Wise Insights

Pinterest’s Customer Support Team Leveraged the Brave New World of Automation Technology

To many customer support teams, endless ticket queues are a fact of life. Yet, huge backlogs make it feel impossible to catch up and difficult to deliver a reliable response time. Implementing web contact forms can open the floodgates, and volumes can get out of control. Add a low customer satisfaction (CSAT) score and you have a recipe for low agent morale. How can you turn things around?

Bringing on new automation technologies is one option. Sure, it can seem like a gamble. After all, tried and true methods may seem less disruptive to your customer base. However, complacency can lead to stagnation, and inaction can be more detrimental in the long run.

If your customer support team is looking for ways to speed response times, improve agent productivity, and reduce ticket volume, consider investigating the many tools available to help you:

  • Solve problems
  • Reduce errors
  • Improve customer satisfaction

Here, we’ll take a look at Pinterest, a top client of ours who is leading the way in this new world of automation — and score impressive results at the same time.

HiRes.jpg“The year of machine learning” presents opportunity

According to a recent article in InformationWeek, machine learning and artificial intelligence (AI) are gaining traction and will be included in more types of platforms and software in the coming years, meaning there will be more opportunities to implement this technology in the support realm.

Still, customer support leaders tend to be somewhat conservative in their approach to new technology. Too often the perspective is that there’s too much at stake, too much risk in trying new technologies or changing routines.

And yet, what is there to lose?

Learn how Machine Learning puts an end to customer service problems in our guide to customer success.

The risk and reward of new technologies

One of our clients, Maggie Armato, is the head of customer service at Pinterest. Her goal is to make the most of automation technologies without losing the personal touch of an agent-generated response.

Having recently hired four new agents and faced with thousands of tickets in the backlog, she wanted to learn how she could take support to a scalable new level without resorting to hiring additional agents. “There was little to lose and a lot to gain,” Armato says.

We knew right away that we could help solve some of her most important issues, including:

  • Improving time to response while reducing handling time
  • Delivering templated recommendations (macros)
  • Reducing ticket volume by implementing automated responses to common questions from their customers base (known as “pinners”)
  • Reducing re-open rates and improving CSAT on keyword-based automated replies

Armato tackled the conundrum with a pioneering spirit. “We wanted our customer support experience to be the best it could be,” she says. Her solution-oriented work ethic and experimental mentality — supported by a fast-paced Pinterest office environment — led her to explore new and existing technologies, including Wise.io Support.

Why exploring new tech is effective

It’s not always a single technology, but often multiple technologies that can simultaneously provide solutions to a support team’s goals.

In Pinterest’s case, the support leader was willing to leverage tools within her current system and experiment with new solutions in order to stabilize her overloaded team. Armato explored several layers of automation solutions for a combination of tools that when used in concert, could deliver the high level of customer service pinners expected.

This progressive approach led her to incorporate four different technologies in her support technology stack:

  • ZenDesk: Pinterest uses ZenDesk as a customer-facing solution.
  • Directly: Pinterest incorporates Directly as a crowdsourced solution in order to route helpdesk tickets and community questions to their expert “pinners.” This reinforces their values of community and sharing.
  • Unbabel: When Pinterest opened their first office overseas, they considered hiring a multilingual agent to respond to incoming tickets in languages other than English. Instead, they chose to leverage another crowdsourced tool called Unbabel. This translation app works with professional translators and native speakers — bolstered by artificial intelligence — to translate, spot check, and respond to thousands of tickets a day, usually within an hour or two.
  • In-house autoresponder: Pinterest’s own autoresponder works using keyword triggers, although they hope to phase out this solution in favor of more advanced AI.
  • Wise: Incorporating Wise’s machine learning support solutions helped create customized macros, effective recommended responses, and reduced response time to improve customer satisfaction.

Measure metrics to monitor improvements

Trying new technology is only a risk if you come at it unprepared. You wouldn’t run a marathon and never measure your mileage or time your splits during training. Measuring your current statistics, setting new goals, and optimizing tools for improved results is how you’ll measure success — or how you’ll know to try something else.

In Armato’s case, her metrics included automation rate, time to response, and CSAT, in addition to her own agents’ onboarding schedules and employment satisfaction. The goal of bringing on new technology included the ability to prove ROI through metrics — all part of the overarching aim to scale support without surpassing the budget.

Armato kept a keen eye on her numbers, and watched the ROI of her unique support technology stack grow.

An unforeseen advantage of agent retention

A common misconception in the industry involves technology and agent retention. It may seem that new technologies could learn to outperform humans on the support team, effectively replacing them with machines.

Certainly, sophisticated automation technologies can do wonders in reducing workload and preventing the need to increase personnel. In addition, new technologies facilitate innovative ways of working, onboarding, and promoting employee growth with a company. Machine learning and support automation can actually improve employee satisfaction by introducing new training methods, improving team mentality, revealing customer appreciation, and reducing employee turnover."

The new way forward includes machine learning

Armato’s key message is clear: take the time to explore existing technologies and how they can help your team. Sure, it’s tempting to spend that time answering the tickets piling up in the queues. But taking an hour to discover new ways to make things easier in the future can have a big effect on tickets down the line.

Pinterest wasn’t aiming for a magic solution. Instead, the team pursued customer support strategies that combined multiple automation tools to improve the process without losing personal touch. They leveraged the tools available to them, researched new options, and took risks. This approach yielded big results, cut costs, and even helped make the customer support team happier while improving overall CSAT.

Implementing a new system can be unsettling. Weigh the risk against the reward. You might find a powerful mix of tools that will work side-by-side with your agents, giving your support team renewed control and facilitate innovation you never thought possible. Download our Case Study to learn more.

How Pinterest Leveraged Machine Learning to Transform Customer Support

Topics: Customer Support