Sub-20 percent SLAs. Agents stretched thin, morale tanking. And customer reviews piling up with complaints that no one could reach support. For multi-brand ecommerce retailer Aterian, juggling a dozen Amazon accounts alongside Shopify and Walmart, each on different platforms, the chaos was real.
Dan Yawitz, Head of Technology at Aterian, and his team faced a crossroads: either keep patching together fractured legacy systems or take a step toward consolidation and automation. With fragmented tools, endless tabs, and SOPs scattered across platforms, the experience for agents and customers was slipping fast, Yawitz told CX Today in an interview. Agent morale was low, response times were even lower, and a seasonal sales rush was looming.
Fast forward a year, and the transformation is striking. By centralizing emails, chats and phone calls into one unified platform and layering in AI-driven tools to speed up repetitive tasks, SLAs have soared past 90 percent. Handle times have dropped 25 percent, and agents are happier. Even phone support, which had been largely paused, is back online. None of it came at the expense of headcount.
How did Aterian, which counts Squatty Potty among its brands, turn things around?
The Breaking Point: When Fragmentation Becomes Crisis
The fork in the road came when poor employee experience began affecting customer service, Yawitz said.
“We knew we had a problem. Agents were burning out… and it was starting to bleed into our customers’ experience. We were starting to get reviews that said, ‘I needed a replacement for this item, but I couldn’t get customer service on the phone.’ That’s when we said, here’s the reason we need to make a change.”
Aterian’s CX ecosystem was fractured. Some replies were handled in Genesys, others in Zendesk, and phone support ran through a dialpad. SOPs were scattered across spreadsheets and manuals, forcing agents to juggle endless tabs.
“Our reply systems were fragmented… People were logging into a million tabs, tracking their SOPs and all of these,” Yawitz said.
Aterian put out a request for proposals (RFP) to vendors to help consolidate and automate its customer service operations.
“We needed to make a one-stop shop. We also wanted the ability to develop on it. Because my tech team likes getting hands-on,” Yawitz explained. “We like building things in-house.”
Aterian selected Genesys because it could consolidate email, chat and phone interactions, and give Yawitz’s team that flexibility to build.
“They had a built-in cloud platform that could be extensible and let us not only build things within their ecosystem, but also had enough open APIs that… we could carry things over.”
Aterian signed the deal in October 2023 and following a four-month implementation period, it introduced the new system in February 2024, which was the peak season for sales of its biggest product, dehumidifiers. That schedule was timed carefully to avoid disruptions, Yawitz said.
“We don’t make any big changes during peak, because everyone is just scrambling to keep up with the volume.”
The Reality Check: We Still Had a Fragmentation Problem
Early wins were promising, Yawitz said.
“We got to the season, and we saw improvements like SLAs went from 20s up to 50s and 60s, which is good.”
But the team quickly realized that fragmentation persisted, especially around order lookups, SOP access, and redundant safety flows.
During the subsequent low season, Aterian’s tech team held a series of shadowing sessions with service employees to understand the pain points in their workflows. The company doesn’t have a dedicated contact center, as its CX teams work remotely.
“So we need to make sure that people aren’t suffering in silence. We need to make sure that they can get connected with other people, like me, who can build and change processes for them,” Yawitz said.
Aterian then started rolling out a series of adjustments.
Laying the Foundation with Early Wins
The first breakthrough came from listening to agents explain how long it took to retrieve an order. Lookups often took minutes, and information from email threads, manuals, and SOPs was scattered—forcing context switching that slowed every interaction.
“That’s an easy win for us,” Yawitz said.
Yawitz’s team reallocated database resources, built a lightweight API, and initially connected it to an internal Google chatbot called Fetch. Contact center agents could then instantly pull customer and order details with a simple copy and paste action to speed up the process.
The second phase brought that same capability directly into Genesys, so that agents could access it inside their existing workflows.
“They have a workflow that is built directly into the email portal that does four things for them.”
The workflow inside Genesys now summarizes emails and customer tone, drafts replies with suggested language and empathy cues, automatically extracts order and product information and populates external databases for wrap-up.
With order retrieval solved, the team turned to a bigger time sink: context switching. Long-running email threads, technical product issues, and documentation scattered across manuals and SOPs meant agents spent too much time piecing together history.
Yawitz’s team saw the opportunity to convert the text into a format searchable with AI tools.
“We took all of our knowledge documents, all of our product manuals, all of our SOPs. We ran them through a series of LLMs with good feedback loops, converted them into FAQs,” Yawitz said. The result was an app that delivers faster answers, with less tab-hopping.
Once the system was implemented the team spent several weeks fine-tuning.
“We set it up with a series of feedback loops of different AI models, reviewing it and scoring each other’s work… going through three or four lifecycle iterations… helped us clean it for them.”
That stage was crucial to ensuring the AI-based system runs as intended. Aterian’s team put in the work to build feedback loops, test and retune models, and iteratively validate outputs against real agent workflows, rather than treating AI as a one-time deployment. As Yawitz explained:
“These are best practices in AI development to avoid hallucinations and always give answers that are good… It’s not just a question of one and done with AI… things don’t always work the first time.”
“It’s not enough just to pass your product document into ChatGPT and say, ‘turn this into FAQs,’” Yawitaz said. “It’s taking the tool but accepting it needs to do more than what is advertised as out of the box. It’s very powerful.”
Once the system is refined and customized, it also takes time to train employees on new systems and processes, Yawitz added.
From Incremental Gains to Sustained CX Performance
The adjustment work paid off.
“That’s when we started to see the most dramatic improvement in handle times and increase in our SLAs… we saw handle time drop, even during our peak season, by 25 percent and it’s held like that since the summer, which is amazing for us,” Yawitz said.
“We’ve got our SLAs from that sub-20, where it was in the fragmented era, to the 60s… Now it’s above 90, which has been essential for ecommerce.”
“In ecommerce, you need to respond within 24 hours, because people keep moving through to the next product. And they have very high expectations for the brands they interact with or buy with on Amazon.“
Rather than replacing human agents, Aterian was able to improve its customer service performance without adding to headcount.
“We never look at AI as a tool to replace what an agent is doing… it’s important to us that our customers still have that human experience.”
“We feed an AI-suggested response to every email, but the agents need to review every single one, so we have a human in the driver’s seat on every single interaction.”
Genesys has a built-in Agent Copilot tool to assist agents with pointers during voice calls that enabled Aterian to expand its service to taking phone, “which we hadn’t done for most of 2024, and we did without additional hiring,” Yawitz said.
“Customers in general are very sensitive to AI interactions. How many times have you been on the phone either stuck in a phone maze or chatting with an AI agent. Our experience and our hypothesis is that drives down NPS just to have those interactions to begin with, which is why we always want our interactions to be human-to-human.”
“But if we can speed up the human on our side to make their life less painful, we’re going to look at opportunities to do that.”
What Buyers Need from Vendors
Aterian found that strong vendor and partner support proved to be critical.
“Genesys is a very rich ecosystem. There’s a lot you can do that’s built in there, but it’s easy to get overwhelmed by everything it can do unless you have a good team walking you through what’s in the tool.”
“Having good partners has helped us work with those challenges,” Yawitz said, citing Amplix and Inflow as key consulting partners for training and uncovering deeper features.
“They help us uncover those deeper features that may be in the documentation, but you may need a hand to walk through how they work.”
From keeping 100 tabs open to a single Genesys portal, agents now work more efficiently. That has translated into increased job satisfaction. “Over this whole process, we’ve seen a big increase in our regular agent NPS scores,” Yawitz said.
“Another place where we’ve used these additional handle time reductions has been giving more time to the agents to experience other parts of our business… this has opened up opportunities for career growth for them as well.”
An element of gamification in the Genesys platform has further boosted agent morale. Tracking metrics like handle time at peak times like Black Friday or Cyber Monday earns agents rewards such as one-on-one sessions with executives to discuss career growth. “It gives them an opportunity to feel more connected to the rest of the business,” Yawitz said.
Those time savings and workforce engagement have allowed the company to launch new subscription-based product lines in recent months, Squatty Potty wipes and a skincare line, without sacrificing customer service quality.
“Anytime that you roll out a new product, you need to dedicate more time into documenting what works, what customers ask about, so that we have good SOPs in place for agents to respond to our customers,” Yawitz noted.
“We can dedicate the agents time to those new products with the time that we’ve saved via these automations across our existing product line.”
Four Strategic Lessons for CX Buyers
What can other buying teams learn from Aterian’s experience? There are four key takeaways that leaders can apply:
- Start with people, not platforms. Shadow agents, listen to frontline managers, identify real pain points.
- Respect your timeline. A four-month implementation was just the foundation; avoid changes during peak periods, and use low season to tune processes.
- Prioritize extensibility. Open APIs allowed rapid development and iterative improvements.
- Build momentum with quick wins. Solving an order lookup problem created immediate value and prepared agents for broader change.
Yawitz stressed that success requires careful governance of AI:
“The differentiator between those that succeed and those that do not are those that succeed have people either experienced with AI or tools that understand the capabilities of what AI can do and it is not a one-and-done replacement.”
“You need to narrowly—like the gutters on a bowling lane—define what you give to the AI for it to be successful. We are very prescriptive and narrow about the individual tasks we give AI to do.”
Consolidation was never the endpoint; it was the foundation for growth. Time savings were reinvested into product launches, career development, and customer experience.
By unifying channels, optimizing workflows, and applying AI responsibly, Aterian turned chaos into a platform that scales with ambition, not headcount.
