There are no second chances when it comes to live events: Mentimeter's solutions have to work perfectly every time. Their audience engagement features must be accessible via mobile devices without user sign-in. They must also be fast. And they need to scale effortlessly to cope with huge spikes in demand; a single event can drive connections from zero to 70,000+ participants in a matter of seconds.
Mentimeter's engineers designed their systems to cope with those exacting demands, but as the business experienced rapid growth, its realtime infrastructure provider struggled to keep pace. The platform's performance at scale started to suffer.
Eventually, a tipping point came when a spike of a relatively small number of concurrent connections – ~35,000 – caused part of their realtime system provider's network to crash. Luckily, Mentimeter had fallback solutions in place for realtime communication so their services continued to operate, but with a degraded user experience for several hours. According to CTO Johan Bengtsson: "An outage like that affects the entire customer experience because the core of efficient audience engagement requires reliable and fast feedback from the audience. A degraded service in this area shakes customer confidence in our service. The provider was unwilling to guarantee to fix these scaling and reliability issues, nor could it provide dependable SLAs. We knew it was time to move on."