Which process should a small business automate first?

The one that passes the 5-check test: it repeats 10+ times a week, follows explainable rules, arrives digitally, is time-critical, and nobody on the team loves doing it. In most service businesses, that's inquiry handling — sorting, answering and booking incoming leads.

How do I calculate whether automation pays off?

Multiply the hours per week a task consumes by the hourly cost of the person doing it, then by 52 weeks. Compare that yearly cost against the automation's build cost plus running cost. If the automation pays back inside 6 to 12 months, build it; if not, wait.

Do I need to replace my existing tools to automate?

Usually not. Good automation connects the tools you already use — email, calendar, CRM — through their APIs. Be skeptical of any proposal that starts with replacing your whole stack: that's a migration project wearing an automation costume.

What are the most common automation mistakes small businesses make?

The big four: automating a broken process instead of fixing it first, starting with the hardest workflow instead of the fastest payback, buying tools before mapping the process, and building without guardrails so the first AI mistake kills trust in the whole system.

AI Process Automation for Small Businesses: Where to Start

// 01Why "AI transformation" is the wrong goal

Small businesses keep being sold the same story: everything changes, transform now or die. The result is usually a workshop, a slide deck, and nothing shipped. Meanwhile the actual opportunity is boring and concrete: somewhere in your company, a person spends two hours a day on work a system could do in seconds.

Our rule for every automation project: remove one specific, expensive bottleneck. Measure it. Then take the next one. Companies that automate this way build momentum and trust. Companies that "transform" build slide decks.

// 02The 5-check test

A task is a strong first automation when it checks all five boxes:

five-checks.test

run automation-candidate --check

repeats: same pattern, 10+ times a week

rule-based: explainable to a new hire in an afternoon

digital input: arrives as email, form, document or data

time-critical: faster handling = more revenue or happier customers

unloved: the team cheers when it's gone

→ 5/5 = build it · 3-4/5 = maybe · below = leave it alone

the test we run in every strategy call — steal it

// 03The usual suspects, by department

Every business is different, but after enough automation projects you see the same tasks pass the test again and again:

Area	Task that usually passes	What the automation does
Sales / inquiries	Sorting & answering incoming leads	Classifies by topic and intent, replies with the right info and a booking link, creates the CRM entry
Customer service	The same 10 questions, forever	Chatbot with your own knowledge base answers them; edge cases route to a human
Back office	Invoices, orders, forms → system	Extracts data from documents, validates it, files it where it belongs
Scheduling	Appointment ping-pong by email	Booking link plus automatic calendar and record updates
Reporting	Compiling the same numbers weekly	Pulls from your tools automatically, delivers the report on schedule
Internal knowledge	"Ask Martina, she knows"	Searchable knowledge system that answers with sources — even when Martina is on vacation

// 04The payback math (copy this)

Before anything gets built, do this calculation. It takes five minutes and kills bad projects before they cost money:

payback.calc

# example: inquiry handling, 40 inquiries/week

time_per_inquiry = 12 min (read, sort, reply, log)

hours_per_week = 40 × 12 min = 8 h

yearly_cost = 8 h × €35/h × 52 = €14,560

automation: build (one-time) + running (monthly)

payback question: does it pay back in < 12 months?

in this example: yes — several times over

adjust the numbers to your case — the structure stays the same

Two honest notes on this math. First, the freed hours only count if they get used — for sales, for service, for growth. Second, some benefits don't fit the formula: a lead answered in two minutes instead of two days converts better, and that's often worth more than the saved hours.

// 05What starting actually looks like

A first automation project, done right, is small and fast:

Map the current process — who touches what, in which order, where it stalls (1 workshop)
Define guardrails and the success metric — what the system may do alone, when it hands over, what number proves it works
Build and connect — to your existing email, calendar, CRM (2–4 weeks)
Warm-up phase — the system runs supervised, a human reviews its decisions until confidence is earned
Measure, then extend — the monthly report shows saved hours and response times; the next bottleneck is next

That's the pattern behind the EBZ Energie lead pipeline: it started as "answer inquiries faster" and grew into a system where the first human touchpoint is the consultation itself.

// 06The four mistakes that kill automation projects

anti-patterns.log

WHAT KILLS PROJECTS

Automating a broken process ("garbage in, faster garbage out")
Starting with the hardest workflow to "prove the value"
Buying tools first, mapping the process never
No guardrails — the first AI mistake kills all trust

WHAT MAKES THEM STICK

Fix the process first, then automate the fixed version
Start with the fastest payback, build momentum
Map first, choose tools second
Boundaries, handover rules and logs from day one