n8n vs Make vs Zapier for Singapore SMEs in 2026

Most n8n-vs-Make-vs-Zapier comparisons are written by US bloggers benchmarking US SaaS pricing for US teams. None of them tell you what an SG SME with PDPA obligations, SGD budgets, and a Xero-plus-HubSpot stack should actually choose.

This guide does. Real SGD pricing (with annual discounts factored in), the integrations that matter for Singapore (Xero AU/SG, HubSpot, Shopify, WhatsApp Business, Talenox, PayNow), PDPA and data residency considerations, and an honest call on which tool fits which kind of business.

Key Takeaway: For most Singapore SMEs in 2026, Make is the fastest no-code prototype to get a workflow running in days, n8n (self-hosted) is the right choice if you need PDPA-grade data control or your workflows have grown complex, and Zapier is only worth its price if the rest of your stack already lives there. There is no "best" — each wins in a different scenario, and we'll break those scenarios down explicitly.

Written by Derek Chua, founder of Magnified Technologies. Magnified builds AI automations for Singapore SMEs and has shipped production workflows on all three platforms across F&B, professional services, healthcare, and B2B SaaS clients.

The 60-second answer

Skip the rest of this article if you already know your situation. Here's the short version.

• You're a 5–20 person SG SME running a few standard SaaS tools (Xero, Gmail, Shopify, HubSpot) and want a fast first automation: Make. Cheapest entry point, easiest to learn, fast to ship.
• You're a 20–100 person SG SME handling sensitive customer data, processing > 10,000 operations/month, or building workflows with conditional logic and complex routing: n8n self-hosted. Better data control for PDPA, no per-operation pricing ceiling, more power for complex flows.
• Your team is already on Zapier across the business, you're not pushing volume limits, and your workflows are simple: Zapier. Don't switch unless cost or limitations force you to.
• You're an engineering-led team that wants full code control and the workflows are core to the business: n8n self-hosted (or write it in code, but that's a different article).

For everything else, the long version is below.

The three platforms in one paragraph each

n8n is an open-source workflow automation tool that you can self-host on your own infrastructure (free, plus server costs) or use their cloud-hosted version (paid). It uses a node-based visual editor — you drag nodes onto a canvas and connect them — and it includes a JavaScript code node for anything the no-code blocks can't do. n8n is the most flexible of the three and the only one where you can fully control where your data lives (which matters for PDPA, regulated industries, and large data volumes).

Make (formerly Integromat) is a cloud-only no-code visual automation platform. It uses a "scenario" editor that's slightly more powerful than Zapier's linear-step model — you can route, iterate, and aggregate within a single scenario. Make's pricing is based on "operations" (each step counts as one), which works in your favour for low-volume workflows and against you when volume scales. It's the easiest to learn for non-technical users and the fastest to prototype.

Zapier is the original cloud-only no-code automation platform. Its strength is the breadth of integrations — Zapier has built connectors for almost every SaaS tool you can name. Its weakness is that the workflow editor is linear (step-by-step, no branching without paid add-ons), and the pricing scales aggressively once you cross volume thresholds. Most workflows that start on Zapier eventually get rebuilt on Make or n8n once volume or complexity grows.

Pricing — in SGD, with the catches

US pricing pages list everything in USD with monthly billing. Here's what it actually costs in Singapore as of mid-2026, with the annual-discount math done.

n8n

• Self-hosted (Community Edition): Free for the software. Server cost: ~S$30–S$60/month on a basic DigitalOcean droplet, Hetzner, or a small AWS/GCP instance. That's the realistic cost for SME volume. Pro: you control everything, no per-operation limits, full PDPA data control. Con: you need someone who can run a Docker container.
• n8n Cloud Starter: ~S$30/month (paid annually). 2,500 workflow executions/month, single user.
• n8n Cloud Pro: ~S$70/month (paid annually). 10,000 executions/month, 5 users.
• n8n Cloud Enterprise: Custom (typically S$1,000+/month). SSO, audit logs, advanced governance.

Make

• Free tier: 1,000 operations/month, 2 active scenarios. Useful for testing but not for production.
• Core: ~S$13/month (paid annually). 10,000 operations/month.
• Pro: ~S$22/month. 10,000 operations/month with faster execution and more features.
• Teams: ~S$45/month. 10,000 operations, multi-user.
• Operations math: A single workflow run typically uses 5–20 operations depending on complexity. 10,000 operations covers ~500–2,000 workflow runs/month. Above that, jump to higher tiers fast.

Zapier

• Free: 100 tasks/month, single-step Zaps only. Almost useless for real work.
• Professional: ~S$25/month (paid annually). 750 tasks/month. Multi-step Zaps.
• Team: ~S$85/month. 2,000 tasks/month.
• Company: ~S$133/month. 50,000 tasks/month.
• Task math: Each step in a Zap counts as one task. 750 tasks covers ~100–250 workflow runs/month at typical complexity. Zapier's per-task cost is 3–10× higher than Make's per-operation cost at equivalent volumes.

What the pricing actually says

For an SG SME running 5–10 small workflows with modest volume, Make at S$13/month is the cheapest, fastest way to start. Same workflows on Zapier would cost S$85–133/month for the higher tiers needed to cover the volume. Same workflows on n8n self-hosted would cost S$30–60/month plus the time of setting up the server (which is a one-off, maybe 2–4 hours if you've used Docker before).

If you scale to 10–50 active workflows or move into volume territory (50,000+ operations/month), n8n self-hosted wins on cost decisively — there's no per-operation pricing ceiling. Make at that volume would cost S$300+/month; Zapier would cost S$500+/month; n8n self-hosted is still S$30–60/month.

Integration depth — which platform connects to what you actually use

For Singapore SMEs, these are the integrations that come up most often in our scoping calls.

Xero (accounting)

• n8n: Full native integration. All standard endpoints (invoices, contacts, bills, bank transactions). Good.
• Make: Full native integration with the largest action library of the three. Best.
• Zapier: Full native integration. Most popular Zapier integration for SG SMEs. Good.

All three handle Xero well. Make has the deepest native action library, which matters if you're doing complex accounting workflows. For straightforward "new invoice paid → update HubSpot deal" workflows, any of the three works.

HubSpot

• n8n: Native integration with most common endpoints. Custom API node available for anything missing.
• Make: Deep native integration. Strong.
• Zapier: Deepest native integration of the three (HubSpot and Zapier have a particularly tight partnership).

Zapier wins on HubSpot if you're building HubSpot-heavy workflows. The native triggers and actions cover more edge cases than the other two.

Shopify

• n8n: Solid native integration.
• Make: Solid native integration with multiple variants (Shopify Plus, multi-store).
• Zapier: Solid native integration.

All three are good. Pick based on the rest of your stack, not on Shopify alone.

WhatsApp Business (Meta Cloud API)

• n8n: Native node available, plus full HTTP API access for any edge case. Best for complex WhatsApp flows.
• Make: Native module available. Good for standard send/receive flows.
• Zapier: Integration available but less feature-complete. Workable for basic flows.

For WhatsApp Business specifically, n8n is the strongest because you can fall through to direct API calls for anything the native node doesn't cover — and the Meta Cloud API has a lot of edge cases.

Talenox (SG payroll)

• n8n: No native integration. HTTP API access works if Talenox exposes an API.
• Make: No native integration. Same — webhook/API integration is doable but custom.
• Zapier: No native integration.

None of the three has a native Talenox integration as of mid-2026, despite Talenox being one of the most-used SG payroll tools. If your workflow needs Talenox, you'll be doing webhook or HTTP API work regardless of platform.

PayNow / NETS / SG bank integrations

• n8n: No native integrations. Most SG bank APIs are accessed via OAuth + HTTP — n8n's HTTP node handles this with proper auth handling.
• Make: Same as n8n. No native; HTTP works.
• Zapier: No native; webhook + HTTP works.

This is the gap in all three platforms. For SG-specific finance integrations, all three are equal — they all require you to handle the integration manually via HTTP. The difference is n8n's JavaScript code node lets you do more sophisticated request signing and response parsing inline; the others can do this but it's clunkier.

LLM providers (Claude, GPT, Gemini)

• n8n: Native nodes for OpenAI and Anthropic. Direct HTTP for anything else. Best for advanced prompt management and chained LLM calls.
• Make: Native modules for OpenAI, Anthropic, Google AI. Good.
• Zapier: Native integrations for OpenAI and "AI by Zapier" wrapper. More limited for advanced use cases.

For AI-heavy automations, n8n is the strongest because you have full control over the API call, can chain models, do RAG with vector stores natively, and handle streaming if needed. Make is solid for single-call use cases.

PDPA and data residency — the part the US blogs don't cover

If your workflow processes personal data of Singapore residents, the PDPA applies. Three things to know about each platform:

Where the workflow executes (data residency)

• n8n self-hosted: Wherever you put the server. AWS Singapore region, Hetzner Singapore, your own datacenter. You control this completely. Best for sensitive data.
• n8n Cloud: Hosted in EU and US regions. As of mid-2026, n8n Cloud does not offer a Singapore region. This means PDPA-sensitive data leaves Singapore during processing.
• Make: Hosted in EU. Data leaves Singapore during processing. Same PDPA cross-border transfer rules apply.
• Zapier: Hosted in US. Data leaves Singapore. Same rules.

For most SME workflows, cross-border processing is acceptable under PDPA as long as the cross-border data transfer rules are met (the receiving country has comparable protections, or specific contractual safeguards are in place). EU-hosted (Make) is easier to justify than US-hosted (Zapier) because EU GDPR is comparable to PDPA. But for sensitive workflows — health data, financial records, large customer datasets — self-hosting n8n in Singapore eliminates this whole question.

LLM provider data flow

If your workflow sends data to Claude, GPT, or Gemini, the LLM's data handling matters more than the automation platform's. All three platforms call the LLM the same way — over HTTPS to the model provider's API. Anthropic and OpenAI's enterprise APIs do not train on inputs by default; their consumer products do. Choose the LLM provider and configuration carefully and document it in your data processing register, regardless of which automation platform you use.

Audit logging

• n8n self-hosted: You control the logs, retention, and access.
• n8n Cloud Enterprise: Built-in audit logs.
• Make: Audit logging on higher tiers only.
• Zapier: Audit logging on Company tier and above.

For PDPA-regulated workflows that need an audit trail, this is a real consideration. Self-hosted n8n is the only option that gives you full log control out of the box.

When to pick each one

Pick Make when

• You're starting from zero on automation and want a working first workflow in 1–2 weeks
• Your workflows are low-to-medium complexity (less than 15 nodes per workflow)
• Your monthly volume is under 10,000 operations
• Nobody on your team has Docker experience or wants to learn
• The data sensitivity is normal commercial data (not health, financial, or regulated industries)

Real example: a 12-person SG marketing agency using Make to capture inbound leads from a contact form, enrich them via Clay, score them with Claude, and post the qualified ones to a Slack channel. Costs S$22/month. Took 1 week to build. Runs flawlessly.

Pick n8n self-hosted when

• You've outgrown Make's pricing or hit operation limits
• Your workflows touch sensitive data (customer health records, financial information, large PII datasets)
• You need workflows with complex branching, loops, sub-workflows, or programmatic logic
• You have an engineer or operations person comfortable running a Docker container and maintaining it
• You expect to build 10+ workflows over the next year

Real example: a 50-person SG B2B SaaS company self-hosting n8n on AWS Singapore, running 23 active workflows including AI-driven helpdesk triage, sales call transcription with HubSpot writeback, weekly Google Ads commentary, and an internal Slack bot over their knowledge base. Server cost: S$45/month. Time to migrate off Make once they outgrew it: 3 weeks for the engineer.

Pick Zapier when

• The rest of your stack is already configured in Zapier and switching has a real cost
• Your workflows are simple (3–5 steps, no branching) and low-volume (< 750 tasks/month)
• You're heavily integrated with HubSpot and want the deepest native integration
• You have no internal technical capacity and need the most beginner-friendly tool

Real example: a 6-person SG professional services firm with 4 Zaps connecting HubSpot to Calendly, Xero, Slack, and Mailchimp. Total monthly cost: S$25. Total monthly volume: 400 tasks. Working fine. No reason to change.

Same workflow, three platforms — what does it actually look like?

Take a simple workflow: "When a new contact form is submitted on Webflow, enrich the lead with Clay, score it with Claude, and post qualified ones to Slack."

• On Make: 6 modules. Setup time: 90 minutes. Cost per run: ~15 operations.
• On n8n self-hosted: 6 nodes. Setup time: 90 minutes (assuming the server is already running). Cost per run: 0 marginal (server is fixed cost).
• On Zapier: 6 steps. Setup time: 60 minutes (Zapier's UI is the most beginner-friendly). Cost per run: ~6 tasks.

At 100 form submissions/month: Make costs S$13/month, Zapier costs S$25/month, n8n self-hosted costs S$45/month. Make wins at low volume.

At 1,000 submissions/month: Make costs S$22–45 (needs higher tier for 15,000+ operations), Zapier costs S$85 (needs Team tier for 6,000+ tasks), n8n self-hosted costs S$45/month. Make and n8n tied at medium volume.

At 10,000 submissions/month: Make costs S$300+/month (Pro or Teams tier scaled up), Zapier costs S$500+/month (Company tier), n8n self-hosted still costs S$45/month. n8n wins decisively at high volume.

Migrating between platforms

A common pattern is starting on Zapier, hitting the cost ceiling, and migrating to Make. Or starting on Make, hitting PDPA or volume limits, and migrating to n8n. Both happen often enough to plan for.

Things to know about migration:

• No platform has a "one-click migrate from competitor" tool. You will rebuild each workflow on the new platform.
• For 5–10 simple workflows, migration takes 1–2 days of work if you know both platforms.
• The hidden cost is in testing. Workflows that worked on the old platform have edge cases that may not transfer cleanly. Budget twice the build time for testing.
• Document your workflows before migrating. A workflow you can describe in plain English is one you can rebuild in a day. A workflow held together by tribal knowledge in someone's head is the one that breaks for 6 weeks after the migration.

If you're scoping a migration, our AI automation services include migration projects from Zapier or Make to n8n, typically taking 2–4 weeks depending on workflow count and complexity.

Frequently asked questions

What's the absolute cheapest way to start with workflow automation in Singapore? Make's free tier (1,000 operations/month, 2 active scenarios) is genuinely usable for testing. If you have one or two simple workflows running at low volume, it can stay free. The Core tier at S$13/month covers most SG SME first-real-workflow needs. Avoid Zapier's free tier — 100 tasks/month and single-step Zaps only is too limited to be useful.

Is n8n really free? The Community Edition (self-hosted) is genuinely free software with no per-execution limits. The only cost is your server (~S$30–60/month on basic cloud hosting). The Cloud version is paid. Self-hosting requires basic Docker familiarity — if you've never deployed a containerised app, budget 4–8 hours to learn or pay someone for the initial setup.

Can I run n8n on my existing AWS / DigitalOcean / Hetzner / Vercel? On AWS, DigitalOcean, Hetzner, GCP, and Azure — yes, easily. n8n has Docker images and one-click deploy templates for most providers. Vercel isn't a good fit because n8n needs a persistent server, not serverless functions. For SG SMEs, DigitalOcean Singapore region or Hetzner are the most cost-effective; AWS Singapore is fine if you're already on AWS.

Which is the best for AI automation specifically? n8n. The reasons: native nodes for Claude and GPT, full JavaScript code node for prompt chaining and RAG, native vector store integrations, streaming support, and the flexibility to handle the long-running operations that AI workflows often require. Make is solid for single-LLM-call workflows. Zapier is the most limited for AI use cases of the three.

Can I use these for ABM or sales engagement workflows? Yes. All three connect to HubSpot, Salesforce, Apollo, Clay, and Outreach. Zapier has the deepest HubSpot integration. n8n has the most flexibility for custom logic and high volume. Make is the cheapest for moderate volumes. The "best" depends on your stack and your scale.

Are there PDPA-specific certifications for these platforms? None of the three has a Singapore-specific certification. n8n self-hosted lets you sidestep this question entirely by keeping the data in Singapore. Make and Zapier rely on standard cross-border data transfer mechanisms under PDPA, which are workable for most commercial workflows but require documentation. For regulated industries (financial services, healthcare), discuss with your DPO before committing to a cloud platform.

Should I use Power Automate (Microsoft) instead? Worth considering if your team is heavily on Microsoft 365 and uses Outlook, Teams, SharePoint, and Dynamics 365 day-to-day — Power Automate's integration with the Microsoft stack is unmatched. For everyone else, Power Automate's UI is dated, the workflow editor is less intuitive than Make or n8n, and the pricing is hard to predict. We've left it out of this comparison because it's effectively a different category for non-Microsoft-first businesses.

What about Pipedream, Workato, Tray.io? Pipedream is closer to n8n (code-first, developer-friendly, has a free tier). Workato is enterprise-tier pricing (S$10k+/year minimum). Tray.io is also enterprise. For SG SMEs in the S$2M–S$30M revenue band, the practical choices are still Make, n8n, or Zapier. Pipedream is worth a look if your team is engineering-led and you don't want to host n8n yourself.

How do I decide if a workflow is worth automating in the first place? Three criteria: it happens at least 10 times a week, it's repetitive and rules-based, and the cost of a wrong first draft is low. Our AI Automation Agency buyer's guide covers this in more detail, and our 15 AI Automation Workflows for SG SMEs lists 15 concrete examples that meet these criteria.

---

If you're evaluating which platform to commit to for a multi-workflow automation programme, our AI automation services start with a free 30-minute scoping call covering the platform decision in your specific context. Either we'll help you scope a build or we'll tell you which tool to use and what to watch out for — no sales pitch.