Question 1

What makes n8n different from tools like Zapier or Make?

Accepted Answer

Do not list features — speak on three axes and name the boundary yourself at the end. "Control. n8n can be self-hosted, so the data can stay on our own infrastructure. On projects with compliance or data-residency requirements that is often the deciding argument on its own. Flexibility. A Code node, direct HTTP calls, conditions, loops, sub-workflows — the difference shows once the logic goes beyond simple if-this-then-that. I would add the AI components here too: agents, tools, memory and vector stores live inside the workflow, with no separate system to stand up. Friction. On this axis n8n is not ahead — simpler no-code tools get you started faster and are easier for a non-technical user. n8n assumes an understanding of APIs and JSON. Close with: 'For a one-step integration on a team with no technical person, I would not pick n8n — a simpler tool means less friction there.'" Note: avoid asserting specific prices or features — they change often, and being wrong costs credibility.

Question 2

When would you choose to write code instead of using n8n?

Accepted Answer

This tests objectivity. Give concrete criteria. "I would write code in four cases. 1. When the logic is the product. Where automation is a bridge between two systems, n8n is ideal. But if the logic is the company's core business rule, it needs a versioned, tested, reviewed codebase. n8n orchestrates; it does not become the product. 2. When there is a performance requirement. Latency measured in milliseconds, or very high throughput, is ordinary service work. 3. When the data volume is large. Processing millions of rows is not an orchestration tool's job — there are specialised data platforms for that. 4. When the logic gets very complex. A fifty-node canvas is no longer readable. Then you either split it into sub-workflows or move part of it into its own service and call that from n8n. I would emphasise that last option: it is not an either/or but a hybrid. Complex logic is built as an API and n8n orchestrates it and connects it to everything else. In practice that gives the best result."

Question 3

A company wants to migrate from Zapier to n8n. How would you approach it?

Accepted Answer

This is a project-management question, not a technical one. Answer in phases. "1. Clarify the reason. Is the move about cost, a compliance requirement, or the current tool running out of room? The answer sets the migration's priorities. Without a clear reason, the migration stalls half-finished. 2. Inventory the existing automations. How many are there, which actually run, which have been forgotten? In my experience a good share of such a list is already obsolete — those should be deleted rather than migrated. 3. Prioritise. I would start with one or two of the simplest, highest-value processes. An early win earns the team's trust; starting with the hardest can sink the project in week one. 4. Run in parallel. For critical processes I would run both tools side by side for a while and compare results. That needs care: both can react to the same event and perform the action twice — so during the parallel period I would keep the n8n side read-only and comparison-only. 5. The infrastructure decision. Cloud or self-hosted? If compliance is the reason, the answer is self-hosted anyway. 6. Set conventions up front. Naming, folders, error workflows. A migration is the ideal moment — going back to 80 workflows later costs far more. The most important expectation to manage: n8n is more flexible but demands more responsibility — error handling, updates and monitoring are now your job."

Axis	Where n8n sits	When that is an advantage
Control	Self-hosting is possible; the data can stay on your own infrastructure	When compliance or data residency requires it
Flexibility	The Code node, HTTP Request, conditions, loops, sub-workflows	When the logic goes beyond simple if-this-then-that
Friction	A visual canvas, but it does assume some technical understanding	When the team understands APIs and JSON
AI components	Agents, chains, tools, memory and vector stores — inside the workflow	When AI is part of the automation rather than a separate system
Data engineering	It is an orchestration tool, not an ETL platform	Not the right choice for processing terabytes

Comparison questions: n8n versus other tools

📚 Sources and documentation