n8n is a workflow automation platform that connects apps and APIs in a visual editor, and it can run on n8n's hosted cloud or on a server you manage yourself. For recruiters it fits best where information moves between systems when something happens, such as a new application, a status change or a weekly report. It fits less well where people need a polished screen to work in all day. The decisions that matter most are where it runs, who maintains it and how candidate data is handled. This guide explains what n8n is, how its pricing and licence work (checked against n8n's own pages in October 2026), which recruiting workflows suit it, where it breaks, and how to build a first workflow safely.
What n8n is in plain terms
A workflow in n8n is a chain of steps, called nodes, that starts with a trigger and then reads, changes or sends data in other tools. You build it on a canvas, and you can reach almost any tool that has an API.
| Piece | What it does | Recruiting example |
|---|---|---|
| Trigger | Starts the workflow when something happens | A candidate is created or moved to a new stage. The Workable Trigger node offers the events Candidate Created and Candidate Moved |
| Webhook | Gives you a URL that another tool can send data to | Your application form posts each new applicant |
| App node | Reads or writes in a specific tool | Add a row to a tracker, send an email, update a CRM field |
| HTTP Request | Calls any REST API, including tools with no dedicated node | Create a candidate in an ATS that has an API but no n8n node |
| Error workflow | A separate workflow that runs when another one fails | Alert the owner and log what failed |
Two details are worth knowing early. The HTTP Request node supports several authentication types (Basic, Custom, Digest, Header, OAuth1 and OAuth2 among them) and has pagination options for long result lists, which matters for exports from an ATS. And n8n counts usage in executions: its pricing page defines an execution as a single run of the entire workflow, regardless of how many steps it has or how much data it processes.
Pricing and licence, as of October 2026
Prices change, so confirm them on the pricing page before you decide. At the time of writing it lists:
| Option | Where it runs | Listed price | Included executions per month |
|---|---|---|---|
| Starter | Hosted by n8n | 20 EUR a month, billed annually | 2,500 |
| Pro | Hosted by n8n | 50 EUR a month, billed annually | 10,000 |
| Business | Self-hosted | 667 EUR a month, billed annually | 40,000 |
| Enterprise | Hosted or self-hosted | Custom | Custom |
| Community Edition | Self-hosted, code on GitHub | No fee listed, but you pay for the server and the upkeep | Limited by your own server |
The page also states that all plans include unlimited users and workflows.
Because every run counts, the trigger you choose affects cost. A workflow that runs once per new candidate uses 300 executions in a month with 300 candidates, however many steps it has. A workflow that checks for changes every five minutes runs 12 times an hour, 288 times a day and 8,640 times in a 30-day month, which is more than three times the Starter allowance. That is simple arithmetic from n8n's definition of an execution, so check how your plan treats scheduled runs that find nothing new. Where the source tool can send a webhook, that is usually the cheaper and faster route than polling.
On licensing, n8n publishes its code on GitHub under the Sustainable Use License. The licence allows you to use the software for your own internal business purposes or for non-commercial or personal use, and it limits distributing it or offering it to others. n8n's licence FAQ says that building, running and maintaining workflows on your own n8n instance on behalf of clients is permitted, provided the clients do not access that instance directly. It also says that hosting n8n as a service so that clients can build their own workflows is not allowed. These terms matter if you plan to resell automation or run workflows for clients, so read the licence and the FAQ, and contact n8n if your case is unusual. This is general information, not legal advice.
Which recruiting workflows fit n8n
A useful test is whether the work is data moving between tools when something happens. If it is, n8n is a strong candidate. If the work needs a screen that people use all day, it is not.
| Workflow | Typical trigger | Fit | Watch for |
|---|---|---|---|
| New application to tracker, with a recruiter alert | Webhook or ATS trigger | Strong | Duplicate applicants, and consent for the data you store |
| Stage change to a client or candidate update | ATS status change | Strong | Wrong stage mapping, and messages sent too early |
| Interview confirmations and reminders | Booking event | Strong | Time zones and double bookings |
| Weekly pipeline report to email or chat | Schedule | Strong | Bad data in, bad report out |
| CV intake and parsing into fields | Email or form | Good | Parser errors, so add checks and a review path |
| Job posting and syncing across boards | Job created or changed | Mixed | Each board's API and terms differ, and some have no API |
| Outreach sequences | Schedule or stage change | Mixed | Deliverability, consent and each platform's terms of use |
| AI-assisted summaries or scoring | New CV or profile | Good, with limits | Human review and a bias audit |
| Client portal or live dashboard | Not event-driven | Weak | A dedicated app is usually the better tool |
These ratings are a judgement based on how each job is shaped, not a benchmark. The workflows that depend on data quality are only as good as the data underneath them, so it helps to run the 30-CV parsing test before automating intake, and to use the weekly shortlist audit before automating anything that ranks candidates.
Connecting your ATS and other tools
There are three routes, and most setups use all of them.
- A dedicated node or trigger. Check n8n's integrations list for your ATS first. Some tools have one (Workable has a trigger node), and others do not.
- A webhook. The Webhook node gives you a URL that another tool can call. It has a test URL, registered while you listen for a test event, and a production URL, registered when you publish the workflow, so it is easy to point the other tool at the wrong one. Authentication options are Basic, Header or JWT, or none, and you can restrict callers by IP address. The documented maximum payload is 16 MB.
- The HTTP Request node. When a tool has an API but no node, call the API directly. n8n also lets you reuse a stored credential for such calls, described in its page on custom API actions, which saves you from setting up authentication again.
Before you commit to a tool, read its API documentation for three things: which fields it exposes, whether it can send webhooks, and its rate limits. An API often exposes fewer fields or actions than the screen does, so confirm that the ones you need exist.
Where recruiting workflows break
Most failures are ordinary, and most can be designed around.
| Failure | What happens | Safeguard |
|---|---|---|
| Silent failure | A step fails and nobody notices for days | Attach an error workflow. It starts with the Error Trigger node, which receives details such as the workflow name, the error message and a link to the failed execution. The Stop and Error node lets you fail on purpose when data is bad |
| Duplicate processing | The same candidate is created twice after a retry or a re-send | Look the candidate up by email before creating, and keep one unique key |
| Expired credentials | A connection lapses and every run fails | Alert on failures, and review connections on a schedule |
| Changed fields | A vendor renames a field and data lands in the wrong place | Validate required fields at the start and log what was received |
| Rate limits | A bulk import hits the tool's limit | Process in batches and respect the documented limits |
| Test and live mix-ups | Real data goes to a test sheet, or the other way round | Use separate test credentials and clear naming |
| Candidate data in logs | Execution history keeps personal data longer than needed | Decide how long history is kept and who can open it |
Hosting, security and candidate data
The main choice is between n8n's hosted cloud and running it yourself.
| n8n Cloud | Self-hosted | |
|---|---|---|
| Who runs it | n8n | You or your provider |
| Updates, backups and uptime | n8n | You |
| n8n's GDPR role | n8n describes itself as both controller and processor, and a data processing agreement is part of its terms for Cloud customers | n8n says it is neither controller nor processor, because it does not manage your data |
| Where data sits | Check n8n's sub-processor list and terms | On your server, in the region you choose |
| Cost shape | Subscription based on executions | Server, time and maintenance, plus a fee on the Business plan |
These roles come from n8n's privacy and security documentation, which also notes that self-hosted instances send pseudonymous telemetry that can be turned off with environment variables, and that n8n says it does not collect workflow data or credential contents.
If you self-host, three practical points apply. First, n8n encrypts credentials before saving them, using a key it generates on first launch or one you set with the N8N_ENCRYPTION_KEY variable. Second, the backup guidance says a full backup includes the .n8n folder (which holds that key) and your PostgreSQL database if you use one, and that without the key a restored database or credential export cannot be decrypted. Third, if you outgrow a single instance, queue mode splits the work between a main instance and workers, and it needs Redis and a database such as PostgreSQL. A distributed setup over SQLite is not supported.
Candidate CVs and notes are personal data, so a workflow that handles them is part of your data protection setup, and so is every tool it connects, including email, spreadsheets and any AI provider. A provider that processes data on your behalf is usually a processor, and Article 28 of the GDPR requires a contract that binds it to act only on your documented instructions. This is general information, not legal advice, and it is worth confirming your setup with a qualified adviser.
Build a first workflow in an afternoon
Pick something frequent, dull and low risk. A good example is a new application that creates a row in a tracker and alerts the recruiter.
- Write it down first. Note the trigger, the inputs, the outputs and who owns the workflow, before you open the editor.
- Build on test data only. Use the test URL while you build, and switch to the production URL when you publish.
- Validate at the start. Check required fields, the email format and whether the candidate already exists.
- Attach an error workflow before going live. A failure should reach a person within minutes.
- Run ten realistic cases. Include messy ones: a missing email, a duplicate, a very long text field and a file in an unexpected format.
- Go live and watch for a week. Review the executions daily and compare the time spent with the manual process.
- Document it. Record what it does, who owns it, where the credentials live and how to switch it off.
The same approach works outside recruiting
A trigger, some checks, an action and an error path is the shape of almost every business automation: client onboarding, invoice intake, lead routing, support triage and reporting. The details change, but the safeguards carry over to all of them. Validate early, avoid duplicates, alert on failure and limit who can see the data. If you can run a reliable application workflow, you can run most of the others.
Common mistakes
- Choosing the tool before the workflow. Define the job and its failure cases first, then see whether n8n fits.
- Polling when a webhook is available. It burns executions and adds delay.
- Going live with no error alert. A failing workflow that nobody sees is worse than no automation.
- Using real candidate data in tests and screenshots. Use made-up records.
- Self-hosting without tested backups or the encryption key. A backup you cannot decrypt is not a backup.
- Treating the licence as an afterthought. It matters if you resell or run workflows for clients.
- Automating a broken manual process. Fix the process first, or you will repeat the mistakes faster.
Next step
Pick one workflow from the table, write down its trigger, checks and owner, and build it on test data this week. If you would like an outside view of an automation or integration setup in your business, whether in recruitment, sales or operations, you can book a short call. For more guides, see twelve processes you can automate in a recruitment agency, what to automate first and what to keep human, how to set up interview scheduling automation, and the rest of the blog.
Sources
- n8n, Plans and Pricing
- n8n Docs, Sustainable Use License
- n8n Docs, License FAQ
- n8n Docs, Webhook node
- n8n Docs, HTTP Request node
- n8n Docs, Custom API actions for existing nodes
- n8n Docs, Workable Trigger
- n8n Docs, Handle errors gracefully
- n8n Docs, Enable queue mode
- n8n Docs, Set a custom encryption key
- n8n Docs, Back up and restore
- n8n Docs, Privacy and security
- Regulation (EU) 2016/679, Article 28 (Processor)