At Bird, while the micro mobility industry was just starting to pick up, Hudson was responsible for product localization, making sure that Bird’s app, website, and content were available in every language needed. By working with other product teams, operations teams, legal, and engineering, Hudson was able to productize and automate Bird’s localization into 19 languages by his departure.

This process involved automated string management for Bird’d ios and android apps, website, marketing materials, help desk articles, email, and phone support. Bird’s localized support system were handling 50k tickets per week across those 19 languages, with agents sitting in 3 countries.

The localization process was also directly integrated into the engineering CI/CD flow, complete with human expert reviews of each string before publishing.

At Bird, Hudson’s team was responsible for support team efficiency. As part of this process, we built and tested several automated chatbots aimed at support deflections and cost savings.

Through rider and financial analysis, we discovered that our average cost per support interaction was actually higher than the average refund amount. From this, we built a preliminary “trust score” for each rider, which would decide if a refund request could be automatically granted or not. This process resulted in full automation of nearly 70% of all support tickets regarding ride refunds, and let a single supervising agent mange the workload that previously required 12+ agents.

Additionally, while never approved for final deployment, Hudson’s team built a support chatbot that, during initial testing, was able to deflect or automate close to HALF OF ALL inbound support tickets. Had this program been released, estimated cost savings to the org would have been over $1.5M/year in BPO costs.

At HelloTech, a 2-sided marketplace for in-home technology services, success of the automated dispatch algorithms was measured by the “True Success” metric. If dispatch failed, and needed human intervention from our fulfillment team to get the job completed, it was called “Partial Success”. Hudson’s team was responsible for maximizing the True Success metric, getting it as close to 100 as possible.

By analyzing the greatest drivers of dispatch failure, and the most common fulfillment agent tasks, Hudson’s team created a series of product and support automations that drove True Success from the high 60’s, to the low 90s over 1.5 years.

Additionally, because human intervention in dispatch was so dramatically reduced, our team was able to implement far more robust A/B testing infrastructure to our dispatch algorithms. Through this experimentation, we were able to modulate customer pricing and working payouts by job type, by market, by date/time/etc, resulting in a nearly 10% additional gain in gross margin, with no reduction in worker engagement, and no increase in worker churn.