Automated website testing that safeguards NZZ's digital revenue

Introduction

Neue Zürcher Zeitung (NZZ) is one of Switzerland’s leading premium publishers, reaching a global readership through The NZZ Website, The E-Paper, Bellevue, and The Market.

For a digital publisher of NZZ’s scale, the paywall is the engine of the business. Any misfire, a regional edition that won’t load, a browser-specific break, a premium article served for free, is silent revenue lost. Across browsers, devices, and geographies, the test matrix is challenging to cover.

---

The Challenge

NZZ’s paywall isn’t a simple on/off switch. It adapts to geography, device, and user state and coexists with dozens of other moving parts: share buttons, search, comments, dark mode, newsletters, ad delivery.

Before Klarent, NZZ ran a layer of Playwright and Next.js scripts. As products evolved, those scripts demanded constant rewrites and most regression testing was still manual and on-request. A paywall that silently broke in a specific region, on a specific browser, on a specific article type was effectively invisible until a reader noticed.

Beyond reliability risk, the manual workload was a release bottleneck. The QA team worked through test checklists line by line, ticking off entries manually leaving little bandwidth for the deep, exploratory testing that catches critical issues before they reach production.

---

The Solution

From manual spot-checks to a continuous QA heartbeat

NZZ deployed Klarent as an autonomous QA agent that interacts with its apps exactly like a real reader: clicking through the paywall, accepting cookies, scrolling articles, completing newsletter sign-ups — and reports back on what works and what doesn’t.

High-frequency, scheduled testing. Tests run on an hourly schedule, and every 15–30 minutes on revenue-path flows and user experience features to ensure their uptime and availability.

Klarent allows us to test multiple devices across multiple browsers with scheduled instances. We can test different geo-specific paywalls and schedule tests every 15/30 minutes, so we detect issues in a moment’s notice.

Geo-specific paywall verification. Klarent simulates users across global markets, ensuring regional paywall rules trigger as intended.

Full cross-browser, cross-device coverage. Every test runs across Chrome, Firefox, Edge, and Safari — in portrait and landscape on mobile and tablet.

Product- and business-authored tests. Because Klarent’s agent is prompted in natural language, more than half of NZZ’s end-to-end tests are now written directly by Product and Business teams, not just QA engineers.

The result. Automated tests covering rendering, geo-redirection, search, article loading, social sharing, paywall triggers, login and newsletter flows, dark mode, and ad delivery — across The NZZ Website, The E-Paper, and The Market.

AI-powered self-healing keeps the suite stable. Klarent’s self-healing capabilities automatically adapt to UI changes, so tests don’t break every time the interface evolves. Once the smoke test suite is configured, the team doesn’t need to constantly rebuild it as the product grows.

---

The Impact

From running blind to real-time resilience

The shift shows up across three dimensions: speed, revenue protection, and how QA spends its time.

Smoke testing that was previously done manually now runs 75% faster. Deployments happen more rapidly and the overall quality of the end product has improved.

We shifted our perspective on how we do QA. Autonomous testing is a tool for making QA lives easier. QAs spend more time investigating and reporting on bugs, which leads to better performance and stability.

Engineers who used to click through smoke tests one browser at a time now focus on exploratory testing and deep bug investigation. The optimized framework also makes onboarding new QA engineers significantly faster.

---

What’s Next

Driven by the ROI on web, NZZ is now expanding Klarent into their mobile app development pipeline, bringing the same automation benefits to their mobile QA team. Success continues to be measured against two strategic metrics: maximizing automated test coverage and driving team productivity across the board.

The tool is quite intuitive and we can learn on the fly. The Klarent team has always gone the extra mile: early access to features, listening to our feedback and quick response time when we need help.