Building a Service Bot Requires Manual Work, Too
Hannah Bertiger is a Product Manager at Walmart, focusing on building internal products that help associates do their day to day work. When not thinking about new features that help employees save time and money, Hannah can be found doing yoga, cooking, or hanging out with her dog. She is based in Austin, Texas.
Hannah will also emcee Her Product Lab’s Virtual Summit on March 4, 2021.
Like most companies, when the pandemic began, Walmart U.S. associates shifted to working from home. With a workforce of over 1.2 million, there’s a lot of moving pieces to consider in regards to technology and support needed to help employees get their work done.
The Global Tech Team at Walmart significantly ramped up our chatbot solution, Support Chat, to provide automated support to associates regarding technical issues. Whether it was connecting to servers, needing a mobile hotspot for remote areas, or troubleshooting video-calls, our product soon became the go-to before filing a ticket with our internal support team.
In order to have the most empathy for the problem you’re trying to solve, you need to eat your own dog food.
Our north-star is to provide a consumer-grade experience chatbot to assist associates with technical issues, but to better align with that goal, we needed to understand the intricacies of support teams and associate’s frustrations with the current processes. In order to have the most empathy for the problem you’re trying to solve, you need to eat your own dog food —
and that’s exactly what we did.
Understanding the problem: firsthand experience of the user journey
As a product manager, you’re told you need to fall in love with the problem you are trying to solve. But if you don’t actually experience the issue firsthand, how can you develop a solution? As part of the workforce at Walmart, our team also experienced issues with technology. When trying to troubleshoot we went about finding a solution a few ways:
- Asking a co-worker or posting in a chat channel
- Pinging our very overwhelmed dev or go-to IT person
- Searching through knowledge articles on our intranet
And what did we find? Sometimes the right answers, sometimes a 404-page, and most of the time more frustration. If we, power users, who know all the right channels to scan for an answer couldn’t find it, we knew a good percentage of our workforce was also experiencing the same.
Try manual solutions to understand the best paths to build the product
Support Chat’s purpose is to automate solutions so you don’t need to go digging for an answer. But before building out all the bells and whistles, we found it was also important to understand how our current support teams were helping associates troubleshoot.
To start, we listened to recorded conversations from our call centers. We wanted to understand the bar of customer support teams were held to, how they think through solutions, and, most importantly for a chatbot, how they spoke to our associates about the issues they were experiencing.
We want to ensure we develop a human-centric and friendly chatbot that mimics the best parts about calling a help desk.
The second path we ventured down was answering support emails. On our products help page, we include a forum to ask questions as well as an email where associates can write in with issues. By working through some of these problems ourselves, we were better able to understand the need to automate simple tasks and how this would free up the support teams to conquer more pressing or complicated technical issues. Even more so, we understood the importance of providing simple and precise instructions for troubleshooting technical issues. Too much fluff leads to more confusion.
Developing your feedback loop
As with all good products, you need to iterate on your MVP. While talking to our associates can be challenging as they are also heads-down with work, eating our own dog food, allowed for a feedback loop to be developed.
Each time we deployed a new feature update or added functionalities, we could see how well they were working. If we saw our inboxes or forums flooded with questions- it was an indication that support tickets would be going up and we needed to fine-tune. If the updates or topics weren’t trending, we were provided qualitative data that we were heading in the right direction.
No matter your product or solution, it’s important to understand customer pain-points and the various ways to solve the problem. By eating your own dog food- or better said, diving headfirst into using the existing products or solutions firsthand, you can expand your perspective on what you need to be delivering on your roadmap. If you just develop solutions and have no idea how they work in the real world, are you really solving the problem?
