The Automation Guide · Free · No sign-up
What to automate first — and how to decide.
A plain-English guide for small business owners. How AI automation works, a simple way to score your tasks, the payback math, the tools, and the mistakes that sink most first projects.
Six chapters · 15-minute read · Updated September 2026
What it is
What AI automation actually is
Automation is a rule: when this happens, do that. A form is submitted, so a row is added to a spreadsheet and a welcome email goes out. It’s been around for years, and it’s great at moving information from one place to another.
AI adds the part that used to need a person: reading, sorting, summarizing and drafting. It can read an email and tell whether it’s a new lead, a complaint or an invoice. It can pull the numbers off a PDF, or draft a reply in your tone.
Put them together and you get a workflow where the automation does the moving, the AI does the reading and writing, and a person handles the exceptions. For example:
- A new inquiry arrives through your website at 9 p.m.
- AI reads it and pulls out the name, what they need and how urgent it is.
- The details go into your CRM — nobody retypes anything.
- A friendly reply goes out with a link to book a time.
- If it’s urgent, you get a text with the details.
The question that matters most isn’t “AI or no AI.” It’s whether you own the system or rent it.
What to automate first
What to automate first
The best first build passes a simple test. The task is:
- Frequent — it happens every week, ideally every day.
- Clear — the right answer is usually obvious; it follows rules.
- Costly — it eats real time, or money leaks out when it’s slow or forgotten.
Here’s how to find that task in five steps.
Write down the repeat work
For one normal week, you and your team jot down every task you do more than once. Don’t judge yet — just list it. Most owners are surprised by how long the list gets.
- How often it happens
- Roughly how many minutes each time
- Where information gets retyped from one place to another
Score each task
Give each task a 1–5 score on the four factors in the table below, then multiply them. The highest product is your best first build.
- Score honestly, not hopefully
- Multiply, don’t add — one weak factor should sink a task
- Break ties by picking the easier build
Do the payback math
Put a dollar figure on the top two or three tasks using the formula below. If a task won’t pay back within about a year, it isn’t your first build.
- Use real hours from your list
- Count money that leaks out, not just time
- Subtract the monthly software cost
Build the narrow version, where you own it
Automate one task, end to end, with a clear handoff to a person for anything unusual. Build it in accounts in your business’s name and test it on real examples before it touches customers.
- One trigger, one outcome
- A person handles the odd cases
- An easy way to pause it
Measure, then pick the next one
After a month, compare against your week-one list. How many hours came back? What still needs a person? Then go back to your scored list and pick the next task.
- Hours given back
- Leaks closed (missed calls, late invoices)
- Next task on the list
The score
Rate each task from 1 to 5 on four factors.
| Factor | Scores a 1 | Scores a 5 |
|---|---|---|
| How often | A few times a year | Many times a day |
| Time each time | Under a minute | An hour or more |
| Cost of a miss | Nobody notices | A lost customer or real money |
| Ease to build | Needs judgment or new systems | Clear rules, tools already connect |
Multiply the four scores (1–625). Multiplying punishes a weak spot the way real life does: a frequent, expensive task that needs judgment every time — or lives in software that can’t connect — isn’t your first build, however much it hurts. The Owner’s Playbook has a worksheet that scores twelve common workflows for you.
The payback math
Three lines on the back of an envelope.
Monthly value = hours saved per month × what an hour is worth + money that stops leaking
Monthly net = monthly value − monthly software & AI costs
Payback = build price ÷ monthly net
Example (illustrative): quote follow-up takes 6 hours a week — about 26 hours a month. At $50 an hour that’s $1,300 a month. Subtract $60 of software and the net is $1,240. A $3,000 build pays back in under three months, and keeps paying after that because you own it.
“What an hour is worth” is what you’d pay someone to do it, including payroll taxes — or, for your own time, what you could earn in that hour instead.
Where small businesses usually start
Front desk & sales
- Missed-call text-back
- Instant reply to new leads
- Self-booking with reminders
- Quote follow-up
Operations
- Intake forms into your system
- Job and order status updates
- Reading documents and PDFs
- Handoffs between team members
Money
- Invoice follow-up
- Receipt and bill entry
- Payment reminders
- A weekly numbers email
Marketing & reviews
- Review requests
- Win-back emails
- First drafts of posts and emails
- Past-customer check-ins
The tools
The tools, in plain English
Most small business automations run on one of three platforms. They all connect your apps; they differ in ease, pricing and who controls the data. The AI part — reading and writing — usually comes from a business AI service (like OpenAI, Anthropic or Google) connected with an account in your name. Business accounts from these providers don’t use your data to train their models by default.
Zapier
01The biggest library of app connections and the gentlest learning curve. Pricing is based on how many steps run each month, so costs grow as volume grows.
Make
02A drag-and-drop canvas that handles branching and multi-step logic well. Usually cheaper than Zapier at moderate volume.
n8n
03Can run on a server your business controls, which keeps data in-house and avoids per-task pricing at high volume. More technical to set up — which is where a builder helps.
You don’t need to pick. We choose based on your volume, budget and how sensitive your data is — and set it up in your business’s name.
Hire, rent or build
Hire, rent or build
There are three ways to get repetitive work off your plate. Each is the right answer sometimes.
/ Option A
Hire help
- Cost
- Wages, every year
- Productive from
- Weeks, after hiring and training
- Hours covered
- Business hours
- Best for
- Work that needs judgment and relationships
/ Option B
Rent software or a retainer
- Cost
- Monthly fees that continue
- Productive from
- Days, if it fits your process
- Hours covered
- 24/7
- Best for
- Standard processes a template fits
/ Option C
Build and own
- Cost
- From $3,000 once, plus your software & AI usage
- Productive from
- 30 days
- Hours covered
- 24/7
- Best for
- Repeatable work that’s specific to how you run
Your first 30 days
Your first 30 days
A first build doesn’t need a six-month project plan. Here’s how a typical Promptolution project runs from kickoff to live.
- Days 1–5
Discover
We learn how your business actually runs — talking with you and your team, mapping the work step by step, and ranking where AI will save the most time and money. Discovery is a paid first step ($1,000), deducted from the build price if you hire us to build. The map is yours either way.
You get: Workflow map · Prioritization map · Build costs
- Days 6–10
Design
A plain-English plan: what gets automated, which tools connect, what the AI will and won’t do, and when it goes live. You approve it before we build anything.
You get: Build plan · Tool choices · Timeline
- Days 11–22
Build
We build it inside accounts you own, then test it against real examples from your business — including the odd cases — until it holds up.
You get: Working automations · Connected tools · Test results
- Days 23–26
Launch
We switch it on, train your team, and hand over simple how-to guides — so it gets used on day 31, not forgotten.
You get: Go-live · Team training · How-to guides
- Days 27–30
Improve
We watch how it performs, measure the time it gives back, and tune what needs tuning. You get a clear list of what to automate next.
You get: Monitoring · Results review · Next opportunities
What goes wrong
Six ways automation goes wrong
Most failed automation projects fail for the same few reasons. All of them are avoidable.
Automating a mess
01If the process is unclear by hand, automation just makes the confusion faster. Tidy the steps first — often that alone saves time.
Starting with the hardest job
02The most painful task is rarely the best first build. Win with something frequent and clear, then take on the tricky one.
No plan for the odd cases
03Every process has exceptions. A good build sends them to a person with the context attached instead of guessing.
Nobody owns it after launch
04Automations that nobody watches quietly break. Name an owner, set up alerts, and check the results monthly.
Costs that grow faster than you do
05Per-task pricing looks cheap at 100 runs and expensive at 10,000. Check what the bill looks like at twice your volume.
Building in someone else’s account
06If the workflows live in your vendor’s account, you’re renting them. Insist on your business’s name on every account from day one.
Where to start
Rather skip the homework? We’ll score it with you.
Thirty minutes, no pitch deck. We’ll look at how your business runs, find the busywork worth automating first, and explain how Discovery works and what it costs. If AI isn’t the answer, we’ll say so.
- From $3,000
- Live in 30 days
- Yours from day one
- No lock-in