Set Your Automation Triggers — Designing the "When" of Every Workflow

OUTCOME — What you'll have afterward

Following this tutorial leaves you with three things.

• Three triggers for your real work sketched in notes — one event, one schedule, one condition
• One of those taken to "Test trigger" success inside Zapier or Make (action wiring comes next tutorial)
• The automation grammar "when A happens, do B" — an eye that reads any automation tool the same way

📖 Term: What is a trigger? "Trigger" literally means the pull of a firearm. In automation, it is "the condition or signal that starts an automation." As pulling a trigger fires the bullet, a trigger firing starts the automation. A trigger defines the when of an automation. Without that "when," the automation doesn't even begin. Triggers are the heart of every automation.

Metadata

Item	Value
Time	30 min (fast: 20, unhurried: 45)
Level	Intermediate (1-1 & 1-2 recommended first)
Tools	One free account in Zapier / Make / n8n
Coding knowledge	None
Payment	None — free tier is enough
Prerequisite	1–2 repetitive tasks from your real work

The first wall is the "Choose a trigger" screen: hundreds of apps, dozens of subtly different events each. Without a framework, people close the tab. This tutorial collapses the choice to three types.

Where you are now: prerequisites next.

---

PREREQUISITES — What you need

Three things.

• One free account: Zapier (easiest onramp), Make (most flexible, 1,000 ops/month free), n8n (self-host = free forever). No preference? Start with Zapier
• 1–2 repetitive tasks from real life: not hypothetical. Things like "log incoming DMs to a spreadsheet," "Monday morning roll up last week's revenue"
• A notes app: somewhere to save trigger blueprints

Not required: coding skills, API knowledge, paid plan, destination apps (Slack, etc.) — this tutorial focuses on triggers only.

📖 Term: Task and operation Zapier counts one automation run as a "task"; Make calls it an "operation." 2026 free tiers: Zapier 100 tasks/month, Make 1,000 operations/month, n8n self-hosted unlimited. A "daily 9 a.m. report" burns ~30 tasks monthly — comfortable on any plan. You only brush the ceiling later with high-frequency automations like "check inventory every 5 minutes." Nothing to worry about today.

Environment check (30s): sign in. If you see a "Create Zap" or "Create a new scenario" button, you're ready. Don't click yet — we'll build together in Step 4.

Where you are now: tool, tasks, and notes app ready. You can skip WHY and still finish, but reading it gives you a diagnostic vocabulary when something breaks three weeks from now.

---

WHY — Why triggers are the heart of automation

Automation sounds intimidating at first. The good news: it has a grammar, and it's absurdly simple. However complicated English looks, the skeleton is "subject + verb + object." Automation is one sentence:

"When A happens, do B."

That's the whole thing. Inquiry arrives → send notification. User signs up → send welcome email. 9 a.m. → compile yesterday's data. Every automation is a combination of that sentence. A is the trigger, B is the action. Trigger defines when; action defines what.

Reason 1 — Without a trigger, automation doesn't even start. Actions are usually clear in your head. The hard half is when it should happen. Sloppy triggers mean the same action runs too often, fails when it should run, or fires at the wrong moment. A trigger is the alarm clock of the automation.

Reason 2 — Confusing types tangles your design. "Every morning at 9, send new inquiries" reads like schedule, but the real intent is event + batch wrapper. Blur the two and you get neither real-time nor calm batch — worst of both worlds. Pick the type first.

Reason 3 — Bad triggers cost money. "Check every 1 minute" = 1,440 tasks/day — burns Zapier's free monthly quota (100) in an hour. Make's free (1,000) in under a day. Granularity errors cost 10×–100× in spend.

Where you are now: now we build a decision rule.

---

STEPS — A five-stage walkthrough

Step 1. Distinguish the three types (5 min)

Three types cover every real automation.

Event — starts when something happens. "Form submitted," "email arrived," "payment completed," "Notion page added," "Linear issue created." In 2026, most SaaS pushes these instantly via webhook — no polling needed. Most common in business automation.

Condition — starts when a condition is met. "Inventory < 10," "rating > 4.5." A number or state crosses a threshold. Used for monitoring and alerting.

Schedule — starts at a fixed time. "Daily 9 a.m.," "every 15 minutes." The clock hits and the automation runs regardless of outside events. In 2026, cron expressions are mainstream in Make and n8n — "last business day of the month at 18:00" is a one-liner.

Three-type comparison

Type	Alternate names	Fires when	Canonical examples	Primary use
Event	Instant / on-X	External actor acts	Form submit, email, payment, Notion page add, Linear issue	Real-time response
Condition	Threshold / state-change	Number/state hits threshold	inventory < 10, rating > 4.5	Monitoring, alerts
Schedule	Cron / time-based	Clock reads specified time	Daily 9 a.m., every 15 min, 0 9 * * 1-5	Reports, rollups

Drill: 1) "When a new subscriber signs up, send welcome email." 2) "On the first of every month, compile last month's revenue into Slack." Answers: #1 event (external actor). #2 schedule (clock).

Step 2. Event vs. schedule (5 min)

The most confused pair. Same goal, two designs. "Handle incoming inquiries":

• Event: every submission → immediate Slack. Real-time; alert storm in busy hours
• Schedule: 9 a.m. daily → 24-hour summary. Calm; one-day delay

Situation	Recommended	Why
Real-time required (urgent support, payment)	Event	Delay = loss
High volume, low urgency (feedback)	Schedule (batch)	Protects attention
Low volume, medium urgency (2–3/week signups)	Event	Nothing to batch
Periodic reports, independent of events	Schedule	No real choice
100+ routine log events/day	Schedule (batch) or silent event	Real-time becomes poison

📖 Term: Real-time vs. batch Real-time = "process the moment it happens." Batch = "collect for a window, then process together." Event triggers are real-time; schedule triggers are batch. Neither is universally better. High volume + low urgency = batch protects your attention. Low volume + high urgency = real-time prevents loss.

Pick yours: pick one of your prepared tasks now. Event-shaped or schedule-shaped? This tutorial continues with "when a new Instagram DM arrives, log it to Google Sheets" — unambiguously event (external actor = the person sending the DM).

Step 3. Granularity — too wide vs. too narrow (5 min)

Once type is set, granularity determines whether your free tier survives the month.

• Too wide: "when a new email arrives" → spam included, 200/day firings, 199 unnecessary
• Too narrow: "subject exactly '[URGENT-VIP]'" → filters out real inquiries

Three granularity questions

• Q1: How many times a day? (target: 10–100)
• Q2: Fraction actually needing the action? (target: 80%+)
• Q3: Safely includes the cases you can't miss? (If false negative is catastrophic, widen)

Applied to our example

• Too wide: "any new Instagram activity" → likes/follows included, dozens of firings
• Too narrow: "DM with subject 'inquiry'" → Instagram DMs have no subject (design flaw)
• Right: "new DM on the business account" → 5–20/day, all relevant

Pseudo-filtering

on_new_instagram_activity(activity):
    # Too wide: process everything
    # process(activity)  ← burns tasks on every like/follow

    # Right granularity
    if activity.type == "direct_message":
        if activity.sender_type != "bot":
            trigger_automation(activity)
    # else: ignore, save tasks

Task consumption simulation (2026 free tiers)

Design	Fires/day	Tasks/month	Zapier free (100)	Make free (1,000)
"All activity" (too wide)	80	2400	24× over	2.4× over
"DMs only" (right)	12	360	3.6× over	Comfortable (640 left)
"Business-hours DMs" (narrower)	5	150	1.5× over	Comfortable (850 left)
"Daily 9 a.m. batch" (schedule)	1	30	Comfortable (70 left)	Comfortable (970 left)

Takeaway: for Zapier free tier, real-time event handling still needs paid or a schedule workaround. Make's free 1,000 ops/month handles "DMs only" comfortably. n8n self-hosted = no limit at all. The tradeoff is decided at design time.

Step 4. Move into Zapier / Make UI (7 min)

Zapier walkthrough, 7 clicks. Make = same structure, different words.

1. Create Zap → editor opens
2. 1. Trigger block at top → click it
3. App & Event search → "Instagram" (or your service)
4. Pick event — e.g., "New DM in Instagram for Business"
5. Account: "Connect Instagram for Business" → approve in popup
6. Trigger (configure): optional filter (defaults usually fine)
7. Test trigger: Zapier fetches one recent sample from your account. Success = trigger live

📖 Term: Webhook triggers — the 2026 standard If your service isn't in Zapier's event list, register that service's webhook URL externally to create a trigger. In 2026, nearly every SaaS supports this: Notion, Linear, Slack, Gmail, Stripe, Tally, Typeform, and hundreds more. In Zapier: "Webhooks by Zapier → Catch Hook." In Make: Custom Webhook module. Copy the URL Zapier/Make gives you, paste it into your service's webhook settings — done. Non-developers: this is literally copy-paste. AI agents (like an n8n AI Agent node or a Claude workflow) can also call these webhook URLs directly, enabling AI-initiated triggers — the pattern is mainstream in 2026.

Three common stumbles: (1) wrong event variant — "New message" vs "New DM" vs "New comment" look alike. Read word-by-word (2) no test data — if no recent DMs, empty result. Send yourself one, wait a minute, retry (3) personal account connected — "Instagram for Business" requires actual business account.

Step 5. Save Off, write the blueprint (5 min)

If Test passes, save the Zap but keep it Off. No-action-yet Zaps on ON pile errors. Actions come next tutorial.

In notes, save this blueprint — for each of three triggers.

[Trigger blueprint #1] Instagram DM logger
- Type: event
- Purpose: catch every customer DM in a spreadsheet, never miss one
- Fires when: new DM on business account
- Expected fires/day: 5–15
- Fits free tier?: Zapier N (150–450 tasks/mo) / Make Y (150–450 ops, 1,000 free)
- Next action (later): add row to Google Sheets + ping Slack #cs

The point: three months from now, you can reconstruct the "why" instantly.

Where you are now: three triggers on paper, one tested. Let's verify.

---

VERIFICATION — What "done" looks like

Five "yes"es and you're complete.

1. [ ] My 1–2 tasks written as "when A happens, do B" sentences in notes
2. [ ] Three triggers classified event / condition / schedule. Can you name each type in two seconds?
3. [ ] Daily firing estimate + free-tier fit written down. Answered Q1–Q3?
4. [ ] One trigger hit "Test trigger" success. Saw real sample data in the panel?
5. [ ] Zap/Scenario saved in Off state. Don't turn on a no-action Zap — errors only

If any is "no," common stalls: #2? Re-apply Step 1's question. #4? Revisit Step 4's "three stumbles" — usually wrong event variant or account permission.

Paper design → test pass → Off save = 80% done. Actions follow next tutorial.

---

VARIATIONS — When the situation differs

Variation 1 — Webhook trigger: pick "Webhooks by Zapier" → "Catch Hook." Zapier gives a unique URL; paste it into your service's webhook setting. In 2026 this covers almost every SaaS — Notion, Linear, Slack, Gmail, Typeform, Tally, Stripe, and more. For non-developers this is literally copy-paste.

Variation 2 — Condition via filter: "Daily 9 a.m., check 'inventory' → if any < 10, ping Slack" = schedule + condition filter. Zapier Filter(paid), Make Router/Filter(free).

Variation 3 — Manual trigger for safe testing: use Webhooks Catch Hook that you call yourself. Verify, then swap in the real trigger. Safer for customer-facing work.

Variation 4 — Cron: Make and n8n support cron (e.g., 0 9 * * 1-5 = weekdays 9 a.m.) for schedules like "last Friday of each month at 18:00." In 2026 this is a standard UI option — no terminal needed. Start with "daily 9 a.m." — learn cron when needed.

Variation 5 — AI agent trigger: an n8n AI Agent node or any HTTP-capable AI workflow can call a webhook URL directly, making AI the trigger source instead of a human action. Example: Claude classifies incoming email as urgent → calls your automation webhook → Slack alert fires. This pattern is mainstream in 2026 and is what "agentic automation" looks like in practice.

---

TROUBLE — Common stuck points

❓ Test trigger returns nothing Correct trigger still needs a recent real event. Send yourself a test DM and retry in a minute.

❓ Same trigger fires infinitely (loop) Most common disaster. Cause: action re-creates what the trigger watches. Example: "new Sheets row → Slack + add row to same sheet" → new row re-fires. Fix: different tab or filter "source != automation." Loops burn a free tier in ten minutes.

❓ "Task limit reached" Free quota gone. (1) Usage tab — which Zap consumed most? (2) Granularity too wide? (3) Move to schedule/batch? (4) Consider switching to Make (1,000 ops free) or n8n self-hosted. (5) Only then upgrade Zapier. Most "Zapier is expensive" complaints trace to granularity.

❓ Long delay (5+ min) Zapier free polls every 15 min. Real-time options: (a) upgrade paid (2-min polling) or (b) webhook-based Instant triggers. Make and n8n receive webhooks instantly — zero polling delay.

❓ Connection keeps dropping OAuth token expiry. Settings → Connections → Reconnect. Prevention: dedicated work account, skip 6-month password rotation.

❓ Still can't decide type "When does this happen?" External action → event. Threshold → condition. Clock → schedule. Stuck? Default to schedule — safest, easiest to debug, lightest on quota.

---

NEXT — What comes next

Thirty minutes ago you might have been staring at "Choose a trigger" about to close the tab. Now you've designed three triggers and tested one. A trigger alone does nothing. Next tutorial wires action (B) in and ships your first complete automation.

SPIN chapter 3 (Process phase), tutorial 2:

1. What is automation? ✓ (3-1)
2. Set your triggers ✓ (3-2 — this tutorial)
3. Wire your actions (in prep)
4. Agent actions — AI-powered action steps (in prep)
5. Running scenarios at scale (in prep)

Best use: over the next week, finish one trigger all the way to action — not three in parallel. Confirm one runs stably seven days, then start the next. Automations compound, but badly designed automations compound as debt. That's why you saved blueprints.

on_new_instagram_activity(activity):
    # 너무 넓은 트리거
    # process(activity)  ← 매 활동마다 task 소모

    # 적절한 granularity
    if activity.type == "direct_message":
        if activity.sender_type != "bot":
            trigger_automation(activity)
    # 그 외는 무시 → task 절약

[트리거 설계도 #1] 인스타그램 DM 기록
- 유형: 이벤트
- 목적: 고객 DM을 Sheets에 자동 기록해 놓치지 않음
- 발동: 비즈니스 계정 새 DM
- 예상 발동: 하루 5~15건
- 무료 한도 안: Zapier N (월 150~450 task) / Make Y (월 150~450 op, 1,000 무료)
- 다음 액션: Sheets 행 추가 + Slack #cs 알림