How to plan a four-week automation pilot
AurIA42 · Published 2026-08-10 · Updated 2026-08-10 · 9 min
A four-week pilot automates a single real process, with a closed scope and measurable success criteria agreed before starting. The first week is spent analysing and deciding; the next two building and integrating; the last testing with real data, fixing issues and deciding whether to scale up or stop.
What a pilot is (and isn't)
A pilot is a real, limited and measurable implementation of an automation applied to a single process. It's not a demo, nor a theoretical proof of concept: by the end of the pilot the process runs on the company's real data and someone on the team uses it in their day-to-day work.
The limited scope is deliberate. Four weeks won't reorganise a company: they solve one specific piece, measure what has changed, and let you decide with data whether it's worth rolling out further.
- —Yes: one process, one team, one success criterion, real data.
- —Yes: documentation of what was built and how to maintain it.
- —No: replacing an ERP, migrating an entire system, or automating five departments at once.
- —No: promising a percentage saving before the starting point has been measured.
How to choose the right process
The main reason pilots fail is choosing the wrong process. A good candidate is repetitive, frequent, stable enough to describe with clear rules, and painful enough that someone notices when it's gone.
- —It happens at least weekly and consumes identifiable hours.
- —It has a clear input and output (an email, a file, a form, a record).
- —Exceptions are known and don't account for most cases.
- —The data needed is accessible without prior infrastructure projects.
- —A responsible person is available for the four weeks.
Defining scope and success criteria
Before writing a single line of code, it's essential to put in writing what's in the pilot, what's out, and how success will be judged. Criteria must be measurable using information the company already has, or that can be gathered during the initial diagnostic.
Measuring the starting point matters as much as measuring the result. Without a baseline there's no possible comparison, and any final figure is just an opinion.
- —Time spent on the process before and after, in hours/week.
- —Volume processed without manual intervention relative to the total.
- —Number of errors or subsequent corrections detected.
- —Response time to the customer or department receiving the output.
Week-by-week timeline
The timeline below is what we apply by default. It can be adjusted, but the logic stays the same: understand first, then build, then test against reality and decide.
| Week | Objective | Deliverable | Client involvement |
|---|---|---|---|
| 1 | Analyse the real process, access and data; set scope and success criteria | Scope document, measured baseline and solution design | 3–5 hours (interviews and access) |
| 2 | Build the core automation and the main integrations | First working version with sample data | 1–2 hours (criteria validation) |
| 3 | Cover exceptions, alerts, error logging and human oversight | Candidate version with traceability and error handling | 2–3 hours (review of real cases) |
| 4 | Test with real data, fix issues, document and train the team | Pilot live, documentation and result measurement | 3–4 hours (testing and training) |
Roles and expected involvement
A pilot doesn't work if the company only shows up to the kick-off meeting. Total client involvement is usually between 9 and 14 hours spread over four weeks, concentrated at the start and the end.
- —Process owner: knows the real exceptions and validates the result.
- —Decision-maker: approves scope and budget without long delays.
- —Technical contact or external supplier: grants access to systems, email or databases.
- —AurIA42: analysis, development, integration, documentation and training.
Risks that make a pilot fail
The problem is almost never technical. These are the risks to watch from day one, along with the preventive measure for each.
| Risk | Warning sign | Preventive measure |
|---|---|---|
| Scope creeps during the project | Every meeting adds a new case to the pilot | Closed scope list and a backlog of ideas for later phases |
| Process isn't stable | Every person does it a different way | Standardise and agree the process before automating it |
| Access doesn't arrive on time | Week 2 with no credentials or real data | Set access as a week-1 requirement |
| Poor data quality | Duplicates, blank fields, inconsistent formats | Automatic validation and an exceptions report |
| Nobody uses the result | The team keeps using the old spreadsheet | Training in week 4 and a named owner |
Indicative cost and what's included
The indicative range for a functional four-week pilot is €4,000 to €12,000. The exact figure depends on the number of systems to integrate, whether a custom interface is needed, the volume of exceptions, and security or traceability requirements.
The initial two-hour diagnostic is free and exists precisely to place the project within that range before committing to anything. Simpler one-off automations can start from €600, and ongoing maintenance from €79/month.
- —Included: analysis, development, integrations, testing with real data, documentation and training.
- —Not included: third-party licences, API or AI model usage costs, or hardware.
- —Not included: reorganising processes outside the agreed scope.
What happens after the pilot
By the end of week four there are three legitimate outcomes: scale up, keep as is, or stop. Stopping isn't a failure if the pilot has shown with data that automating the process isn't worthwhile; that's exactly the value of keeping it small and short.
If it scales up, the usual next step is extending the automation to neighbouring processes or higher volumes, with maintenance covering system changes, incidents and ongoing evolution.
Frequently asked questions
Is four weeks enough to see results?
It is for a specific, well-defined process. It isn't for automating several areas at once. If the scope doesn't fit in four weeks, we say so before starting and propose splitting it up.
What happens if the pilot doesn't meet the success criteria?
We analyse why: a poorly chosen process, insufficient data, or unforeseen exceptions. With that information we decide whether to adjust the scope or stop. The decision is based on the measurement taken, not on impressions.
Do we need to change the software we already use?
Generally not. We work with existing systems and connect the automation to them. We only suggest changing a tool when the current one technically blocks the process.
Who maintains the automation afterwards?
It can be maintained internally with the documentation we hand over, or you can contract maintenance and evolution with us, from €79/month depending on criticality and volume.
Can everything be done remotely?
Yes. We work remotely with SMEs and freelancers, and hold in-person sessions when the process requires it.
Related service
Operational Efficiency Consultancy in BarcelonaBefore we propose any automation, we map your processes, measure where the time actually goes and prioritise by impact and effort.
How to automate an SME in Barcelona: 7-step guide
How much does SME process automation cost?
Want to plan your pilot?
In the free two-hour diagnostic we review the candidate process, the baseline, and whether it fits into four weeks.