01The problem it solves
Automating a process that is not defined only makes it faster. If nobody knows what happens with an exception, the flow will repeat it as many times as it runs, and the error will arrive sooner.
n8n is an orchestration tool, not a backend. It is not there to hold up a product with its own rules, to replace a well-built form, or to process live data without a legal basis and permissions.
There is also an operating cost: someone has to review runs, update credentials, handle failures and decide when the flow is switched off. That owner is named before, not after.
If the process exists only to avoid deciding how it should work, n8n does not fix it: it speeds it up. First define the process and its owner; then automate what already makes sense.
02When it adds value and when it doesn't
Can add value
- The process repeats at a known frequency and is done by hand today.
- The systems involved expose stable APIs or integration mechanisms.
- Volume and manual cost have been measured and justify implementation.
- A human owner is responsible for the flow and for what must happen when it fails.
- Tools that do not share data natively have to be connected.
- The process can handle retries, alerts and a switch-off plan.
Doesn't add value
- The goal is to replace the website form: a well-designed form and human review solve it better.
- It is a one-off send or a task that does not repeat.
- The plan is to automate with no owner, no monitoring and no switch-off criteria.
- Live data has to be processed without a legal basis, permissions and security measures.
- It is used to avoid defining the process or making a decision about it.
- A product backend is needed: n8n orchestrates tools, it does not replace a custom API.
The four decisions
-
01
Do not automate
If the task happens once, does not repeat or depends on human judgement nobody has defined, it is not automated. A better form or a manual review can solve it with less cost and less risk.
-
02
Document first, automate later
If the process repeats but is not written down, it is documented first with the person who runs it. Automating without that step moves the improvisation into the flow; with the process clear, implementation can be assessed.
-
03
n8n in the client's environment
If the client wants to hold the data and credentials, n8n is installed in their environment and they take on operations. Access to build and test is agreed and removed at the end.
-
04
n8n managed by skserviciosweb
Only if expressly agreed. It means someone operates the instance, reviews runs and responds to failures; it is not an implicit service and is not assumed in the proposal.
03What we do
n8n isn't the service: the service is what gets built with it. These are the usual pieces, each one with its scope and its cost.
-
01
Process map: steps, inputs, outputs and exceptions.
Go to Process automation -
02
Flow design with triggers, conditions and limits.
Go to Process automation -
03
External credentials, outside the repository and documentation.
Go to Process automation -
04
An environment separate from production.
Go to Process automation -
05
Error handling, retries and idempotency.
Go to Process automation -
06
Tests with synthetic data, including failure cases.
Go to Process automation -
07
Documented rollback and criteria to switch the flow off.
Go to Process automation -
08
Monitoring and alerts.
Go to Process automation -
09
Operations runbook and handover in the agreed environment.
Go to Process automation
04Performance and maintenance
Can be measured
- Volume of runs and data the flow moves.
- Real frequency of the process and whether the trigger keeps up with that cadence.
- Errors per run and how many are caught before reaching the recipient.
- Retries, timeouts and what happens when an external system does not respond.
Can't be promised
- A time or cost saving without having measured the current process.
- That automation removes errors: it reduces some, but adds its own failures to watch.
- An availability or response time for external systems I do not control.
Maintenance
Review runs and errors, update credentials, maintain the flows and the n8n version, and respect the licence. Someone also has to decide who operates the instance and what happens when the flow is switched off. The cost is agreed with the project.
Price and timeline
Quoted after diagnosis. It depends on systems, exceptions, tests, operation and the delivery model. There is no per-process rate or fixed package.
See the speed and WPO service05Proof: stated capability
Declared capability. There are no publishable cases or verified metrics for n8n automations on this project, and evidence per technology is still pending.
What I can explain is the reasoning: how the process is mapped, how failures are handled and what is documented so the flow has an owner.
This page describes when n8n fits and when it does not; it does not claim specific experience or promise savings that have not been measured.
See the projects06Frequently asked questions
What can be automated?
Repetitive tasks between systems with accessible interfaces: moving data, creating records, sending notifications or syncing states. Anything that depends on undefined human judgement, happens once or has no clear owner is left out.
Is n8n required?
Not always. n8n makes sense for orchestrating a process across tools when its operating cost is accepted. If a direct integration, a custom API or a better form is enough, that is considered first.
What happens if it fails?
It is defined before building: expected errors, retries, who receives the alert and how it is switched off. A flow without a failure plan is not delivered.
Who holds the credentials?
The client, in their own environment. They are configured outside the repository and do not travel in the documentation or the code. Access to build and test is agreed and removed at the end.
Does it connect to WordPress?
Yes, through its API or webhooks, when the process requires it. n8n is not installed on the corporate website and does not replace its form; the website keeps working without it.
How much does it cost?
Quoted after diagnosis. It depends on systems, exceptions, tests, operation and the delivery model, which is decided case by case. There is no per-process rate or fixed package.