You have read that Make runs your business while you sleep. Then you open the scenario builder, stare at the canvas and close the tab. That first attempt is the gap between the claim and the reality.
Here is the reality: five small scenarios are enough to cover the jobs that eat a two-person service business. Measured on a two-person home cleaning company with 60 leads, 12 open quotes, 15 invoices and 45 jobs a month.
Each one gets the same two questions. What does a run cost in credits and hands-on minutes against doing it by hand, and exactly where must the automation stop and hand back to you?
The five together use 557 credits a month, under the Free plan's 1,000. But Free allows only two active scenarios, so the real decision is which two to switch on first.
The Build at a Glance
1. Lead Reply Within 90 Seconds
The catch before you build: the AI draft will promise a slot your calendar does not have, and mail from a brand-new address often lands in spam. Send from a domain you have authenticated, and read every draft yourself for the first week.
The flow is short: the website form fires a webhook the moment someone submits it. That's the form pinging Make with the details, no polling. A webhook is the trigger; the form does not wait for anyone to check it.
Next an HTTP step posts those details to an AI email API. That's the plain interface an AI service exposes so software can send it a request and get text back.
The prompt stays a fixed template, never free-form writing: [name], [service] and [preferred day] filled straight from the form, and an instruction to offer no time slot the calendar has not confirmed. Pasted into the API step, it reads:
Prompt to copy
You write one short reply to a new cleaning enquiry. Use only these filled slots: [name], [service], [preferred day]. Confirm we will call within the day to agree a time, never promise a slot. Under 70 words, warm and plain, no prices, no free-form additions.
Gmail sends it and the CRM gets a new record. Each run costs 4 credits; 60 leads is 240 credits a month. By hand the reply takes 12 minutes per lead.
The AI takes 8 seconds. Reading the draft and checking the calendar takes 2. The email API runs about $0.02 a message, so $1.20 a month.
A lead answered inside 90 seconds is one the owner never sees at 10 p.m. They sleep, and the company is still the first cleaner that prospect hears from.
💡 Tip: Send from an authenticated domain, not a brand-new address, or Gmail batches the reply straight to spam and the speed buys you nothing.
💡 PRO TIP
Cap the AI's role, not its voice
Keep the prompt a template with slots filled from the form. Letting the AI draft freely means a friendly but wrong offer of a Saturday morning you do not have, which you then have to retract by phone.
2. Quote Follow-Up That Nudges Twice
Twenty-two weekday checks a month, twelve of them finding a live quote, cost 68 credits in total. An empty check burns 2 credits and 2 minutes of your own time, so the 9 a.m. search runs whether or not anyone is waiting on an answer.
On a match, 4 credits buy a drafted nudge that goes through an HTTP step to Gmail, and the CRM logs it.
A day 3 branch and a day 7 branch, then the scenario leaves that quote alone; a reply cuts the branch dead, so nobody gets chased twice.
Building it, a matched quote drops from about 10 minutes of drafting to 10 seconds of AI time plus a minute of your checking.
Ledger: 22 checks and 12 matches run 68 credits a month, and a dozen quiet quotes get two polite nudges with nothing to remember.
| Workflow | Trigger | Credits per run | Runs a month | Credits a month | By hand per run |
|---|---|---|---|---|---|
| Lead reply | Website form webhook | 4 | 60 | 240 | 12 min |
| Quote follow-up | Weekday 9 a.m. schedule | 2 empty / 4 match | 22 | 68 | 2 empty / 10 match |
| Invoice chaser | Daily schedule | 2 empty / 4 match | 30 | 90 | 2 empty / 8 match |
| Review request | Job-complete trigger | 3 | 45 | 135 | 4 min |
| Sunday owner brief | Sunday 6 p.m. schedule | 6 | 4 | 24 | 45 min |
3. Invoice Reminder Before It Goes Late
An invoice reminder is a scheduled Make scenario that reads your accounting app and sends a plain email. Every morning it looks for invoices due in three days and any already past due, then drops paid and disputed ones before Gmail sees them.
The limit is the endgame. A final demand, a collections threat, anything that pushes toward legal steps should never leave an automation, and a customer who proposes a payment plan switches the scenario off that hour.
That last conversation belongs to the owner, not to a scheduled trigger. Thirty checks and fifteen matches cost 90 credits a month, and a matched invoice takes 8 minutes by hand against 1 minute of review.
The rest of the job runs itself. Fifteen calm reminders land before a balance becomes an awkward call, and the relationship stays intact because the firm part was never handed over.
💡 Tip: Write the reminder as a statement of fact, not a nudge: invoice number, amount, due date, payment link. No apology, no exclamation marks.
4. Review Request After Every Clean
Forty-five review requests go out this month, and nobody types one of them. The trigger is the job-complete event in the scheduling app.
Make catches it, sends the email through Gmail with the review link in it, then writes a line to the CRM so the request is on the customer's record.
One template does the whole month. It carries the customer's first name and the service they booked, pulled from the job record, and nothing else changes between requests.
No per-job wording, no rewrite, no AI step at all, because a review ask does not need one.
Three credits a run across 45 runs is 135 credits a month. Doing it by hand runs about 4 minutes a request, so the review you always skipped on a busy Friday takes a minute of checking instead of four minutes of writing.
🛠️ Build Order for a Two-Scenario Start
Pick your two scenarios
Choose based on the leak that costs you most: cash flow or quiet quotes.
Connect the accounts
Authorize the form, Gmail, CRM and accounting app inside Make's connection settings.
Build the trigger first
Set the webhook or schedule, run it once and confirm it fires with sample data.
Add the modules one at a time
Run the scenario after every module so you see exactly where a field is missing.
Watch the first week
Read every AI draft and log the credit spend before you trust it overnight.
5. Sunday Night Owner Brief
Yes, and it only reads what the other scenarios already wrote down. A Sunday 6 p.m. scheduled trigger runs three searches, one each in the CRM, the accounting app and the calendar, then an HTTP step passes those totals to an AI summary.
The email lands in Gmail or Slack. No module writes back, so the brief cannot change a single record.
Six credits a run, four runs a month is 24 credits. By hand, pulling the week's leads, cash and jobs takes about 45 minutes of tab-hopping.
The AI writes the summary in about 20 seconds, and checking it against the source apps takes 5 minutes. Keep customer names, card data and private notes out of the AI step, totals only.
This is where a privacy slip would cost you real money.
The two times: 45 minutes by hand against 20 seconds of AI plus 5 minutes of checking. One email then gives you the week's leads, cash and jobs, and Monday starts with decisions instead of five open tabs.
💡 Tip: Build the brief only after the other four scenarios are logging. It summarizes their output, so an empty CRM gives you an empty Monday email.
⚠️ COMMON MISTAKE
Do not run all five on Free
Make's Free plan allows only 2 active scenarios at once, and schedules no more often than every 15 minutes. The 557 credits fit under 1,000; the five active scenarios do not.
Frequently Asked Questions
Do I need to know how to code to build these?
No. Each scenario uses Make's built-in modules and one HTTP step to call an AI email or summary service. The HTTP step needs a URL and a key pasted from that service's dashboard, not code you write.
What happens when my Make credits run out?
Scenarios stop until the cycle renews or you add capacity, and extra credits cost more than your plan's own rate. Pay attention in the first month so a quiet weekend does not turn into a silent week.
Can I run the AI email reply without paying for an API?
You can, but you will swap API cost for manual copy-paste. The API route is what keeps the reply under 90 seconds, because the whole chain runs without you touching it.
Why does an empty scheduled check still cost credits?
On Make, the trigger module spends a credit each time it runs, and scheduled checks that find nothing still count. That is why the quote and invoice scenarios show 2 credits for empty checks, not zero.
Can I move these scenarios to Zapier or n8n later?
You will rebuild by hand, because none of the three tools imports another's workflows. Make blueprints export and re-import as JSON inside Make itself, so keep a copy if you plan to experiment.
Two Scenarios First, Then Decide on the Paid Plan
If cash is the leak, build the lead reply and the invoice chaser first. The lead reply costs 240 credits and protects the 90-second window; the invoice chaser costs 90 and keeps 15 invoices a month from aging.
If quiet quotes are the bigger drain, swap the chaser for the quote follow-up at 68 credits. Both pairs sit under the Free plan's 1,000 credits, because the limit that bites is the two active scenarios.
All five together use 557 credits a month, which fits the credit allowance. Running them needs a paid plan, and Make's current paid prices are on its pricing page. Start with two, read every draft for a week, then add the third.