1. Home
  2. Blog
  3. 5 shop tasks you can automate this week (with the step-by-step recipe)

Automation · 8 min read

5 shop tasks you can automate this week (with the step-by-step recipe)

You don’t need a big AI project to get started. These five tasks come round every day in any shop. For each one you get the problem, the step-by-step recipe, the tool, the usual mistakes and what you can do today.

In 30 seconds

  • Five repetitive shop tasks you can automate without coding using Make, Zapier or n8n.
  • Each recipe has a trigger, steps and an outcome, with the tool and the data you need.
  • Start with drafts and human review: the automation suggests, the person decides.
  • Don’t do all five at once: the one-week plan leaves you with one running and another ready.

In a shop, the manager’s time is one of the most expensive things there is. Every hour spent copying data from one system to another is an hour off the shop floor, away from the team and the customer. Repetitive work can be automated with tools such as Make, Zapier or n8n, which connect apps without any need to code.

Three words you’ll see in every recipe:

Starts

Trigger

What sets the automation going: a fixed time (every day at 8:00) or an event (a form comes in, an email arrives).

Runs

Steps

What it does next, in order: read data, calculate, write to a sheet, send an alert.

Needs

Data needed

The information that has to exist and be accessible. If the data isn’t there, no automation is possible.

Example The time savings are estimates for a hypothetical shop, not measured results. Use them as an order of magnitude and measure your own.

The daily sales report

The problem: imagine a clothes shop with 3 branches. Every morning someone exports the sales from the POS, pastes them into a sheet, compares them with last week and sends an email. If that person is on holiday, the report doesn’t arrive.

TriggerEvery day at 7:30
Step 2Read the previous day’s sales
Step 3One row per shop in Sheets or Excel
Step 4Compare with last week and last year
Outcome3-line summary per shop
Lowdifficulty
~1.5 hestimated weekly saving (about 15 min a day)
Zapier or Makeif your POS has a connector; n8n if it only exports files
How to read the sales: if your POS or online shop has a connector or API (Shopify, WooCommerce, Square and others do), they’re read directly. If not, have the POS send the report by email and the automation picks it up from the inbox. In the sheet: date, shop, sales, transactions and average ticket. The summary goes by email, Telegram or Teams, not as an attachment.

Data needed: sales per shop and day, a history in the sheet to compare against and the list of people who receive the summary.

Common mistakes: sending the whole sheet instead of a summary (nobody opens it on their phone); not allowing for closures and bank holidays, so the comparison comes out at −100 %; mixing sales with and without VAT depending on the source.

Your first step today

  • Write down on a piece of paper the three figures you look at every morning. Just those three. That’s your report.

Cashing up the till

The problem: the close ends up in a notebook or in a sheet with amounts typed in by hand, and discrepancies are found days later, when nobody remembers what happened.

TriggerClosing form
Step 2Look up that shop’s POS sales
Step 3Calculate counted vs. recorded
Step 4Over the threshold? Alert the manager
OutcomeRow in the control sheet
Lowdifficulty
10 minestimated saving per close, plus the time spent hunting down old discrepancies
Makefor its visual threshold filter; Zapier with a filter step
The form (Google Forms, Microsoft Forms or Tally) asks for six fields: shop, date, cash counted, card total from the card terminal, till float and notes. You set the threshold, for example 5 euros. Below it, the row is simply saved.

Data needed: POS sales per shop and day, each shop’s fixed till float and the tolerance threshold.

Common mistakes: alerting about everything (if every cent triggers an alert, by the third day nobody reads them); not recording who did the close; using it to monitor people rather than to help them. Explain it to the team before you launch it.

Your first step today

  • Create the form with those six fields.
  • Ask for tonight’s close to be done with it. You can connect the automation tomorrow.

Orders and stock alerts

The problem: every week someone counts, notes down and writes the order from scratch. Or nobody notices a product has sold out until a customer asks for it.

TriggerMonday 9:00 or stock change
Step 2Read the stock for each item
Step 3Filter those below minimum
Step 4Group by supplier
OutcomeDraft order to review
Mediumdifficulty
1 to 2 hestimated weekly saving with a few suppliers
Make or n8ngrouping by supplier means looping through lists
First of all, prepare a master sheet with every item: code, supplier, minimum stock and usual order quantity. The stock is read from the management system or the online shop. The draft (one email or document per supplier) goes to the person in charge of purchasing, and is only sent to the supplier once they approve it. Zapier can do it, but the flow gets complicated sooner.

Data needed: up-to-date, reliable stock, the master sheet with minimums and each supplier’s email address.

XLSXTemplate: stock master sheetExcel with the columns ready and the formula that flags what to order.Download ↓
Common mistakes: automating on top of unreliable stock (it’ll order badly, just faster); sending the order without human review, when you should always start with a draft; using the same minimums all year round, without taking the season into account.

Your first step today

  • Pick the 20 items you sell most.
  • For each one, note the minimum stock and the quantity you usually order.
Real case · Make

An Amazon seller stops handling orders by hand

A German Amazon supplier handled around 30 batch orders a week with a spreadsheet. The consultancy Techflow.ai used Make to automate order confirmation, shipping labels, invoices, warehouse notifications and real-time stock updates, with a forecast based on sales and production lead times.

15 h → 0weekly hours of manual order handling
30 % → <1 %orders with stock discrepancies
€900,000 → €360,000tied-up stock

The takeaway: the big saving didn’t come from the hours, but from having reliable stock. With good data it was possible to order less and with fewer mistakes.

Source: Make, Techflow.ai case ↗. Figures published by Make.

Team rotas

The problem: holidays, shift swaps and sick leave arrive as scattered WhatsApp messages. Someone has to sort them out and tell each person. Messages get lost and gaps appear that nobody covers.

TriggerRequest form
Step 2Save in the sheet as “pending”
Step 3The manager approves or rejects
Step 4Update the shared calendar
OutcomeNotice with the updated rota
Mediumdifficulty
~1 hestimated weekly saving, plus the mistakes avoided
Zapier or Makewith Microsoft 365, Power Automate too
Details: the form asks for name, type (holiday, shift swap, absence), dates and, if it’s a swap, who with. Approval can be an email with links or a card in Teams or Slack. The calendar can be Google Calendar or Outlook. If it’s a swap, the colleague is notified too. Optional: every Friday, each person receives their rota for the following week.

Data needed: team list with email or mobile, base rota and a shared calendar.

Common mistakes: launching the form without saying that WhatsApp changes no longer count (if both channels coexist, it fails); not allowing for minimum cover, when it’s enough to check how many people are left that day; storing health data about sick leave. “Absence” is enough.
Real case · Data protection

A €42,000 fine over a work WhatsApp group

The Spanish Data Protection Agency (AEPD) fined LVMH Iberia for adding an employee’s personal mobile number to a work WhatsApp group without her consent, and for adding her again after she had objected in writing. The proposed fine was €70,000, reduced to €42,000 for voluntary payment and admission of the infringement.

The takeaway: if rota notices go to someone’s personal mobile, ask for permission in writing and offer another channel, such as work email or a company app.

Source: ECIJA, October 2025 ↗

Your first step today

  • Create the request form.
  • Share it with the team with one clear sentence: “from Monday, changes are requested here”.

Google reviews with AI

The problem: Google reviews are the shop’s digital shop window. Many go unanswered because nobody sees them in time, and unanswered negative ones carry weight.

TriggerNew review on Google
Step 2Extract rating, text and shop
Step 3Draft with ChatGPT or Claude
Step 4Save in a sheet
OutcomeThe manager reviews and publishes
Mediumdifficulty
5 minestimated saving per review; with 10 a week, almost an hour
Make or n8nthey have OpenAI and Anthropic modules; so does Zapier
Fixed instructions for the AI: brand tone, maximum length, sign off with the shop’s name, promise nothing specific and, if the review is negative, invite them to get in touch privately. 1- or 2-star reviews are flagged as urgent. If your tool has no direct connector, use the notification email Google sends as the trigger.

Data needed: access to the Business Profile, an AI API account and a tone guide of 5 or 6 sentences.

Common mistakes: publishing automatically without review (an unfortunate reply to a complaint is highly visible); identical replies for everyone, when you can ask the AI to mention something specific from the comment; including customer data or details of their purchase.

Your first step today

  • Write your tone guide: how you greet, how you sign off, what you never say and how you reply to a complaint.
DOCXTemplate: tone guide for reviewsWord document with gaps to fill in, examples and the instruction for the AI.Download ↓
Real case · AI reviews

Everytable, 38 outlets: from one person answering everything to AI drafts

This takeaway food chain receives hundreds of reviews a month. Its customer service lead answered all of them to keep up the quality, which was impossible to sustain. They started using AI-generated drafts tuned to the brand’s tone.

70 %less time writing each reply
38 % → 60 %reviews answered
50replies in the first two weeks

The takeaway: AI drafts let you share out the task without losing the tone. That’s exactly what the recipe above suggests.

Source: Marqii, Everytable case ↗. Figures published by the tool’s provider.

All five, compared

TaskDifficultyRecommended toolEstimated weekly time
Daily sales reportLowZapier or Make1 a 1,5 h
Cashing upLowMake1 h
Orders and stock alertsMediumMake or n8n1 a 2 h
Team rotasMediumZapier, Make or Power Automate1 h
AI reviewsMediumMake or n8n0,5 a 1 h

The times are estimates for a hypothetical small or medium-sized shop. In your case they may be higher or lower.

One-week plan

Don’t try to do all five at once. This plan leaves you with one automation running and another ready.

Monday

Measure and choose

Note down how much time you spend on each task. Pick the one that takes up the most time and is low difficulty.

Tuesday

Gather the data

Logins, sheets and lists. Create the spreadsheet or the form.

Wednesday

Build

Open a free account on the tool you’ve chosen and set up the automation with test data.

Thursday

Test it for real

With real data, but sending alerts only to you. Fix whatever goes wrong.

Friday

Switch it on

Open it up to the team and explain in two minutes what changes and what doesn’t.

Saturday

Check

Check in the tool’s run history that it has worked every day.

Sunday

Measure and repeat

Note the time freed up and pick the next task for the following week.

With one automation that works and a real time figure, deciding on the next one is much easier. And once the team sees it working, they’re often the first to suggest it.

Collaborate

Did you find this useful?

I write about retail, operations and AI. If you’d like to comment on the article or collaborate on something similar, drop me a line.