• slider
    CASE STUDIES
    Insight & Impact

Tracking a Wearable Product Launch

Case Study 1

A global consumer electronics company was on the verge of releasing a brand new wearable product. Given such a high-profile launch, the company needed to be ready to resolve any issues that might crop up and much more quickly than traditional methodologies could handle.

They needed a solution that could detect issues automatically, allowing them to instantly prioritize and route incoming support tickets with ease. The problem was that most natural language vendors relied on a keyword-based approach. This meant they'd have to manually build and maintain a list of ontologies to tell the system what to look for. The weeks to months-long setup time for this approach would not work with the company's product launch schedule. In addition, consultants would have to regularly update and maintain the ontology, adding to the time and cost of the overall solution. Most concerning was the fact that any unexpected issues, for which a keyword had not been programmed, could be missed entirely.

The company decided to work with Luminoso because of its foundation in artificial intelligence and natural language understanding. Instead of scanning text-based feedback for matching words or terms, Luminoso's solutions focus on understanding the underlying concepts in data - no matter which specific words are used to describe that content.

As early as 24 hours after the product launch, the company began reaping benefits.

Fast Time to Insight: It took less than ten minutes to set up in contrast to the weeks to months it would have taken with another technology. The company was able to begin analyzing incoming support tickets and finding insights right away.

Multilingual Capabilities: Luminoso supports 15 languages, all of which are included in the software. No extra time, training, or work was required to get Luminoso's solution up and running in the five major languages in which the company was receiving feedback.

Identify & Categorise Intent: It accomplished this in near-real-time and at a high level of accuracy.

Uncover Unexpected Insights: Within 24 hours, the software surfaced several unexpected insights. For example, It uncovered a subset of complaints related to a "scratched" or "cracked" screen. While the matching label - "Repairs and Physical Damage" - seemed obvious, the underlying cause of the issue was unexpected. Rather than being due to user damage, the technology was arriving with already-cracked or otherwise damaged screens. Determining the root cause of this issue enabled the company to take appropriate action and resolve the issue.

In addition, the software was able to uncover entirely unanticipated issues that would have gone unnoticed using a keyword-based approach. One such issue was a dial on the product that was critical to the use of the device but which users reported as "sticking" or being entirely "unresponsive." Even for consumers who had no issues using the dial, many found that the direction the dial turned was counterintuitive. Because this issue was quickly surfaced and isolated, the support team could take immediate action.

Monitor Trends: Luminoso's software monitored the prevalence of assigned labels and produced easily digestible reports, which the compan's analysts could use to quickly identify emerging issues. Key issues were monitored to see if they were trending upwards or downwards, enabling the support team to prioritize which problems were most pressing.

With Luminoso's software, the global technology company was able to analyse and monitor their support processes in a more accurate, rapid, and streamlined way. The findings uncovered in the weeks following the product launch were critical in helping the technology company smoothly and successfully introduce its new product to market.

<

Ticket-Wrangling with Mobile Game Developers

Case Study 2

In the world of mobile game development, there is a very real and looming threat of game-breaking bugs slipping into the latest release, causing users to frantically swat at the bee in their bonnet via support tickets.

For one particular developer, this threat was made much more manageable with the help of both Luminoso Compass and Analytics. That combination helped them move these bees from user's bonnets, to a hive with enough evidence for a new feature, perhaps a bug fix, or ideally, a link to the right FAQ page to enable self-service.

With 4-5 big game updates every 2-4 months, their servers are hit with up to several hundred thousand tickets daily, and in multiple languages. The problem was that their previous keyword-based system couldn't quite keep up with the frequently-changing language used in their games. So, they needed a way to automatically send customers with simple issues to a helpful FAQ page, ultimately preventing the team from drowning in a sea of tickets.

For Luminoso Compass, the ability to understand underlying concepts, rather than match keywords, made deciphering this domain-specific language possible. Everything from in-game items and classes, misspellings, or acronyms could then be caught and automatically routed to the right response. And if a ticket was assigned a label that didn't match an existing self-help article, the ticket would be automatically routed to a support representative for review. This essentially doubled the number of tickets they were able to deflect to self-help pages, all while maintaining 90% accuracy - a huge win for the support team's time-to-resolution.

On the other side of the coin, Luminoso Analytics was used to analyse this huge influx of tickets and unearth issues that were more complex than a simple FAQ page. For those tickets, Analytics would identify and quantify the issues before routing them to game specialists for further action. The game specialists would use this data to start tackling the issues that appeared in the data most often. From there, the summary of topics (listed as relevant terms) and percentage of voice among users could be shared with the rest of the team. This also enabled the team to track specific topics as well as the actions support agents took to solve issues related to those topics.

With the ability to intake, process, classify, and tag data in real-time, the developer was able to handle high-volume feedback 24/7, and in the eight languages they cared most about. Not only that, Luminoso was set up in a day and trained in less than 10 minutes using past support tickets in multiple languages before being connected to the company's ticket management platform. The impact this workflow change had on the business was huge, helping the developer keep gamers happily gaming and support teams confidently supporting.

Email contact@kmqanalytics.com for more on our machine learning solutions.

Contact Us