Making a mobile app? Iterate with mobile web first
There is a lot of talk about the Lean Startup in which you create a minimal viable product, get it in front of customers and iterate rapidly. With the Web, this process was much easier to implement as you could make one product and it would work cross platform. Sure, you have to support multiple browsers but it is no where near the burden of having to support two mobile OS’s, one of which has an approval process for every update, and a multitude of different devices and resolutions.
Although I had developed mobile apps in my job, I had yet to set foot on developing a mobile app on my own until Chinchin. Chinchin came about because I noticed many of my coworkers would set each other up on blind dates. While this is nothing out of the ordinary, it was interesting to see them show each other their Facebook friends list and find someone they found interesting.
Minimal Viable Feature
I quickly realized it would be interesting to see a list of a bunch of people who were friends of my friends. People who were connected to me via two degrees of separation. I talked to my friend and we decided to mock this up with a website. There was not much to it but when you registered, you would see a list of people who were friends of your friends. You could not do anything else. No extra photos. No detailed profile information. Simply a persons name, their photo and the knowledge they were somehow connected to you by a friend.
The initial response was positive. Here is a sample comment during usage:
This is so fun. There’s nothing you can do but this is so fun.
It was clear we were on to something. We also received valuable feedback for improvement:
Is it possible to find out who the mutual friend is?
Minimal Viable Product
We decided to put more effort and actually create a dating service where you would be able to connect to someone. We added more profile information, photos and the ability to see who your mutual friend was. Then, we added the capability to like the user. All along the way, we gathered feedback from users to guide us along the way.
We knew from the beginning Chinchin was going to be a mobile app but then the question came, when do we go native? We made a decision early on we would develop in two phases:
Product/Market Fit. We would choose a key metric and only when a key metric hit a certain threshold, we would consider ourselves to hit product/market fit.
Product Optimization. Once we hit product/market fit, we would then optimize by going native for a better user experience and optimize other key metrics.
Mobile App Development
We created iOS and Android versions of Chinchin by simply embedding our mobile webpage into the app. 100% of our service was mobile web. We were a bit nervous we would be rejected by Apple but we managed to get our app approved.
While UX was horrible, we were not concerned with this. We decided we simply wanted to measure how engaged users were with our product. The key metric we chose was the average percentage of members who would like someone on the service.
Gathering Data & Iteration
We installed Mixpanel to measure various types of data but we were most concerned with our key metric. There were good metrics but our key metric underperformed. While people would view a number of profiles, not many would actually like anyone.
We went back to our friends and watched them use the app in realtime and realized, people did not know you could like someone due to an unintuitive UI. We went back to the drawing board and redesigned the service.
Originally, we showed a list of 6 people who were friends of your friends. You could then view someone’s detailed profile and like the user from their profile. We decided to make a change so you would only see one profile at a time and you would either have to like or pass on that profile to see someone else.
All we had to do to change the mobile apps was to change the mobile website. It actually took less than a day and we got it back in front of users and our one key metric jumped.
If we had to deal with building native versions of Chinchin for iOS and Android, we would probably be a month or two behind where we are now. While our friends complained about the UX, it was more important that we build something people wanted. The data and user feedback seems to indicate we are really close to a product people find fun, interesting and useful.
If you would like to try Chinchin, visit http://chinchinapp.com where you will find links to our apps for iOS and Android.
I was inspired to write this blog post by two other blogs I have stumbled upon recently.
Premature optimization by Sam Altman
Mobile app startups are failing like it’s 1999 by Andrew Chen