Revision 1
n8n
@namkyu · Aug 31, 2026, 9:11 AM
Updated n8n pricing, usage guidance, workflows, best practices, business uses, limitations, and alternatives.+19,987−0
1
n8n is a workflow automation platform for connecting applications, APIs, databases, and AI services. It uses visual workflows made from connected nodes, while still allowing more technical users to work with APIs, JavaScript or Python, webhooks, databases, and self-hosted infrastructure.2
3
Its main advantage is flexibility. A simple workflow might copy a website lead into Google Sheets, while a more advanced workflow could validate the lead, enrich company data, update a CRM, call an AI model, and notify a sales team.4
5
# Quick Facts6
7
8
| Item | Details |9
| ---------------------- | ----------------------------------------------------- |10
| Main use | Workflow automation and AI orchestration |11
| Best for | Businesses, technical operators, developers, agencies |12
| Beginner friendly | Moderate |13
| Free option | Self-hosted Community Edition |14
| Starting cloud price | €20/month billed annually |15
| Included Starter usage | 2,500 workflow executions/month |16
| Coding required | No, but useful for advanced workflows |17
| Deployment | n8n Cloud or self-hosted |18
| API | Yes |19
| Integrations | 2,000+ listed integrations and nodes |20
21
22
# Pricing23
24
n8n has an important pricing difference from many automation platforms: paid plans are primarily based on **workflow executions rather than charging for every normal step in a workflow**.25
26
An execution is a complete run of a workflow.27
28
For example:29
30
Webhook → Validate lead → Check CRM → Create contact → Send email → Notify Slack31
32
That can still represent one n8n workflow execution even though several nodes run during it.33
34
Current main plans include:35
36
37
| Plan | Starting Price | Included Executions | Deployment |38
| ----------------- | --------------------------: | --------------------------------: | -------------------- |39
| Community Edition | Free | No n8n Cloud execution allowance | Self-hosted |40
| Starter | €20/month billed annually | 2,500/month | n8n Cloud |41
| Pro | €50/month billed annually | 10,000/month | n8n Cloud |42
| Business | €667/month billed annually | 40,000/month | Self-hosted |43
| Enterprise | Custom | Custom | Cloud or self-hosted |44
45
46
The Starter plan also includes five concurrent executions and one shared project. Pro increases concurrency and adds features such as additional shared projects, workflow history, global variables, and execution search.47
48
Business is currently self-hosted and adds organizational features such as environments, Git-based version control, SSO options, and scaling capabilities.49
50
All paid plans currently include unlimited workflows, users, and workflow steps. Usage is primarily determined by executions.51
52
The free Community Edition can make self-hosting attractive, but it should not be treated as zero-cost automation. You may still need to pay for a server, database, backups, storage, monitoring, domains, and third-party services.53
54
External API costs are also separate.55
56
For an AI workflow, the realistic cost may be:57
58
n8n subscription or hosting + AI API usage + database + other external services59
60
This matters because AI model charges can exceed the n8n cost in AI-heavy workflows.61
62
# What It Does63
64
n8n is mainly used to move data and trigger actions between systems.65
66
A sales workflow might look like:67
68
Website form → Validate lead → CRM → Sales notification69
70
An e-commerce workflow could be:71
72
New order → Check payment → Update inventory → CRM → Fulfillment notification73
74
An AI support workflow might be:75
76
New support message → AI classification → Route request → Draft response → Human review77
78
n8n becomes particularly useful when a workflow needs more than a basic trigger-and-action automation. Conditions, branching, APIs, databases, data transformations, custom code, and AI steps can all be combined in the same workflow.79
80
# Getting Started81
82
For most beginners, **n8n Cloud is the easiest way to start** because it avoids server installation and maintenance.83
84
A practical first project is:85
86
Webhook → Google Sheets87
88
Start by creating an n8n Cloud account and a new workflow.89
90
Add a Webhook trigger, send it some test data, and inspect the fields n8n receives.91
92
Then connect Google Sheets using your Google credentials and map fields from the webhook into spreadsheet columns.93
94
Run the workflow and confirm that the new row appears.95
96
This simple project teaches several important concepts at once:97
98
- triggers99
- incoming data100
- credentials101
- field mapping102
- nodes103
- executions104
- testing105
106
After this works, add a condition or another application.107
108
There is little benefit in starting with an AI agent or a large multi-branch workflow before understanding how normal data moves between nodes.109
110
# How to Use111
112
The basic n8n model is:113
114
Trigger → Data → Processing → Action115
116
A **trigger** starts the workflow. Examples include a webhook, scheduled time, incoming message, new database record, or event from another application.117
118
A **node** performs an operation. One node might read Google Sheets, another call an API, another transform data, and another send a Slack message.119
120
Data flows from one node to another. Mapping lets you take output such as a customer's email address from one node and use it as an input in another.121
122
Conditions and branches allow different actions depending on the data.123
124
For example:125
126
New order → Check amount127
128
- Over $1,000 → Notify sales manager129
- Otherwise → Standard processing130
131
**Webhooks** are useful when another application can notify n8n immediately when an event occurs.132
133
**Schedules** are useful for recurring jobs such as daily reports or database synchronization.134
135
The **HTTP Request** node is especially important once you move beyond beginner workflows. If an application has an API but its native n8n node does not support the operation you need, the HTTP Request node can often call the API directly.136
137
Execution history should be used to inspect successful and failed workflow runs and understand what data entered and left each node.138
139
# Practical Workflows140
141
## Lead Capture and CRM Automation142
143
Website form → Webhook → Validate → Check CRM → Create or update lead → Notify sales144
145
This removes manual CRM entry and gives sales teams faster access to new leads.146
147
Common integrations include HubSpot, Salesforce, Google Sheets, Gmail, and Slack.148
149
The main failure risk is duplicate processing. A form may be submitted twice or the same lead may already exist in the CRM.150
151
Check for an existing contact before creating a new record.152
153
## AI Customer Support Routing154
155
New message → Extract text → AI classification → Validate result → Route team → Log result156
157
An AI model can classify messages into categories such as:158
159
- Billing160
- Technical161
- Cancellation162
- Sales163
- Other164
165
This can reduce manual ticket triage.166
167
AI API usage creates an additional cost, so avoid sending unnecessary conversation history or other irrelevant text to the model.168
169
AI output should also be validated before it triggers important customer-facing actions.170
171
## Automated Business Reporting172
173
Schedule → Fetch sales data → Fetch marketing data → Combine metrics → Calculate KPIs → Send report174
175
This is useful for daily or weekly reporting that would otherwise require manually copying values from several platforms.176
177
Typical destinations include Slack, email, Google Sheets, Notion, or a database.178
179
The main problems are often not workflow logic but inconsistent dates, currencies, API pagination, and time zones.180
181
# Tips & Best Practices182
183
## Prefer Webhooks When Available184
185
Polling repeatedly asks an application whether something has changed.186
187
For example:188
189
Every minute → Check for new order190
191
A webhook instead allows the source application to notify n8n when the order actually arrives.192
193
New order → Webhook → Workflow194
195
Webhooks usually reduce unnecessary requests and executions and can also respond faster.196
197
## Filter Before Expensive Operations198
199
Avoid sending every record to paid APIs or AI models when simple logic can remove irrelevant data first.200
201
Instead of:202
203
1,000 leads → AI analysis → Filter204
205
prefer:206
207
1,000 leads → Basic filter → 80 relevant leads → AI analysis208
209
This can substantially reduce external API and token costs.210
211
## Handle API Pagination212
213
Many APIs return only a limited number of records per request.214
215
A workflow can technically succeed while retrieving only the first page of results.216
217
When processing large datasets, check whether the API uses:218
219
- page numbers220
- cursors221
- next-page URLs222
- page-size limits223
224
Continue until no additional results remain.225
226
## Use Batching for Rate-Limited APIs227
228
Sending hundreds or thousands of requests at once can trigger API rate limits.229
230
Process records in smaller batches where necessary.231
232
For example:233
234
1,000 records → batches of 50 → API → next batch235
236
This is especially useful when the external provider limits requests per second or minute.237
238
## Prevent Duplicate Actions239
240
Some events can be delivered more than once.241
242
This is particularly dangerous for:243
244
- payments245
- orders246
- emails247
- invoices248
- account provisioning249
- refunds250
251
Store a unique event or transaction ID and check whether it has already been processed before performing an irreversible action.252
253
This is a practical form of idempotency.254
255
## Build Error Handling Into Important Workflows256
257
A production workflow should not fail silently.258
259
An error workflow can send information about failures to Slack, email, or an incident log.260
261
Useful information includes:262
263
- workflow name264
- execution ID265
- failed node266
- error message267
- time of failure268
269
Test failure cases such as API timeouts, HTTP 429 rate-limit responses, missing values, malformed data, and unavailable external services.270
271
## Keep Credentials Out of Workflow Logic272
273
Use n8n's credential system instead of placing API keys directly inside ordinary fields or code.274
275
This makes credentials easier to manage and rotate and reduces accidental exposure.276
277
## Treat Time Zones Explicitly278
279
Schedules and reporting workflows frequently produce errors because services use different time zones.280
281
Choose a consistent workflow or business timezone and convert dates deliberately instead of relying on defaults.282
283
# Business Uses284
285
n8n is useful wherever employees repeatedly move information between systems.286
287
**Sales and marketing** teams can automate lead collection, enrichment, CRM updates, outreach preparation, campaign reporting, and notifications.288
289
**Operations teams** can synchronize spreadsheets, databases, internal systems, and SaaS applications without manually copying records.290
291
**Customer-support teams** can classify incoming requests, retrieve customer information, route tickets, and prepare responses.292
293
**E-commerce businesses** can connect orders, payments, inventory, CRM records, fulfillment, and reporting.294
295
**AI automation** can combine language models with company data and external applications. Common examples include document extraction, classification, summarization, research workflows, internal assistants, and AI-supported customer service.296
297
Agencies and consultants can also sell workflow design, implementation, maintenance, API integration, and AI automation services.298
299
n8n's license allows many normal internal business and consulting uses, but businesses planning to sell hosted access to n8n itself or embed n8n as a major part of a commercial product should review the current Sustainable Use License and commercial licensing options first.300
301
# Integrations302
303
n8n currently lists more than 2,000 integrations and nodes.304
305
Common categories include:306
307
- **Communication:** Gmail, Slack, Telegram, Microsoft Teams308
- **CRM and sales:** HubSpot, Salesforce, Pipedrive309
- **Data:** Google Sheets, Airtable, PostgreSQL, MySQL, Supabase310
- **AI:** OpenAI and other model or AI infrastructure providers311
- **E-commerce:** Shopify, WooCommerce312
- **Payments:** Stripe313
- **Development:** GitHub, GitLab, HTTP APIs314
315
You do not need a native node for every application.316
317
Use the native integration when it provides the operation you need.318
319
When it does not, check whether the service exposes a REST or GraphQL API and use n8n's HTTP Request node.320
321
Webhooks are another common way to connect services that need to trigger n8n in real time.322
323
# Advanced Use324
325
n8n becomes substantially more flexible once you understand APIs.326
327
The HTTP Request node can call external APIs, while n8n itself also exposes an API that can be used to manage workflows and executions programmatically.328
329
Custom JavaScript or Python can be added when normal nodes and expressions are not enough.330
331
n8n can also orchestrate AI workflows by connecting models with APIs, databases, tools, memory, vector stores, and business applications.332
333
Self-hosting is available through the Community Edition and paid self-hosted plans. Docker is a common deployment method.334
335
A simple self-hosted setup may use:336
337
Internet → Reverse proxy / HTTPS → n8n → Database338
339
Production systems may additionally need backups, monitoring, persistent storage, workers, queue infrastructure, and security configuration.340
341
Self-hosting makes sense when infrastructure control or deployment flexibility matters. n8n Cloud is usually simpler when managing servers provides no business advantage.342
343
n8n describes itself as **fair-code licensed**, rather than conventional open-source software. Its Sustainable Use License permits many internal and consulting uses but restricts some forms of commercial redistribution or hosting.344
345
# Limitations & Common Mistakes346
347
n8n is more flexible than many simple automation products, but that flexibility creates a steeper learning curve.348
349
Advanced workflows may require an understanding of APIs, HTTP, JSON, authentication, data structures, databases, and sometimes code.350
351
Complex workflows can also become difficult to debug. A single canvas with dozens of unrelated branches is harder to maintain than several focused workflows or reusable sub-workflows.352
353
Self-hosting transfers infrastructure responsibility to you. Updates, security, backups, monitoring, database maintenance, and downtime become your responsibility.354
355
Execution-based pricing does not mean workflow architecture no longer matters. A workflow triggered unnecessarily every minute can still consume a large number of executions.356
357
External providers also impose their own API limits and charges.358
359
Common mistakes include:360
361
- polling when a webhook is available362
- creating duplicate CRM records363
- ignoring duplicate payment or order events364
- forgetting API pagination365
- putting expensive AI calls inside large loops366
- ignoring rate limits367
- hardcoding credentials368
- assuming a successful HTTP response means the business action succeeded369
- testing only the successful path370
371
The reliable workflow is usually the one designed for retries, duplicate events, missing data, and external service failures—not merely the one that works in the first test.372
373
# Alternatives374
375
The closest general-purpose alternatives are Make and Zapier.376
377
378
| Factor | n8n | Make | Zapier |379
| --------------------- | -------------------------------- | ------------------------------------------ | ------------------------------- |380
| Beginner friendliness | Moderate | Moderate | High |381
| Pricing basis | Workflow executions | Credits, generally based on module actions | Tasks, based on work performed |382
| Advanced flexibility | Very high | High | Moderate to high |383
| Self-hosting | Yes | No standard self-hosted edition | No standard self-hosted edition |384
| API-heavy workflows | Strong | Strong | Supported |385
| Custom code | Strong | Supported | Supported |386
| Best fit | Technical and complex automation | Visual multi-step automation | Fast SaaS automation |387
388
389
Make currently offers a Free plan with up to 1,000 credits per month. Its Core plan is listed at $12 per month for 10,000 credits, with pricing depending on the selected credit allowance. Most normal module actions consume credits.390
391
Zapier currently offers a Free plan with 100 tasks per month, while Professional starts at $19.99 per month on annual billing. Its pricing is task-based, and some newer AI or specialized actions may consume multiple tasks.392
393
Choose **n8n** when self-hosting, APIs, complex workflow logic, custom code, or technical flexibility are important.394
395
Choose **Make** when you want a highly visual workflow builder and are comfortable with its credit-based usage model.396
397
Choose **Zapier** when ease of setup and broad SaaS connectivity matter more than maximum technical control.398
399
# FAQ400
401
## Is n8n free?402
403
Yes. The Community Edition can be self-hosted without paying an n8n subscription.404
405
The server, database, storage, backups, external APIs, and your maintenance time may still create costs.406
407
## Does n8n require coding?408
409
1. Many workflows can be created entirely with visual nodes.410
411
Coding becomes more useful for custom data transformations, unusual APIs, and advanced workflow logic.412
413
## Is n8n good for beginners?414
415
Yes, but it is more technical than some simpler automation tools.416
417
Beginners should usually start with n8n Cloud and a small workflow before learning self-hosting or advanced AI automation.418
419
## Can n8n use AI?420
421
Yes. n8n supports AI-related nodes and integrations that can connect models with data, tools, APIs, vector stores, and normal business workflows.422
423
External model usage may create separate API charges.424
425
## Can n8n be self-hosted?426
427
Yes. n8n provides a Community Edition and paid self-hosted options.428
429
Self-hosting provides greater infrastructure control but also makes you responsible for deployment, security, backups, updates, and monitoring.430
431
## Is n8n better than Zapier or Make?432
433
Not universally.434
435
n8n is particularly attractive for technical, API-heavy, complex, or self-hosted automation. Zapier is often easier for straightforward SaaS automation, while Make provides a strong visual approach to complex scenarios.436
437
# Sources438
439
- [n8n Plans and Pricing](https://n8n.io/pricing/)440
- [n8n Cloud Subscription Features](https://support.n8n.io/article/n-8-n-cloud-subscription-features-per-tier)441
- [n8n Pricing Model Update](https://support.n8n.io/article/updated-pricing-model-august-2025)442
- [n8n Documentation](https://docs.n8n.io/)443
- [n8n Community Edition Features](https://docs.n8n.io/hosting/community-edition-features/)444
- [n8n Webhook Documentation](https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.webhook/)445
- [n8n API Authentication](https://docs.n8n.io/api/authentication/)446
- [n8n Looping Documentation](https://docs.n8n.io/flow-logic/looping/)447
- [n8n Error Trigger Documentation](https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.errortrigger/)448
- [n8n Docker Installation](https://docs.n8n.io/hosting/installation/docker/)449
- [n8n Sustainable Use License](https://docs.n8n.io/sustainable-use-license/)450
- [n8n Integrations Directory](https://n8n.io/integrations/)451
- [Make Pricing](https://www.make.com/en/pricing)452
- [Zapier Pricing](https://zapier.com/pricing)