
On this page
Most automation projects fail before they start, because the first one is too big.
Someone maps out eleven workflows, connects four tools, and by week three the team has turned half of them off because they fired at the wrong time.
Start with one. A small one. Then leave it alone.
Pick the step you hate
Not the most valuable step. The most annoying one.
Candidates that come up again and again:
- Copying a form enquiry into the diary.
- Texting a customer to confirm tomorrow’s appointment.
- Sending the same “here is your quote” message.
- Telling the team that a call was missed.
Any of these is a good first automation, because you already know what “working” looks like.
The shape of a small automation
Every automation is the same three parts:
- A trigger. Something happens.
- A condition. Optional, but usually one rule.
- An action. Something else happens.
Example: when a booking is created (trigger), if it is more than 24 hours away (condition), send a confirmation with a reschedule link (action).
That is it. No branching, no scoring, no third-party integration.
Write the message first
Before you touch the settings, write the exact message you want the customer to receive. Read it out loud.
If it sounds like a robot, the automation will too, no matter how well it is configured.
Keep it short, sign it with a person’s name, and give the customer a way to reply.
Test it on yourself
Put your own number in as the customer. Trigger the automation. Read the message on your phone, at the time a real customer would get it.
Three things to check:
- Does it arrive at a sensible hour?
- Does the link work on a phone?
- Does a reply go somewhere a person will see?
That last one is where most first automations fail. The message goes out, the customer replies, and the reply lands in a mailbox nobody opens.
Turn it on for one service
Do not automate every booking. Pick one service, or one day of the week, and let it run.
This limits the damage if something is wrong, and it gives you a comparison: the automated group and the manual group.
Leave it alone for four weeks
The temptation is to keep tweaking. Resist it for a month, or you will never know which change helped.
After four weeks, look at three numbers:
- How many messages were sent
- How many customers replied
- What changed in the outcome you cared about (no-shows, callbacks, time spent)
If the third number did not move, the automation is not solving the problem you thought it was.
When to add the second one
Add a second automation only when the first one is boring. Boring means it runs, nobody talks about it, and the numbers are stable.
Then pick the next annoying step. Not the next impressive one.
The failure modes to expect
- Duplicate messages. Two automations triggering on the same event. Check before adding.
- Wrong time zone. A 9am message arriving at 4am. Test with a real phone.
- Replies going nowhere. The most common and the most damaging.
- Automating a bad process. If the manual version confuses people, the automated one will confuse them faster.
What not to automate first
- Complaints.
- Anything that needs an apology.
- Pricing negotiations.
- The first reply to a new enquiry, if you can answer it personally within the hour.
Those are the moments where a human voice is the product.
Where BittaFlow fits
BittaFlow’s automation builder uses the same three parts: a trigger, a condition, and an action. You can run one automation on one service, watch the replies in the shared inbox, and add the next one when the first is boring.
Set up one. Test it on your own phone. Then leave it alone.


