Count Your Clicks: Fix Order Screens Before Adding AI

Count Your Clicks: Fix Order Screens Before Adding AI

Your team may switch between four or five screens to finish one order. Daikin cut that and saw 10% better counter efficiency. Here is the click audit you can run this week, and what to automate after.

Most owners ask, "Where can we add AI?" The better first question is, "How many screens does one order take?"

A big HVAC company just showed why. Here is the story, and a 30-minute test you can copy.

What did Daikin actually do?

Daikin's North American business ran finance, logistics, procurement and sales on separate, older systems. EY and SAP helped it rebuild those workflows.

The pilot went live in 10 months. It covered more than 300 frontline users across dozens of branches. According to EY, the early results were a 10% improvement in counter efficiency and a 20% faster financial close.

One detail matters most for small businesses. Counter staff got single-screen workflows. They stopped hopping between systems to finish a sale.

The gain came from fewer screens, not from a clever model.

A caveat: this is an EY write-up of a large company, and the numbers come from the first wave of a pilot. Your results will differ. The method still works at any size.

Why count clicks before adding AI?

If you automate a messy process, you get a faster mess.

Say your counter person opens a chat app, a billing tool, a stock sheet and a delivery note for every order. An AI assistant on top of that still needs all four.

The newsletter that pointed us to this story gave a simple rule. If one order needs more than three screens, you are spending time on navigation instead of selling.

Fix the flow first. Then automate the parts that are left.

The 30-minute click audit

Do this with a pen and paper, or a shared sheet.

  1. Pick your highest-volume transaction. For a pasal, that is a normal sale. For a clinic, it is booking an appointment. For an agency, it is a new client enquiry.
  2. Watch one real transaction from start to finish. Do not ask for a description. Watch it.
  3. Write down every screen, app or paper form touched.
  4. Count the clicks and the times someone types the same data twice.
  5. Circle every place where data is copied by hand from one tool to another.

Those circled copy-paste steps are your first automation targets.

What should you fix first?

Start with the easiest wins:

  • Remove double entry. Enter the order once and let it flow to the sheet, the invoice and the stock list.
  • Put what staff need on one view: customer, item, price, stock.
  • Cut approvals that never get rejected.

Only then ask what an AI step can do. Good small jobs are reading a message and filling the order fields, or writing a daily summary.

What this means for businesses in Nepal

Many Nepali SMEs take orders in chat apps, phone calls and paper notebooks. Then someone types them into a sheet or billing software at the end of the day.

That is the same problem Daikin had, just smaller. The tools differ, but the pattern is the same: one order, many places, lots of retyping.

You do not need a large system. A single shared order sheet, filled automatically, is often enough to start. Count your clicks first, then decide.

This is also why we say a first project should be small and measured. We covered that in our last post on starting an AI project with a small budget.

How to automate this with n8n

Here is a simple version for a small shop that takes orders through a chat app.

  1. Trigger: n8n receives each new order message from your chat channel.
  2. Extract: an AI step reads the message and pulls out customer, item and quantity. Add a rule: if anything is unclear, flag it for a human.
  3. Log: n8n writes one row per order into a Google Sheet. This is your single screen.
  4. Confirm: n8n sends the customer a short confirmation.
  5. Summarise: each evening n8n totals the day's sales and sends you the summary.
  6. Measure: re-run the click audit after two weeks. Count clicks and minutes per order again.

A starting point: n8n template: Log Telegram orders with Google Sheets and send daily sales summaries. Adapt it to the channel you use.

Measure before and after. If the click count did not drop, the automation is not done.

FAQ

How do I measure counter efficiency? Time ten normal orders from start to finish and count the screens. Repeat after the change. Compare the averages. Keep it simple.

Do I need AI to cut the number of screens? Often not. Joining your tools so data is entered once solves most of it. AI helps with reading messy text, like free-form order messages.

Can a small business really copy an enterprise case like Daikin? Copy the method, not the scale. Counting steps and removing double entry costs nothing. Treat Daikin's percentages as an example, not a promise for your business.

Sources

Want help finding the one workflow worth automating first? Book a free automation call at https://solvex.com.np.