Location-based targeting, many pivots later
Post-mortem of Loceo
This story takes place in the mist of the late Web 2.0 era, 2010. The iPhone had just set the mark for the future of phones. People talked about apps. Every month there was a new iphone-killer. Unlimited data-traffic plan were introduced in Sweden for the first time ever (later removed). Facebook was cool and there was no snapchat.
One day I just got a crazy idea. What if you could bombard people with text-advertisement? What if you could locate their phone and do location targeting on the ads? What if also the users would get a split share of the revenue, then everybody would join, right? Right? Take a look around, there is already a company called mginger.com (Acquired by ValueFirst Messaging on June 27, 2012) in India doing this. The other day on Swedish Shark tank, we saw a company called Manna från himlen (1M SEK investment, later bankruptcy), doing something similar. Well this sounds very good, we are on to something. Let’s start hacking.
We where targeting the Swedish market. We easily hook up Ericsson IPX and their API to do cell-triangulation of any phone with consent. Since we where off-loading our users cellphone bill the name Tellifree was great. We where simply letting people speak freely on the phone. The work on a high-performance Java backend began to meet our overwhelming demand of users that were later going to join us. We got a small grant from Swedish ALMI to make a legal review of the business and see if our space was potentially patentable.
Our technology was based around doing analysis of peoples area of movement, area of rest, and current position. When a new incoming position came in, we were supposed to make a fast decision on what ads to send. Additional targeting by user-chosen interests would make it bulletproof. The name was changed to Loceo. The best name I’ve ever come up with.
We did a small launch just to get some users into the service. At the same time there was something big going on, we had a meeting with a business developer at Telenor. We needed their help to setup a solution to deposit money into our users phone bills. Something they needed to implement. At the meeting we got our advise, ‘Come back when you have the advertising agencies with you’. Fair enough, we didn’t have our first customer. We continued back home doing simulations, optimizations, and design. Far too less sales, big mistake.
Summer 2010, my motivation was at it’s bottom. My partner quit, tired of over-working a backend nobody was using. Was this the end of Loceo? Well, not for me. I still liked the name and, the way I saw it, after pivoting it still had a lot of potential. Loceo took a new form as a service provider aimed at developers using geo-location in their apps, much like SimpleGeo (Acquired by Urban Airship on October 31, 2011). The pivot towards infrastructure-as-a-service put out new requirements such as developer documentation, client APIs, and support. Building a developer community failed without much of a try. Meanwhile creating these APIs, also I made a Loceo Spotify App, iOS app, and
Our technology-stack consisted of Redis as a geo-spatial index, a document-store in MySQL, and Memcached. The geospatial index in Redis was implemented using geohashes in sorted sets. This was a good approach and very efficient, using only 64-bit integers. It allowed geo-shapes to be index within multiple resolutions. One shortcoming was however the need of after-processing the results.