What Should a Small Shop Automate First?

01 / Find the work

Measure the task, not the excitement around a tool.

For one ordinary week, note each repeated task: how often it happens, the minutes it takes each time, who does it, and what goes wrong. A five-minute task done twenty times can matter more than a long task done once a month. Include the time spent checking and correcting mistakes.

Mark where an error affects a customer: a wrong stock promise, missed order update or duplicate message has a different cost from a late internal report. Ask what a successful change would improve: staff time, accuracy, response speed or customer confidence. Do not count time saved if someone must spend the same time checking every automated result.

02 / Check readiness

Make sure the task is safe to repeat automatically.

• Rules and stability: can staff describe the trigger, steps, owner and finish point the same way? If the process changes each week, agree the process first.
• Data readiness: is there one dependable source for stock, order status or customer contact details? Check missing, late and duplicate records. Automation can repeat a bad source faster.
• Exceptions: what happens when an item is returned, an order changes, the source is blank or a customer needs a personal answer? Decide when work stops and who takes over.
• Dependencies: list the systems that must connect and who maintains each one. A reminder inside an existing system may be simpler than syncing three systems.
• Burden and benefit: include setup, staff training, subscriptions, monitoring, corrections and future changes. Compare that full effort with the time and errors you expect to avoid.

Do not automate yet when the process is unstable, the source data is poor, the work needs substantial judgement, the task is rare, or setup and maintenance cost more than the likely benefit. Keep a person responsible for decisions about unusual orders, complaints and sensitive customer conversations. Fix the routine or records, then review the idea again.

03 / Worked scenario

Five candidates in one neighbourhood shop.

Imagine a small homewares shop with an in-store counter and online orders. The owner samples a busy week and ranks the work below. This is a fictional ShopsInto scenario, not a universal order of priority.

1. Daily sales summary: the same figures are copied each evening from a dependable sales record. A scheduled report or saved view is a strong small first test: frequent, clear and easy to check against the original.
2. Order follow-ups: staff repeatedly check which paid orders are ready for collection. A reminder could help once order states are consistently recorded. Keep a person for changed, disputed or delayed orders.
3. Repetitive admin: a weekly supplier form uses stable fields. A template may remove more work than an integration; compare the small gain with setup and upkeep.
4. Stock updates: counts disagree between the shelf, counter and online listing. Customer impact is high, but an automatic sync would spread bad counts. First agree one stock record and a correction routine.
5. Customer messages: opening hours and collection rules can use prepared replies. Returns, complaints and unusual requests still need a person; do not auto-send a stock or delivery promise without a reliable source.

In this shop, the daily summary is a sensible pilot because the input and rule are ready. Another shop might start with order reminders if missed collections hurt customers more. Re-rank the candidates when the shop's records and workload change.

04 / Repeatable decision

Use the same questions for every candidate.

1. Volume: how often does it occur, how long does it take, and how much correction time follows?
2. Value: what error or delay would it prevent, and how would customers notice?
3. Readiness: are the steps stable, the rules clear and the input accurate enough?
4. Exceptions: which cases need judgement, and who receives them when the routine fails?
5. Cost: which systems must connect, who will build and maintain the change, and what will it cost over time?

For each answer, mark ready, needs work or poor fit. Prioritise candidates with meaningful value and mostly ready answers. Repair a “needs work” input or handoff before automating it. Keep a poor-fit task manual and revisit it only if the work changes. This is ShopsInto's practical discussion framework, not a scored product recommendation.

Test one small change against the original routine. Record time, errors and missed handoffs before and after, including the time spent monitoring. Keep a way to pause it and return to the manual process if the result is worse.