Project Portfolio: Jetpac
From Idea To Acquisition
In March 2013 Jetpac decided to pivot away from the Jetpac iPad app showing your friends’ best travel photos from Facebook.
The new mission was to launch a new mobile application, this time on iPhone, based on geo-location data with the goal of reaching a million users before July 2014.
Product Manager, UX Design (including User Research and Usability Testing), UI Design Exploration and Mockups, QA, Product Marketing, Customer Service, Community Management and some HTML/CSS coding.
In addition to the app I worked on designing Jetpac's main website, blog, email campaigns, widgets and the app's web end-points.
We started the process by brainstorming different concept ideas based on geo-location data. Before we started building anything we first wanted to test the different value propositions to see if there was a need, and we did that by talking to a lot of people and running Facebook ads to test the response.
Once we had narrowed down to a few concepts I started producing wireframes and UI designs daily for further testing and validation.
Based on our research findings we decided to move forward with a city guide app that would surface the top 10 most popular places around you at all times. This version got the working title "Jetpac Local."
We built a web app prototype and ran around all of San Francisco to test what the result were like compared to other location data apps. Unlike Yelp, Foursquare, Sosh and FieldTrip, Jetpac Local’s recommendations would change depending on your exact location. This seemed great until we kept the app open during a cab drive and noticed the app window looked like a spinning wheel, constantly refreshing the feed as we were moving.
A PHOTO CENTRIC APPLICATION
Our previous app was a photo centric app and we wanted to continue in this field that we felt we already had an expertise in, plus not to forget our valuable image processing and machine learning algorithms that gave us an advantage. We were in addition working on developing an image recognition algorithm that made it possible for us to understand what was in a photo. This technology was a breakthrough and became the foundation for the final concept for Jetpac Local.
When looking at the public photos we had collected from Facebook, Flickr, Google+ and Instagram it was clear that focusing only on Instagram photos for the app made the most sense. The reasons for this decision was because:
1) Instagram photos seemed to be of better quality and more interesting for local discovery than photos on other platforms
2) There were more publicly shared geo-located photos on Instagram than on other platforms
3) It would be easier if all photos had the same dimensions.
UN PHOTO QUIZZES
A concept we ran with for some time was a photo based city quiz for the top locations in each neighborhood in each city. We then tried combining the two, creating an app that would quiz you on how much of a hipster you are based on the places you’re able to recognize from Instagram photos. . We could now use object recognition to spot hipster mustaches in photos and give each venue a hipster score. To add another gamification element, we showed funny animated GIFs after every question answered, either epic fails or hilarious wins to give the user an instant reward and incentive to keep playing. This was the element that received the most positive feedback in product testing with users.
The reason for this niche game angle was that we thought that could be a funny and unusual approach to a city guide that would create a lot of attention and hopefully traction.
People seemed to enjoy the quiz and thought it was funny, but they felt the app served no real purpose and did not solve a problem. We were at a cross-road where we had to decide if we wanted to mainly provide entertainment and double down on the fun and silly features, or provide more useful information about places. We chose to go with the latter because it seemed unlikely to reach a million users only focusing on hipsters, but we wanted to still keep the quiz as a sub feature since people seemed to enjoy the animated GIFs so much.
OUR TARGET USER
We kept iterating and testing the app until we got to a feature set that solved a problem for a particular user group – young active people in cities wanting to find new places to check out.
Our target users were making their decisions based on photos so they could see what a place looked like and get a sense of the type of people hanging out there to judge whether or not they'd like it.
This group of people were currently using services like Yelp and Foursquare, but had a hard time telling from all the reviews from random people if it would fit their own taste or not.
The places they were looking for in their daily life was typically bars, coffee shops and restaurants, but would also look up places for weekend activities like hiking trails and parks or for travel planning venues like hotels, camping spots or museums.
We didn't have a defined persona at the time, but this is what he could have been like:
John is 32 years old and lives in Seattle. He lives an active lifestyle and likes to spend his weekends with his wife and friends, like going to brunch or meet up for dinner. John wants to find cool new places to check out and he cares about what a place looks like and the vibe.
We also realized that as we moved more towards a traditional city guide concept to satisfy this user group, the quiz was not longer the main activity and had to take the back seat visually.
THE SOLUTION: JETPAC CITY GUIDES
We had landed on a solution that leveraged the world's Instagram photos to create city guides for any city in the world showing you the Top 10 most popular places in a city in a certain category based on certain filters. The new technology we had developed allowed us to create not only standard guide categories like restaurants, bars or parks, but also categories based on objects in photos taken there or the type of people taking these photos.
In Jetpac City Guides a user could see guides like Top 10 Bars With A View where the results were aggregated from looking trough all the photos taken at bars, then finding the photos where we could spot a large ratio of blue sky, then sort these places by popularity in terms of the number of people sharing photos from this place.
We hired a visual design agency to provide UI assets based on my mockups and I worked with them for three weeks to get the app ready for launch.
We launched the app in December 2013 and got featured by Apple "Best New App."
Until we reached product market fit we decided to code the app in HTML5 and CSS3 with a native wrapper in order to be able to rapidly test and iterate the app without having to wait for Apple to go through the review process every time we wanted to make a change. The tradeoffs were speed and smoothness and we had to spend some time optimizing loading time as best as we possibly could.
We made weekly changes in the app in the months after launch based on closely monitoring key metrics and findings from usability testing.
Some UX problems around 1.0 we discovered and solved in version 2.0
1. The bottom nav bar was confusing separating top guides, categories and people.
2. Users wanted to see venues in relation to each other in a map view.
3. The limited curated Top Guides did not meet user needs to find cool places
4. Users did not find the favorites collection useful because it was not sorted by city but by time added
5. We didn't know what type of categories worked better than others for different cities
Below are some of the changes we made to solve these problems. We also learned from looking at behavior data in Kissmetrics and reaching out to these users that our most active users were using the app for vacation planning. One reason for pivoting away from the Jetpac iPad app that mainly focused on travel planning was that people don't travel that often. With Jetpac City Guides we wanted tap into the market for local discovery, but if our most active users were using it for travel planning, should we start optimizing for this?
Providing nearby search range options.
Providing guidance and alternatives to unsuccessful searches
Optimizing for growth. Pinterest was our main user acquisition channel.
Favorites grouped by city, optimized for travel planning.