I built the tools that became Siteimp over many years of running a Dev Ops focused consultancy. And over two months in 2021 and 2022, I converted all these smaller tools into one big tool. Siteimp looked at entire websites, hit them with a lot of traffic that collected a lot of different types of data and even added full Google Lighthouse reports per page. The reports were a great way to look at entire websites, diagnose serious problems and slowly prioritize fixes based on overall value.
But in 2021, web performance tooling was nowhere near what it is now. Google had just recently begun using performance as a ranking metric and it took time for the industry to catch up. But it caught up in a big way. Full Siteimp reports became less useful as tooling adapted.
But performance is still important and Siteimp had some features that still helped it shine. The most important was that I could write synthetic tests and collect performance data over real world test cases.
But note the word 'I'. One of Siteimp's main issues was that it was very powerful. I could keep it under control because I built it, but when it was deployed in certain ways it was remarkably close to a denial of service tool. Any bugs in that synthetic process could flood a site with thousands of hits a minute.
And so my use of Siteimp and my clients' use of Siteimp began to diverge. I began to use it mostly for contextual performance testing - I wanted to see what happened to performance when users interacted with pages and I started to run hundreds of tests a day.
Over the years, I started many versions that could be released as a software as a service based product. But it never quite fit. And then I got absolutely fed up with subscriptions.
My subscription related freak out started when I realized how much money I spent to watch sports a year. I'm not even a huge sports fan, but I like watching soccer and baseball. Each was a subscription. Add in three streaming services, Apple Music, my email accounts and various services...and I was spending a metric shit tonne of money every month.
And meanwhile I was trying to figure out how to add in another subscription? So....I started a different way and started building Siteimp as a desktop application. It's roughly the same Siteimp but you can run it from any Windows machine in version 1 and somewhere around 1.2 it will support Linux and Apple. *It would support Linux at v1 but it's really really hard to monetize Linux software and my nine year old can't eat open source projects.*
It also supports different applications of the testing methodology - in this case, it works with web applications and both Electron and Tauri applications in naive and learning modes. Naive mode just beats the heck out of an application. Learning mode is based on a structured log utility and lets Siteimp users play back user sessions.
Funny story - the first implementation of learning mode could only play back user sessions according to exact timestamps. I was using it to test how quickly a user could register and how that registration function operated under load for a web application that I was working on. My Dad called mid test and we talked for about half an hour, then I had lunch and went for a walk. When I got back, I finished logging in. So, that's the story of the time that it took 3.5 hours for a user to enter three fields and register for an application. Siteimp's usability feedback was pretty scathing.
I'm working on Siteimp in tandem with other things now and it's not my main task. And with how it's deployed now, I don't have a lot of space for mistakes so I have no idea when it will be released. It's scheduled to be released when it's ready. But I can't wait to show all of you another way to test. And only charge you once while I'm doing it.