How to pick an automation platform that fits your business
Most businesses pick their automation platform the same way they pick a restaurant on vacation. They go with the name they recognize. Zapier has the brand, Make has the marketing, and n8n has the GitHub stars. The decision gets made on familiarity, not fit. We see it constantly. A 4-person agency signs up for Power Automate because they already pay for Microsoft 365. A solo founder picks Zapier because a blog post ranked first. A CTO chooses n8n because open source sounds right. None of them asked the five questions that actually determine which platform will cost less, break less, and scale with the business over the next two years. Wrong tool. Wrong price. Wrong fit.
The automation market now includes over 200 platforms across business automation, developer tooling, data pipelines, and enterprise iPaaS. The $26 billion market is growing at roughly 9% per year. More options should mean better decisions. Instead, more options mean more confusion, more defaults to brand recognition, and more money wasted on the wrong fit.
We've tracked how businesses choose and how often they switch. The pattern is consistent across industries. Teams that start with the tool end up migrating within 18 months. Teams that start with five honest questions about the business stay put.
The five questions that actually matter
Every automation platform decision reduces to five variables. Get these right and the platform practically picks itself. Get them wrong and you pay twice, once for the platform and once for the migration.
Question one: What is your monthly volume?
Count the total number of actions your workflows will execute per month. Not the workflow count, but total actions across all runs. A single workflow that processes a form submission, creates a CRM contact, sends an email, notifies Slack, and updates a spreadsheet is 5 actions per run. At 1,000 form submissions per month, that is 5,000 billable actions. Zapier charges per task, Make charges per operation, and n8n charges per execution where the whole workflow counts as one. The same workload costs $300 on Zapier, $10.59 on Make, and $24 on n8n Cloud. Volume is the single most expensive variable in automation, and measuring it first prevents every downstream mistake.
Question two: What is your team's technical level?
This requires honest assessment. If nobody on the team has configured a webhook, deployed a Docker container, or written a JSON transformation, a no-code platform is the only realistic option. One developer willing to spend a few hours per month on maintenance makes self-hosted options viable. A full DevOps team opens the entire range. Technical capability determines not just what your team can set up but what they can maintain when something breaks on a Saturday morning.

Question three: How fast will your volume grow?
This is the question everyone skips. Your current volume does not determine your platform. Your volume in twelve months does. We've watched businesses sign annual contracts at comfortable levels, then hit 3x growth by month six and face overage charges that double the original price. Zapier's overage rate is 1.25x the base per-task cost, and Make's overage rate is 1.25x per credit. Plan for your growth. Pick the pricing model that rewards it instead of taxing it.
Question four: What tools do you need to connect?
Count the specific integrations your workflows require. Zapier leads with 8,000+ app connections, Make offers 3,000+, n8n has 400+ built-in nodes plus nearly 2,000 community-built nodes, and Activepieces covers 200+. If your stack includes niche or industry-specific tools, check availability before choosing. A platform with 8,000 integrations is worthless if it does not connect to the ERP system that runs your operation. A platform with 400 integrations is perfect if those 400 cover everything you need.
Question five: Do you have compliance or data residency requirements?
Healthcare, finance, legal, and government organizations face data residency rules that eliminate most cloud platforms immediately. HIPAA, GDPR, and SOC 2 restrict where data can be processed. Self-hosted platforms like n8n and Activepieces keep data on your own infrastructure, which simplifies compliance audits. Cloud-only platforms route data through third-party servers in regions you may not control. Over 60% of enterprises now face data sovereignty regulations on cloud-hosted workloads. If you operate in a regulated industry, this question narrows your options more than any other.
Match your answers to a platform
Your five answers create a profile. Here is how that profile maps to the major platforms.
Low volume, non-technical team, slow growth, broad integrations needed, no compliance concerns. Zapier. The free tier covers 100 tasks per month, and the Professional plan at $19.99 per month (annual billing) covers 750 tasks. The onboarding takes 15 minutes, and nothing else matches the simplicity. Use it until volume makes the math untenable.
Moderate volume, team can handle visual logic builders, steady growth, 1,000+ integrations needed, standard compliance. Make. The Core plan at $10.59 per month covers 10,000 operations. The visual scenario builder handles complex branching, error handling, and data transformation better than Zapier's editor. Unused operations now roll over one month on paid plans. For growing teams that outgrow Zapier, Make is the obvious next step.
High volume, at least one developer, fast growth, AI workflow capabilities needed, infrastructure ownership preferred. n8n. Self-hosted community edition costs nothing per execution, with server hosting as the only expense at $5 to $50 per month depending on scale. n8n Cloud starts at 24 EUR per month for 2,500 executions. The AI Agent node built on LangChain, vector store integrations, and RAG workflow support are the strongest in the category. Over 180,000 GitHub stars and a $2.5 billion valuation after the October 2025 Series C.
Microsoft 365 shop that needs desktop automation. Power Automate. The Premium plan costs $15 per user per month and includes cloud flows and desktop flows in attended mode, while the pay-as-you-go option charges $0.60 per flow run. The deep integration with SharePoint, Teams, Dynamics, and the rest of the Microsoft stack is unmatched. But the pricing gets complicated fast with per-user and per-flow licensing, and the learning curve is steeper than Zapier or Make.
Open source preferred with the simplest cloud pricing. Activepieces. Self-hosted is free, and cloud starts at $99 per month for 50,000 tasks and 5 users. That flat rate removes the per-action math entirely. The integration library is smaller at 200+ pieces but growing fast. For teams that want open-source flexibility without the DevOps overhead of n8n, Activepieces deserves serious evaluation.
Developer who thinks in code. Pipedream. The free tier includes 100 credits per month, and the Advanced plan at $49 per month adds unlimited workflows. Pipedream charges per compute time rather than per action, which benefits lightweight workflows. The platform runs Node.js, Python, Go, and Bash natively. Not a business-user tool. Built for engineers who want code-level control.
The mistakes that cost you twice
We've analyzed hundreds of platform selection decisions across the businesses we study. The same three mistakes appear in every industry, at every company size, and at every budget level. These mistakes account for most of the wasted money and forced migrations.
Mistake one: choosing by brand instead of by fit. Zapier owns the top Google results for almost every automation query, and the brand recognition is earned. But brand recognition is not a buying criterion. We've tracked businesses paying $500 per month on Zapier for workloads that cost $35 on Make, running the same workflows with the same integrations. Same workflows. Same integrations. The price difference comes entirely from the pricing model, not from product quality.
Mistake two: ignoring volume growth. Your current volume is a snapshot. Your pricing model is a trajectory. The per-task model was designed for an era when businesses ran 5 simple automations. That era ended. Every team that takes automation seriously builds more workflows, adds more steps to each workflow, and runs them more frequently. Both axes multiply together, and a business processing 500 tasks today while growing 20% monthly hits 1,500 tasks within six months. That growth changes which plan and which platform makes financial sense.
Mistake three: skipping migration cost analysis. Switching platforms costs real time and real money. A typical migration takes 20 to 80 hours depending on complexity, and at consultant rates of $60 to $150 per hour, that translates to $1,200 to $12,000 in migration expense. Workflows built on proprietary features, custom logic, and platform-specific integrations are harder to move. We've seen businesses delay migration for a year because the switching cost felt prohibitive, while paying $400 per month more than necessary. The annual cost of delay exceeded the migration cost by month three.

The common thread across all three mistakes is the same. The decision was made on what felt easy today instead of what would cost less over the next twelve months. Automation platforms are infrastructure, and infrastructure decisions compound.
The decision in one table
| Your situation | Platform | Starting cost | Pricing model | Best for |
|---|---|---|---|---|
| Non-technical, low volume, broad integrations needed | Zapier | $19.99/mo (750 tasks) | Per task | Quick setup, simple workflows |
| Growing team, moderate volume, visual workflow builder | Make | $10.59/mo (10,000 ops) | Per operation (credits) | Cost-efficient scaling |
| Technical team, high volume, AI workflows, self-host option | n8n | $0 self-hosted, 24 EUR/mo cloud | Per execution | Maximum flexibility and ownership |
| Microsoft shop, desktop + cloud automation | Power Automate | $15/user/mo | Per user or per flow | Microsoft ecosystem integration |
| Open source, simple cloud pricing, small team | Activepieces | $0 self-hosted, $99/mo cloud | Flat rate (50K tasks) | Predictable cost, open source |
| Developer, code-first, lightweight workflows | Pipedream | $0 free, $49/mo advanced | Per compute time | Code-native automation |
Every row in that table is the right answer for a specific situation, and no row is the right answer for every situation. That is the entire point. The right platform exists for every business. The right platform for every business does not.
The real cost of getting it wrong
We track a pattern that repeats across industries and business sizes. A team picks an automation platform without asking the five questions. They build 10, then 20, then 50 workflows. The workflows accumulate steps, the steps accumulate volume, and the bill crosses a threshold that forces a conversation nobody wanted to have. By the time that conversation happens, the migration cost is 10x what it would have been on day one.
The five questions take twenty minutes. The wrong platform costs thousands per year and dozens of hours to escape. We've watched this cycle play out enough times to know the pattern is preventable. Ask the questions first. Match your answers to the table. Pick the platform that fits your volume, your team, your growth rate, your integration requirements, and your compliance constraints. Not the platform with the biggest brand, the best marketing, or the most familiar name.
The businesses that get automation right are not the ones who pick the best platform. They pick the right one.
Pricing verified March 2026 from official platform pricing pages: zapier.com/pricing, make.com/en/pricing, n8n.io/pricing, microsoft.com/power-platform, activepieces.com, pipedream.com/pricing. Market size data from Mordor Intelligence and Coherent Market Insights. Migration cost estimates from Upwork and Digital Agency Network 2026 reports.
Crux helps businesses find the right automation platform for their specific problem. We don't sell automation tools. We help you pick the right one.
Related Posts
How to migrate from Zapier to Make or n8n
I migrated 47 Zaps last year. Took me three weeks, about 40 hours of actual work, and I broke two production workflows along the way. One was a Stripe webhook that silently stopped firing for 36 ho...
How to automate invoice processing without code
I spent two years manually processing invoices for a consulting business. Every Friday afternoon looked the same: download PDFs from email, retype line items into QuickBooks, cross-check amounts ag...
How to self-host n8n when you are not an engineer
I self-hosted n8n on an $8 VPS last month. The whole thing took about 45 minutes, and I'm not a sysadmin or someone who runs servers for a living. I just got tired of watching my n8n Cloud bill gro...