Make and Zapier: mental models and a real build
Two tools, one mental model
Zapier and Make (formerly Integromat) do the same job — they watch for a trigger in one app and fire actions in others — but they think differently, and clients will expect you to be fluent in both. Zapier thinks in a straight line. Make thinks in a canvas.
Everything you mapped in lesson 2 translates directly: your trigger becomes the first module, your conditions become filters and routers, your actions become the steps after. The tool is just your map with login credentials.
How each one thinks
| Zapier | Make | |
|---|---|---|
| Unit of work | A "Zap" — one trigger, then actions in a line | A "scenario" — a visual canvas of modules |
| Branching | Paths (limited, plan-gated) | Routers — branch as much as you want |
| Pricing logic | Per task (every action step counts) | Per operation — usually cheaper at volume |
| Strength | Most integrations, fastest to ship, client-friendly | Complex flows, iterating over lists, data transformation |
| Weakness | Gets expensive and clumsy when logic grows | Steeper learning curve, easier to build spaghetti |
The five building blocks
Every scenario or Zap you'll ever build is assembled from five parts. Triggers — polling (checks every 15 min) or instant via webhooks, a URL that an app pings the moment something happens. Instant matters: a lead answered in 5 minutes converts wildly better than one answered in an hour. Actions — create, update, send, add. Filters — "only continue if budget > 1000" — the condition gate. Routers/paths — multiple branches from one trigger, each with its own filter. Error handling — what happens when a step fails, which is the part amateurs skip and seniors get paid for.
The build: form → CRM → Slack → email sequence
This is the lesson 2 diagram, built for real. A client runs a consultation-request form and keeps losing leads. Here's the Make version, module by module:
Error handling keeps clients
Every automation will eventually hit a bad input, an expired login, or an API that's down for maintenance. The difference between a hobbyist and a professional is what happens next: silently lost leads, or a controlled failure someone hears about.
Do
- Add an error route that logs failures to a Google Sheet with a timestamp and the raw data
- Send yourself (not the client) an alert on failure so you fix it before they notice
- Use "resume" or "ignore" handlers deliberately — decide per step whether the flow should continue
Don't
- Ship a scenario with zero error paths and hope
- Let one malformed submission kill the whole run for everyone behind it
- Retry payment or email-send steps blindly — you can double-charge or double-send
Do this now
Twenty-five minutes. Create free accounts on Zapier and Make. In either one, build a 3-step automation: Google Form (or Typeform) submission → add a row to a Google Sheet → send yourself an email notification. Test it with one real submission and one garbage submission (blank fields, weird characters) and note what happened.
Then record a 2-minute Loom walking through the build: what triggers it, what each step does, what you'd add for error handling. That Loom is portfolio artifact #1 — you now have proof you can do the thing 45,000+ job posts are asking for at its simplest level.
Tip: use your ← → arrow keys.