that need no judgement, and which of those to hand over first.
Listing: deciding and entering are two parts
Listing naturally splits in two.
The first half is deciding: which product, what price, how to write the selling points, which main image. It is different every time.
The second half is entering: open the back end, create a product, put the name, price, and specs into the matching fields, upload the images, pick a category, and submit. The steps, the fields, and the order are fixed. Only what you type in changes.
That second half can be handed over. With iEasyRun you can say it like this:
Open the product back end and create a product. Read the name, price, specs, and description from this sheet. Take the images from this folder on the desktop and upload them in order. Stop before submitting and let me look.
Keep that last sentence. Listing publishes outward. The machine stops before submit, and a person does the last check. You save the time spent entering. The act of publishing stays with you.
Repricing and copying the numbers
These two often get filed under daily maintenance. They should not be handled in the same order.
A pricing mistake is real money. Put a zero in the wrong place and one night can wipe out a month of profit. So it should not be first. After patrol and fixed-field entry are both running steadily, then touch repricing.
Sorting the numbers only reads. Open a few back ends and copy sales, stock, and review counts into one sheet. A copy error means the numbers are off. It does not cause a loss directly.
Do the read-only work first. While it runs you will hit which page loads slowly, which back end is in maintenance in the morning, and which device tends to stick at this step. Those are useful when you get to repricing.
Review replies and private messages
Review replies and private messages are two ends of the same job. A review faces public content, and you care about the wording and how far it reaches. A private message faces one person, and you care about how fast you reply.
Standard questions can be handed over. Anything with emotion, or that makes a commitment, has to be a person.
“When will it ship”, “do you accept returns”, and “is there a discount” have fixed answers. You do not have to judge them on the spot. That kind is most of a day’s messages.
Complaints, bad reviews, and promises that involve compensation: the customer wants to be taken seriously. Leave that part to a person.
The numbers before the end of the day
Sorting the data and sending the daily report looks like a little over ten minutes. The hidden cost is switching from doing the work into sorting, then switching back. One switch costs more than those minutes.
The report itself cannot be skipped. What can be handed over is the sorting: have the machine pull the figures at a fixed time and put them in a sheet, so what you open is already sorted. You only judge which store dropped hard today and whether to adjust.
Most tools you already have only handle how to operate. Turning the data after the operation into a sheet you can read is often where the relief is.
Which part to hand over first
In this order:
- Sorting the numbers. It only reads, it happens most often, and you can check the result. After two weeks you will also get a feel for how the devices and the network behave.
- Fixed-field entry. The back half of listing, batch updates to product details, and similar work. The steps are fixed, and a mistake can be rolled back.
- Replies to standard questions. High volume, fixed answers, a clear boundary.
- Actions that touch money, such as repricing and taking a listing down, last, and with a person confirming before submit.
The basis for that order is one line: start where a mistake has no loss.
For any action, ask two questions. Does this step need a judgement? If it does, leave it to a person. If it does not, it can be handed over. If it goes wrong, what is the loss? If the loss is doing it again, it can be handed over. If the loss is paying money or affecting the account, do not hand it over yet.
Once those two are answered, whether to hand the action over, and where it sits in the queue, is mostly clear.
A related case is how many messages a support person answers in a day. The multi-account listing flow is in bulk e-commerce listing.
Frequently asked questions
- Which part takes the most time?
- It depends on volume. Listing and repricing take longest per action, but data collection and review replies happen most often. Taken together, what actually eats the day is whatever is high-frequency, short, and needs no judgement.
- Can listing be fully automated?
- The back half can. Filling fixed fields, uploading images and applying a template are all mechanical. Choosing products, pricing and writing selling points are judgement calls. The sensible split is the machine enters and uploads, a person decides what goes in.
- Can repricing, which touches money, be handed over?
- Put it last. Repricing, delisting and refunds cost real money when they go wrong. The right order for any automation is patrol first, then fixed-field entry, and modifications last.
- Why hand data collection to a machine?
- Because a person will not get it wrong, but will get tired of it. Manual transcription starts drifting around the sixth page. And this is the most regular category of work there is, which makes it the best automation candidate.
- How is answering reviews different from support?
- Two ends of the same job. Reviews are public and about wording and impact; private messages are about response speed. Both share the same line: standard questions can go, anything with emotion or a commitment cannot.
- Can the daily reporting be skipped?
- The report itself cannot, but the assembling can. Have a machine collect the figures at a fixed time into a sheet and you only make the judgement. What that saves is not the few minutes but the switch into and out of a different mode.
- Which part pays back first?
- Data collection. It carries the least risk because it only reads, it happens most often, and the result is verifiable. Two weeks of it also teaches you how each back end loads, which pays off later.
- How many stores can one person run?
- It depends how much has been handed over. Purely manual, three to five stores is the limit. With patrol and fixed-field entry automated, a dozen is common.
How many times a day?
Hand over the half that needs no thought
Software is completely free and runs on your own computer. Start with the most repetitive part.