Will it be replayed?
A test run twice a year will never pay for its writing or maintenance. A regression test run with every release will.
Test automation
Automation doesn't replace the recipe: it frees up brain time by taking over what is repetitive, stable, and verifiable without judgment. The rest remains human, and that is good news.
Make an appointmentSorting
A test run twice a year will never pay for its writing or maintenance. A regression test run with every release will.
«Is the displayed amount €12.90?» is automatable. «Is the screen clear?» is not automatable, and should not be.
Automating a screen that changes every week means funding permanent maintenance. We are waiting for it to settle down.
Once this sorting is done, there generally remains 60 to 70 % cases in a non-regression campaign — and that's exactly the part that exhausts the teams and that gets sacrificed first when deadlines tighten.
What stands the test of time
Most automation projects do not die from a technical defect, but from their maintenance cost.
Three choices determine the lifespan of an automated base. The anchoring of the elements stable identifiers placed in the application rather than fragile paths that break at the first layout change. The separation between the storyline and the technique — the Page Object model — so that a screen change can be fixed in a single place. Data mastery : an identically reconstructed test set, otherwise failures become random and no one reads the reports anymore.
The real health indicator is not the number of automated tests, it is the time spent fixing them each week. We measure this from the first month.
The tools
Selenium, Cypress, Playwright — depending on what your teams already know how to maintain, not according to our preferences.
Appium for applications, and verification via application interface when it is more reliable than the screen.
Heavy client, kiosk, industrial screen: when no tool works, we write the missing PLC.
What changes
What these figures actually make possible: delivering more often without reducing coverage, check in full what was previously only checked by sampling, and give testers back the time to hunt for the bugs that no script will ever find.
Frank questions
Through the critical user journeys played out in each version: login, ordering, payment. About ten well-chosen cases are already useful and serve as a foundation. Starting out wanting to cover everything is the surest way to deliver nothing.
Someone needs to be able to fix a test that breaks. It can be a trained tester, not necessarily a developer. We train your teams during the project rather than after, and we document in French.
The first scripts take two to three weeks to run. The return on investment shows by the second or third non-regression campaign: that is when the time saved exceeds the time invested. Before that, it is an investment, and it must be stated as such.
She writes a first draft of test cases, maintains selectors, triages failures, and performs bulk comparisons. She does not decide what is worth testing. We have been running her in delivery for two years: what has changed is the volume a consultant covers in a day, not the nature of the job.
A no-obligation audit of your testing process to find out where you really stand.
Make an appointment
English 
French 
Hebrew