ChooseBestStroller is a service, that helps moms to find perfect baby stroller. The problem is that there are thousands of strollers, they are completely different and one that is ideal for one mom is awful for another and vice versa. Because of it for most of mothers and fathers and other relatives it’s almost impossible to find really good stroller. Some of them are lucky to buy something that fits, but other suffer with unconvinient stuff. 
So ChooseBestStroller helps in such situations.
All started from the decision of one b2b company to create a startup inside itself and create something for b2c. It was independent group of managers, researcher, designer (me), analyst. We wanted to research how customers make choice. 
We started 5 days research using Sprint (Jake Knapp) method. We found a lot of interesting insights. A then, making decision what product we should build, we decided to create a service for parents waiting for a (possible) first child and need good advice.
I made fast prototypes to make fast researches with real users, find problems and fix them.
We created a lot of variants of questionnaire, websites and other stuff. Consultants first worked manually, searching for best stroller match for each client. Now we created smart algorythm that searches for best decision, but the possibility of consultation with real persong is still available.
I create simple logo. The ChooseBestStroller website isn’t a masterpiece of design, it’s just a simple template. And a lot of parts were made on templates. And I’m not yet talking about dozens of landings. But despite of it it’s real data driven design — without any stuck with beautiful UI yet, but fully devoted to tests and creating the best UX.
But results of auto-match are crafted with love and is in continuos rework based on researches. Now it looks like that. There are several variants of buttons, because of testing in this exact moment.