Store Check-Ins, Listings and Reviews: What AI Phone Control Can Handle

After closing time, the phone work is still waiting

Someone I know runs a small neighbourhood restaurant — two locations, four people including himself.

He says the worst part of his day is not the lunch rush. It is the hour or so after ten at night. The day’s group-buy orders need reconciling, tomorrow’s new set menu has to be configured, and yesterday’s reviews are still sitting unanswered.

In that hour, maybe ten minutes involve actual thinking. The rest goes into the same loop: open, find, fill in, send, open the next page.

He did the maths once. Three platforms, two locations, and just retyping the same set-menu details three times costs him four or five hours a week.

This article takes those tasks apart: what can be handed over, what cannot, and in what order.

A store’s phone work comes in three kinds

• Copy-and-fill. Check-in records, set-menu listings, store profile updates, opening hours. No judgement needed — the information already exists, it just has to be moved again.
• Copy-and-reply. Answers to common questions, thank-yous for good reviews, booking confirmations.
• Must be you. Bad reviews, complaints, conversations with platform staff, anything involving a promise.

Once sorted, a pattern shows up: the first kind is the largest, the third is the smallest but the one that hurts, and the second sits in between.

Almost all the time goes into the first kind — not because it is hard, but because there is so much of it.

Check-ins: only half of it can be handed over

The check-in itself cannot move. The customer is standing at the counter, and scanning, confirming and saying “you’re all set” has to happen right there.

What can be handed over is the record-keeping afterwards.

Most owners rely on memory or a quick glance at the dashboard, then discover a few mismatched entries at month-end. By then the platform records are painful to search.

This is the part worth handing over: read the check-in records at a fixed time each day and append them to a sheet. No reading, no changing, no acting — just moving numbers.

Described in iEasyRun it looks like this:

Open the order pages for these three platforms, read out every group-buy
order checked in today — order number, set menu, check-in time — and append
them to the end of this sheet.

There is no “how to click” in that description, only what to read and where it goes.

Why is it worth handing over? Because the risk is zero. It touches nothing, and the worst case is a misread number you would spot instantly.

Set-menu listings: the upload is not the problem

The first time you list something, judgement is involved: which set menu, what price, which hours are available. That is a business decision and it stays with you.

The trouble starts afterwards — the same settled information has to be entered on three platforms.

Worse, the field names differ. One calls it “set name”, another “item title”. One asks for “terms of use”, another splits it into “validity” and “blackout dates”. You know it is one thing; your hands type it in three formats.

That second half can be handed over:

Open the listing pages for these platforms and fill in the set-menu details
from this configuration sheet. Use the images in this folder. Stop when you
are done and wait for me to confirm before submitting.

Note “stop and wait for me”. Listing is customer-facing, and one confirmation step moves the cost of a mistake from “wrong set menu published” to “one extra click”.

Review replies: templates for praise, never for complaints

Split this in two.

Good reviews can use templates, but not one identical template forever. Platforms recognise boilerplate, and dozens of identical “thanks for your support, hope to see you again” replies on one store do it no favours. Keep three to five variants, or mention the specific dish.

Neutral and negative reviews must be written by you. Not because a machine writes badly, but because these customers want something else — to be taken seriously rather than politely dismissed.

A more practical suggestion: the part worth handing over is not the reply, it is the alert.

Each morning, check the review pages for these stores and screenshot any
new reviews for me.

That way you never miss one — missing a bad review costs far more than answering it late — and you never answer a conversation with a template that deserved attention.

Store content: distribution yes, what to shoot no

For a local business, posting regularly brings people in. What to shoot and how to shoot it is your call, and it does not get handed over.

But “send the same clip to several platforms” is worth handing over.

One ten-second clip goes to Douyin, then Xiaohongshu, then WeChat Channels. None of those three steps needs thought, but together they take twenty-odd minutes, and each one means picking the file again and rewriting the title.

Post the videos in this folder to these three platforms, using the matching
titles from the sheet.

For the detail, see how bulk posting on Xiaohongshu works — the thinking is the same.

What never gets handed over

• Calming down an angry reviewer. No exceptions.
• Conversations with platform account managers or support. These have back-and-forth and probing. A machine cannot hold them.
• Anything involving a promise: deposit refunds, private bookings, invoices, holding a table. Say the wrong thing and it is a dispute.
• Anything improvised: closing for half a day at short notice, a sudden price change. There is no pattern to learn from, and a machine makes these worse.

A useful test: if sending the wrong sentence causes real loss, keep it. “Hope to see you again” sent wrongly is awkward. “The deposit is refundable” sent wrongly costs money.

What to hand over first

Keep to the order:

• First, anything read-only. Daily check-in reconciliation, review alerts. Zero risk, and you feel the time saved immediately.
• Second, anything copy-and-fill. Listings, profile updates, content distribution. Keep a confirmation step.
• Third, outbound messages. And only the most standard ones.

To judge any task, ask three questions:

• Is the answer fixed? If it has to be thought out every time, handing it over only moves the thinking elsewhere.
• Does a mistake cause real loss? If so it stays with you. This one is not negotiable.
• Does it need live judgement? That stays too.

Pass all three and it can go. Fail one and leave it.

Store work has a particular quality: it is not hard, it is fragmented. Fragmented work suits a machine, because a machine does not mind fragmentation and a person does. That is what this kind of phone automation software is for — it is completely free and runs on your own computer. Start with one location for a week.

For device-side problems, read what to do when a phone keeps going offline; for getting daily data into a sheet automatically, see turning phone screenshots into a spreadsheet.