Ranking what to automate
Two questions per flow: how many manual checks a month does it get, and what does a failure cost? Dealer sign-in, special prices, order confirmation and invoice creation tend to top the list.
Consider a Savinja Valley furniture maker that exports through a B2B portal used by dealers in Italy, Croatia and Hungary. Every dealer has an individual discount, payment period and delivery arrangement. Hardly a week passes without a plugin or library update. One unremarkable update left dealers looking at full catalogue prices for three days, and the firm learned of it only from an irritated customer writing to the head of sales. An automated check would have flagged it within minutes. The script signs in as a particular dealer, puts an item in the cart, compares the total with the discounted figure it expects and halts the release if the two disagree. For us automation is not a box to tick once. It is code, and like any code it needs care. So we do not chase full coverage; we start with whatever your people re-check manually most often and whatever causes the worst damage when it fails.
We want a suite your developers believe, so a red result gets attention instead of a shrug.
Two questions per flow: how many manual checks a month does it get, and what does a failure cost? Dealer sign-in, special prices, order confirmation and invoice creation tend to top the list.
Unit and API checks carry most of the logic; only the journeys that really matter run through the browser. The suite completes in minutes and does not collapse whenever the design is refreshed.
Playwright in browsers, Appium on phones. Selectors are chosen to survive layout tweaks, and the code is kept simple enough for your developers to follow.
Quotes, delivery notes and invoice PDFs sent to customers are diffed against an approved version. A logo that has moved or a table running off the page is spotted straight away.
Scripts create dummy customers, items and price lists and remove them when finished. Nothing is copied from production, which also keeps you on the right side of GDPR.
Checks run in parallel on GitHub Actions, GitLab CI, Azure Pipelines or Jenkins. When one fails, its recording, screenshot and log appear in the team's Teams or Slack channel.
A check that flips between pass and fail is quarantined so it cannot hold up the main pipeline while we find out why. An untrusted check is worse than none.
We begin modestly, show results, then extend. Expect the first automated journeys in your pipeline within weeks.
Current tests, CI configuration, how often you ship and a ranked list of flows.
The top three automated, then a week of evidence that they run reliably.
Further flows in ranking order, with a brief check-in alongside your team every two weeks.
Either we maintain the suite or we train your developers and pair with them until they own it.
Automated checks free your testers; they are not a substitute for them. Scripts confirm that familiar paths still work. Only a person will notice that a new feature is baffling, that a label reads two ways or that users have found a route no one designed. Give the repetitive work to the pipeline and point human attention at exploration.
We sort the useful from the useless. Reliable checks can stay put; ones that fail constantly or that no one reads are rebuilt in Playwright or dropped. Migrating the lot unexamined only relocates old headaches.
We investigate immediately. If the reason is not obvious, the check is quarantined outside the main pipeline and tracked on its own. When developers start ignoring red, the whole suite stops being worth much.
Yes, using Appium or Flutter's built-in tools. Mobile checks run slower and break more easily than browser ones, so we keep the number low and aim them at the journeys that matter most.
The language your team already works in, usually TypeScript. Everything sits in your repository with documentation, and we run a few pairing sessions at handover.
Initial build is billed at €50 per hour plus VAT, based on an estimate agreed beforehand. Ongoing upkeep can be hourly or covered by a Start, Business or Premium plan.
Which checks does your team repeat by hand before each release, and for how long? We will suggest where to begin with the first three flows.
Enquiry received
An engineer will write back no later than the following working day. Faults that leave your staff unable to work are always looked at first.
No town by that name here. Check the letters, or settle for the nearest larger place. Our engineers work only over the internet, so Kočevje and Koper get precisely the same treatment.