Shopify Flow vs Zapier vs Custom Bots: What I Run

Shopify Flow vs Zapier vs custom bots, compared from experience: I run Flow on five stores and my own bot stack. Decision table by use case and scale.

Linh Nguyen · Updated 2026-07-04

I've made this exact choice with real stores on the line. Across the five Shopify stores we operate, Shopify Flow handles the in-admin rules, a custom bot stack we built on the Admin API handles everything cross-store, and Zapier is the tier I evaluated seriously and decided not to keep. "Automation" spans three very different tiers — Shopify Flow (native, free, inside one store), Zapier/Make-style connectors (glue between Shopify and outside apps), and custom bots built on the Admin API (full control, needs a developer). Pick the wrong tier and you either overpay for something Flow does free, or spend months fighting a connector that was never designed for your scale. This post compares all three and ends in the decision table I wish someone had handed me — by use case and store count.

The Three Tiers at a Glance

Now let's look at what each tier actually does — verified against the official docs as of mid-2026, plus what I run in production where I do.

Tier 1: Shopify Flow — Native, Free, Per-Store

Shopify Flow is Shopify's own ecommerce automation platform. Per the Shopify Help Center, it's a free app available on the Basic, Grow, Advanced, and Plus plans — it stopped being a Plus-only perk years ago. Every workflow is built from three blocks:

• Trigger — an event that starts the workflow (order created, product added, inventory quantity changed, or a Scheduled time trigger that fires on a timetable, at most once every 10 minutes per Shopify's docs).
• Condition — optional logic that filters when the workflow continues (order value over $200, customer tagged "wholesale").
• Action — what happens (add a tag, hold fulfillment, send an internal email, update a metafield).

A few capabilities are gated by plan:

• Send HTTP Request action (call an external API from a workflow): Grow, Advanced, and Plus only.
• Tasks from custom partner apps: Plus only.
• Usage limits scale with your plan's API limits, so a Basic store gets less throughput than a Plus store.

Recent updates in Shopify's Winter '26 Editions added the ability to test workflows before going live and to cancel runs — small things that make Flow feel less like a black box.

Where Flow shines: anything that lives entirely inside one store. On our stores that's order tagging, high-risk holds, and inventory pings — the full audit of what I keep in Flow across five stores is its own post. Zero cost, zero latency concerns, no third-party data processor to vet.

Where Flow stops: the moment your automation needs to touch a second store or a tool Flow doesn't integrate with. Flow is installed and configured per store. You can export a workflow as a .flow file and import it into another store manually, and Plus organizations can copy workflows between their stores — but there is no "run this workflow across all 12 stores" button, and no consolidated view of what fired where.

Tier 2: Zapier and Make — Connectors Between Apps

When the job is "when X happens in Shopify, do Y in another app," connectors are the default answer — and before we built anything ourselves, I spent real time evaluating whether they should be ours. Two dominate:

Zapier. The official Shopify integration offers 20+ triggers (order created/paid/fulfilled, new customer, abandoned cart, product updated) and 30+ actions (create orders, customers, products, fulfillments; search and update records), connecting Shopify to thousands of apps — Google Sheets, Slack, HubSpot, QuickBooks, Klaviyo, Notion. Setup is genuinely an afternoon; Zapier's own getting-started guide walks it. One detail I only caught mid-evaluation: Shopify is a premium app on Zapier, which means you need a paid Zapier plan to use it in your Zaps. Your Shopify store itself can be on any plan.

Make (formerly Integromat). The official Shopify integration ships 27 modules — 4 triggers, 18 actions, 5 searches — inside a visual scenario builder connecting 3,000+ apps. Make has a free tier with no time limit, and its branching/iteration model handles more complex logic than a linear Zap.

I'm deliberately not quoting connector prices — both vendors change tiers and task quotas often, so check their pricing pages directly. The structural point is stable, and it's the one that ended my evaluation: connectors bill by usage (tasks or operations consumed per run). That's the trap at scale.

Where connectors shine: one or two stores, moderate order volume, and a need to reach apps Flow can't — pushing orders into a CRM, posting Slack alerts, filling spreadsheets. No developer needed.

Where connectors stopped for us:

• Cost scales with volume. Every order that flows through a Zap consumes tasks. At a couple hundred orders a day across 5 stores, usage-based billing grows uncomfortable fast — that projection, not any one price, is why we went a different way.
• Per-store plumbing. Each store is a separate connection; a 10-store operation means 10 copies of every scenario to build, monitor, and fix when a token expires.
• You still don't own anything. The workflows live in a vendor's cloud. If you care about that trade-off, the own-source vs SaaS comparison walks through it in depth.

Tier 3: Custom Bots and API Automation — Full Control, Needs a Dev

The third tier is the one we live on. Code (or a platform you own) talks directly to the Shopify Admin API. Shopify pushes events to you via webhooks — HTTPS callbacks fired on topics like orders/create or fulfillments/update — and your bot reacts: writes to a database, calls a carrier API, posts to Discord, updates ten other stores. If webhooks are new to you, start with our plain-English guide to Shopify webhooks, then see what a Shopify bot API setup actually looks like in practice.

Where custom shines:

• Cross-store by design. One bot subscribes to webhooks from every store you run and aggregates them — consolidated order feeds, portfolio-level revenue, stuck-shipment alerts across all stores. That last one is exactly how our own alerting works: a 17TRACK-based tracking lifecycle in our stack, because carrier silence is not a Shopify event.
• No per-task billing. Once built, marginal cost per automation run is near zero. High-volume operations that would burn connector quotas run flat.
• Any logic, any integration. Rate-limit-aware bulk updates, custom dispute-deadline logic, AI agents answering ops questions — the agent layer we've run since February 2026 sits on this tier, and building an AI agent for multiple stores is its build log.

Where custom costs you: somebody has to build and maintain it. You need webhook endpoints with HMAC verification, retry handling, and respect for Shopify's per-store API rate limits. For a solo operator with one store, that's overkill. For a portfolio operator, it's usually the only tier that actually fits — API-level automation (order sync, 17TRACK shipment tracking, consolidated finance, a Bot API) is what a portfolio needs, whether you write it yourself or buy something that already did.

Decision Table: Use Case × Scale

My read of the table: Flow first, always — it's free and native, so exhaust it before paying anyone. Connectors for cross-app jobs at small scale. API-level for anything cross-store or high-volume.

The Multi-Store Nuance Nobody Advertises

Here's the part that decided the argument for us: Shopify Flow is fundamentally per-store. Ten stores means ten Flow installations, ten copies of each workflow (manually imported, or copied via the org tools if you're on Plus), and ten separate places to check when something misfires. Flow has no concept of "all my stores." I know the drift this causes firsthand: in a March 2026 audit of our five stores, our high-risk flagging rule existed in three slightly different versions, because two stores got an update the other three never did.

Connectors inherit the same shape — one connection per store — and add usage billing on top. So the real question for a multi-store operation isn't "Flow vs Zapier"; it's when do I move to API-level tooling that treats the portfolio as one system? The trigger points I'd watch for, because they were ours:

1. You're rebuilding the same workflow for the 4th store and dreading store #5.
2. Your morning routine involves opening multiple admin tabs to answer "how did we do yesterday?" — the exact problem a single multi-store dashboard solves.
3. Connector task bills are climbing with order volume while doing the same simple job.
4. You need cross-store logic — dedupe customers across brands, portfolio-level dispute deadlines, one report across all payouts.

At that point, whether you hire a developer to build bots or adopt a platform with the bot layer already built, you're in tier three. The good news: tier three doesn't replace Flow. We kept every one of our Flows the day the bot stack went live. Keep Flow doing free in-store housekeeping, keep a connector or two for lightweight cross-app glue if they're earning their keep, and put the portfolio-level nervous system on the API.

How to Choose in 5 Minutes

1. List your top 10 repetitive tasks. Mark each as in-store (one store), cross-app, or cross-store.
2. In-store → Flow. Build it today; it's free on every plan from Basic up.
3. Cross-app, low volume → Zapier or Make. Prototype on a trial before committing; remember Shopify needs a paid Zapier plan.
4. Cross-store or high volume → API tier. Budget dev time, or evaluate a multi-store platform on your own stores before deciding to build.
5. Re-audit every quarter. Automations rot: stores get added, tokens expire, workflows silently fail. Whichever tier you use, monitoring is part of the job.

Automation isn't one product — it's a stack. On our five stores that stack is Flow plus the API tier, with the connector tier consciously skipped; yours might legitimately keep a connector in the middle. The operators who scale calmly are the ones who chose each tier on purpose, each doing what it's cheapest and best at.