Bees and Pollen’s Alan Avidan on Optimizing your In-game Conversions, Staying Focused and How to Serve Your Users

Alan Avidan has been a major contributor to Bees and Pollen in the past year. As the company’s director he’s been busy driving the HoneyLizer, an optimization platform that collects and parses social and behavioral data to enrich a game’s performance through virality, monetization and engagement.

He reminded us that the platform, a visionary brainchild of Udi Barone and Yaron Cohen, was started with research and patent filings in social networks back in 2005.  Both Udi and Yaron understood the potential of leveraging social data for optimizing users’ experience in websites and apps, and pursued their vision by developing the HoneyLizer technology.

Although Avidan has been involved with startups for the past 20 years, his interest in the video game industry blossomed when he noticed the social features-centered state it was in. “The video games industry is just screaming for help in getting users better engaged and better monetized,” says Avidan. Engagement and monetization were just the tip of the iceberg of problems he saw, and he felt that he could help chip away at those issues.

Learning to Listen

Throughout his career, Avidan learned lessons that have been useful to him now. “Over time, I’ve learned to focus on the real goals and not be distracted by every twist in the road,” says Avidan. “But the most important thing I’ve learned is to listen (really listen) to our customers’ needs and dreams.” Listening and understanding the customers has been a key strategy for Bees and Pollen—an important key to its success.

“There are a number of ways to understand a user.  A well-known, highly-practiced method in the industry is A/B Testing,” say Avidan.  “The problem with A/B Testing is that ultimately your final choice is only the marginally better option of those you’ve tested, which then becomes the ‘one-size-fits-all’ default option.”  According to Avidan, user segmentation can provide specific preferences of subgroups, even increasing key performance indicators, if done right.  He also says that cohort analysis can be effective in tracking user’s performance over time through a common reference point, possibly providing insights into what’s wrong and right with a game.

“More advanced segmentation methods can produce better results but are sometimes not worth the results received,” says Avidan.  “Consider having ten game elements that need to be tested, each with five options. That’s 50 tests, each lasting a couple of weeks. By the time you’re done you need to get started again, as your user base and their preferences may be changing.  You will also need dedicated staff and a good platform to really make that work.”

“HoneyLizer, on the other hand, provides each user with the game options they are likely to prefer, based on algorithms crunching their social and behavioral attributes,” says Avidan, “it does all the heavy calculations seamlessly, automatically and in real-time.  This invariably leads to higher conversions and higher KPIs.”

There are many ways to learn more about users, but choosing to do so is only part of the battle. What to do with the harvested information is a critical piece of the puzzle. “There is not enough I can say for how important it is to understand user behavior to benefit the game,” says Avidan. “The trick to winning, though, is to be able to reliably predict behavior and automatically act on the prediction in real-time so that you can serve users with the options they prefer.”

Comments