Shufl

The predecessor to Bunz, Shufl was a buy/sell/trade application which launched in the summer of 2015. It grew to a small user base in Toronto and got Artifact Labs Inc. a seed round to continue building great products with a great team.

 

Initial Requirements:

  • Show users local content sorted by relevance based on an algorithm which sorted a deck of products by recency, proximity and the user’s interest in the item
  • “Match” products on trade posts - ie: if I liked another user’s Watch and that user liked my Bicycle we would have a match
  • Increase value of your offer by adding cash to your trade transaction
  • Meetup management ie: End-to-end experience - can we ensure safety of the users via location services and also let both parties arrange a date/time to meet up
  • Simple profiles which focus on the goods users are offering for sale or trade
  • A user should be able to accept or decline any offers which are in their inbox

Onboarding was complicated — there was much to explain in terms of how it worked, what to expect and how to send an offer. There was a lot of time spent writing the “perfect” app store copy, working on a pre-signup landing page which explained the value propositions of the product and why to use this over some other brands.

Shufl home screen. There used to be a deck of cards much like Tinder that a user could swipe left or right on,

After a month in market we removed the deck of cards on the home screen as we found if users didn’t get a match or find something they were interested in after 3 swipes then they would stop using the app completely. We then switched to a horizontal list of product cards which worked a lot better for the time and we did not see as much churn as a deck of cards where you couldn’t see what card might be next.

Post view was probably the easiest flow to design within the app. All it needed to be was a way to consume content about one post — view basic information such as the image, item title, description, price, if it was available for trade or just buy only. Be able to view comments, add a comment, see likes on the post and finally make an offer. Since this was a conversion metric for us to increase sales I chose to add a large green “Make an Offer” button on the bottom of the screen which persisted across the 3 tabs in the post view.

Offer flow.

If you could offer an item and also cash on something you wanted but didn’t want to outright buy, would that be enough of a reason to use this app? The value of the seller’s item and the value of your offer as a buyer was left up to each respective party to judge independently. The app just pre-filled the cash input box with the value of the seller’s ask for that item, if it was a trade item without a price attached then the value would be $0.

Upload item flow.

The requirements for this flow were fair enough: Add a title, description, condition of item, price and a user had to choose an option: if the item they were uploading was for sale (Cash Only) or for trade (Accepts Trades). There was an optional checkbox on Cash Only posts which was “Cash Firm” where the make offer flow would not pick up items from your profile. #hashtags were accepted in the description text area.

I will admit, the inbox with 4 separate sections sounded good in theory but failed miserably in practice. Each user we tested this on (about 20 people in total) got so lost within the “Offers” inbox alone that they stopped using the app. The 4 sections were as follows: A pending offers inbox, an accepted offers inbox, a rejected offers inbox and a sent outbox which had 3 sections inside it: offers, accepted and rejected.

To further distinguish the Shufl product from competitors such as Kijiji (local classifieds), eBay and Craigslist we tried to implement a location-aware meet up management system. Either party in the transaction could set a date, time and location to meet up and once the other party accepted we would track a user’s location allowing reviews to be made after both parties were in the geofenced area.

The focus was on the items on a user’s profile but we also tried to make the profiles a bit “social” by including bios, being able to share your Shufl profile on 3rd party apps such as Facebook or Instagram. We hoped that the social shares would help the Shufl platform grow, but if anything it wasn’t worth the development time to the tradeoff of less than 100 shares while Shufl was still active.


Goals:

  • Get user feedback from private beta to improve product: November 2014 - June 2015
  • Get 500 signups within the first 3 months of the app being launched to the public (July 15, 2015)
  • No success criteria for offers made or transactions completed: Any offers/transactions done on the platform were a success to us at the time
  • Allow the app to be used in any location worldwide — heavily reliant on a user’s device location
  • Start collecting user data to feed a work-in-progress algorithm — this data was based on user’s likes on the app, likes pulled from Facebook and the types of posts a user uploaded
  • Build a view-only web app which would showcase all the items from the product

Challenges:

  • Our first 50 users who we were testing the app with were our friends; Anything we did they automatically liked and gave bias feedback
  • We didn’t know what to expect in the first few months of release for our in-app funnels (transactions completed, posts created) so we heavily relied on our instincts
  • All of our data was stored on a free version of Mixpanel which would lock us out pretty fast from the data we were collecting
  • Image uploads and processing was really slow, at this time we were using “Parse” a product that we built our backend on
  • Team size was small so we had to do multiple things all at once, while working part-time
    • Team size was small: 1 iOS Developer, 1 Design/Web Dev (Me), 1 Marketing/Growth, 1 on Finance (CEO), 1 on Legals (COO)
  • Search functionality was very limited, and the amount of items on the platform was less than 200
  • Algorithm didn’t work at all…

Problems Encountered:

  • Since the algorithm didn’t work as we had hoped, the content on the app became stale very quickly
  • The app was very buggy — the iOS developer built the backend on Parse and the data flowing between the backend and the frontend was too large at times to process (image sizes were massive, data was slow to respond).
  • Offer flow from making an offer to accepting/rejecting was too convoluted for a lot of our users — why would you offer your item for trade when you could just sell it outright?
  • Meet-up and the deck of cards to recommend you products wasn’t interesting enough for most users (saw a steep decline in active users after they had signed up)
  • No sense of brand or reinforcement of the vision
  • No retention emails, ability to target our users based on their interests, demographic data or any other means — we were only able to send an email to every user on the platform (around 340 users)
  • Marketing our message was a clear challenge — there was no way we could put it into a single sentence, it wasn’t unique enough to warrant huge marketing campaigns

Some Wins:

  • Able to prove that we could build a product with a small team
  • 2 people in San Francisco used the product after hearing about it on Dribbble
  • Media in Japan picked it up (where the app wasn’t even available), some media in Quebec printed an article about Shufl in the local newspaper: https://twitter.com/thomasjkerjean/status/655363599335071745 
  • Raised $1MM seed round from a local investor with the condition of getting 10,000 users on the app within a 6 month period (between October 2015 - April 2016)
  • Found a line that resonated with a fair amount of people: “The heart of the city”
  • Had 500 people finally sign up after 4 months in market