Every weekend hundreds of people meet at public parks in Argentina to find and trade collectible stickers. We saw a strong community lacking organization. There are many online message boards and WhatsApp groups where people try to find the stickers they need. We want to seize the opportunity to make this process more efficient and to bring people closer together by finding the best match for trading.
What it does
Users can upload collectible stickers by just taking a photo and also input which ones they are missing. Our algorithm will create matches among those people and notify them. They can use our bot to coordinate a meeting point to make the transaction.
How we built it
We chose Messenger bots because Messenger has a high penetration across LatAm. Users get a cross-platform experience, inside a known interface they can natively navigate and we can send push notifications to let them know when we find matches, which is key to the success of the product. The integration with wit.ai gave us NLP capabilities out of the box. These technologies made our life easier allowing us to focus on the product.
Challenges we ran into
As any community-oriented app, there’s a critical mass that we need to get to be able to provide a useful service. We know it’s better to start local so the plan is to create targeted online and offline (Messenger codes) campaigns at major trading centers and make an initial investment in a large number of stickers to make sure people hopping into the platform will find trades quickly. We also face a challenge with privacy, safety, and moderation, since minors might be using the platform, there’s a need for referrals and message moderation to prevent bad actors from getting into the community.
Accomplishments that we're proud of
We were able to focus on one vertical of the main product and build an MVP that served as proof of concept. The machine learning algorithms to recognize the stickers and to find optimal swapping loops for the people in the community provided a delightful experience for our testers. We also added internationalization support (Spanish and English for now) to make sure language is not a barrier to using the bot.
What we learned
Building community is hard and comes with a great responsibility. From listening to the feedback from our early testers we learned that we needed strong moderation to guarantee a safe environment for the people involved. Product and algorithms are just the beginning, it’s how you launch and listen to your users needs what allows a community to thrive.
What's next for Swapr
We decided to focus on collectible stickers because there was a rush during the World Cup but our idea could actually work for any kind of trading where all items have the same value. For example, we think we could use our app for developing a tutoring community where people exchange classes without money. We made the code open-source and agnostic to the specific thing being swapped to allow this in the future. We can also leverage FB Login and the graph API to create a Swapr platform, where anybody can take an FB page, hook it up to our servers and create a new local swapping community in a matter of seconds.