ALFA-ARKIV Mobile App Overview
ALFA-ARKIV is a digital novel masquerading as a game. Combining themes of surveillance culture, genetic modification, and Soviet history, it is as culturally relevant as it is technologically innovative.
Released in 2014, ALFA-ARKIV was featured as a "Best New Game" by Apple and lauded by media outlets such as CNET and the Canadian Broadcasting Corporation as one of the top mobile experiences of that year. More of a digital novel than a game, the core of the experience is a journal written by award-winning novelist Shani Boianjiu, author of "The People of Forever Are Not Afraid". The game requires players to explore the story through different platforms and features an extensive use of a Siri-like chatbot character who is preprogrammed with over 100,000 words of code.
Features
Features an AI chatbot who serves as your guide, hintbook, and salesperson.
Over 100,000 words of discoverable text.
Adaptive, context-sensitive soundtrack.
Unlocks content across mobile and the web, with geospatial data, geo-fencing, image recognition, AI chatbots, and various APIs and SDKs, including augmented reality.
Powered by a feature-rich backend, including a user engagement tracking system.
Situation: We noticed a high drop-off rate before the paywall on our iOS app. Was the user attrition caused by the AI chatbot implementation or the face / image recognition SDK?
Task: Normalize and marry two sets of metrics to identify cause of drop-off.
Action: I looked at timestamp metrics between events, from download through to paywall. I then split the usage stats into quartiles, by session time for each event.
Result: The analysis identified the image recognition SDK as the source of the problem. Based on this finding, our developers adjusted the acceptance threshold for the SDK. The analysis also highlighted the high conversion rate of chatbot for further investigation.
Description: The attached diagram tracks user flow through an AI chatbot-gated paywall on our iOS app using the Hexagram user-state tracking system.
Problem: During initial user tests, we noticed a sharp drop-off between the initial download and the paywall. We suspected either a flaw in the AI chatbot design or the augmented reality face recognition feature as the source of user attrition. However, we had little insight into this flow because we had not yet built a fully integrated and comprehensive reporting system. The only indication of the user attrition issue was the reporting from apple indicating the number of downloads and the number of purchases at the pay wall.
Solution: The attached diagram helped us to identify the camera feature as a major friction point and to identify the AI chatbot as a solid conversion gate with ~5% overall conversion (against industry standard or 1-3%). Even better, later tests with larger, statistically significant user counts have shown up to 20% conversion rates at the AI-gated paywall! I pulled reports from the Apple store and married them with aggregated timestamps of user state changes in our tracking system to get the data used in this diagram. This ad-hoc report could also be turned into an automated reporting tool to identify similar friction points across all inputs for other partners on the Hexagram platform. In the end, we adjusted the threshold settings for this camera feature to allow users to easily progress to the next chapter of the game.
Note: in an effort to preserve confidentiality, some of the data provided in this example has been redacted, abstracted or altered.
CNET: Alfa-Arkiv
Best Mobile Games of 2014
" Alfa-Arkiv is about as ambitious a multimedia project as we've ever seen. The core of it takes place in the iPad app where you, as a new operator at a mysterious organization, are reading through documents pertaining to the detention of a young woman named Rhea, a member of a resistance movement called the Liberation Army of Dagestan.
While it technically falls under the definition of an alternate reality game, Alfa-Arkiv isn't easy to categorize. It's sort of an interactive novel, but it's so much more: nearly 10 years in the making, it will send you crawling the web hunting for clues planted by the development team years before the app's release in July of this year.
It failed to get the attention it deserves, partially because it's not easy to categorize as either a novel or a game; partially because it asks things of the user that go beyond a single screen; and partially because it's so very realistic. It is, however, a spectacularly executed piece of work, and a magnificent experience."
