Hadas Drutman on Customer Service Philosophy: Why Bobbie Wants Parents to Reach Out More

Customer experience is a brand experience

Lisa (0:02) Welcome, Hadas. I’m very excited to chat with you today. I would love to jump straight into it, and I would love if you can introduce yourself and tell us about your role at Bobbie.

Hadas (0:13) So I’m Hadas. I’m Director of Customer Experience at Bobbie. I oversee our internal team, working externally as well, thinking about the customer experience holistically, and thinking about one-on-one interactions and what that infrastructure looks like. Previously, I’ve been consulting in the customer experience and retail experience space, and previously was VP of Member Experience at Seed. Previous to that, I was Head of Retail Experience at Glossier.

Lisa (0:48) And also, I just found out that you are half Romanian.

Hadas (0:50) I am. Which, I mean, actually, you reminded me, because I knew this for a while, which is very funny.

Lisa (0:57) What’s your philosophy on customer experience?

Hadas (1:01) Customer experience ultimately is a brand experience. So it’s your opportunity to make an impact, an impression on a customer, and allow the brand to come through and add value to a customer’s experience. I don’t like calling agents agents, because they’re true specialists that are able to educate the customer, that are able to build relationships, that are able to really show up humanly and empathetically and support a customer, sometimes in a certain emotional state, and find solutions.

I think education plays a huge role in customer experience and how we add value, ultimately. And if you’re showing up genuinely and you’re really putting yourselves in the customer’s shoes, they feel that. And to understand that there’s someone on the other side that truly cares has a lot of impact for the business: feeling that there’s somebody that is super dedicated to supporting you and finding a solution. And sometimes it’s not the easy response of, like, “where is my order.”

This is from a customer-facing perspective, but from a business perspective, a customer experience team has a lot of weight that it carries, and a lot of insights that need to become accessible to the rest of the business. So we play a huge role in that as well.

Why Bobbie wants customer contact to go up

Lisa (2:59) I have recently been on a few demos with other brands, and I’ve heard a lot, “Oh, I really admire Bobbie’s customer experience, and that’s where we want to take ours.” So hearing that from you makes so much sense when it comes to AI. Obviously, one of the reasons why I wanted to invite you to share the story is because you’ve been making incredible progress with AI and Siena. And I feel like most brands want AI to take people out of the conversation, and you’ve always said that you want customer contact to go up. Why is that?

Hadas (3:37) The opportunity to influence and show up for customers has a huge impact on their lifetime value, and their trust, and their attention. I want a customer that is standing in the formula aisle at Target to be like, “Oh, let me reach out to Bobbie so they can support me.” So we have that ability to influence their decision through education, but also support them in being their support system as they’re navigating a lot of decisions that are really heavy, and sometimes challenging and confusing. And be that resource that arms them with information that allows them to make the choices that are right for them in those moments. And that has such a huge impact on the customer.

So obviously, I want contacts like “where’s my order” or damage inquiries to go down, but inquiries that allow us to show up as Bobbie and to educate and support are the ones that we want growing, because deflection, in my mind, should never be the goal for CX, if it’s really engraved in the company’s DNA. Because your ability to be closer to your customer and show the brand through is really meaningful in those one-on-one moments. There’s a lot that you can do in one-to-mass, but one-on-one is really when you can show up as a true differentiator.

Breaking through a 19% automation ceiling

Lisa (5:33) If we dive into your AI journey and usage of AI or automation tools, can you share a little bit about your past experience before you found Siena, and then what convinced you to try Siena and give it a shot? Because I think it’s been only three and a half months since we started working together.

Hadas (5:55) Yeah. A year prior to that, we were working with a different tool. It’s been kind of a headache for the team, I would say. Something that was very hard to manage, very hard to optimize, very hard to scale. And we ultimately hit a ceiling of automation. It was very, very hard to go past 19% automation. My team wasn’t able to move the needle, and wasn’t able to do that independently. We needed the company’s support for any small thing. There wasn’t any self-serve. It became a weight on our team’s day-to-day instead of something that allows us to elevate.

So closer to the end of the contract, we started to evaluate, and we knew we could do better. But this is all that we implemented previously. It was hard to think about all the time that was invested in implementing, and in some ways trashing that and moving forward, because we’d need to do that from scratch.

We were able to triple our automation rate within a few weeks. So we had a very insane deployment and implementation phase, and automation rate was just a piece of it. We were able to do so much more instantly with Siena, including building a full QA program that made such an impact for us: reviewing 100% of tickets, both human and AI. We previously did that manually for our human tickets and were barely able to get a 20% sample. So that made an instant impact, and also we were able to cut BPO headcount. So it was like, okay, here’s an instant 30K in savings annually. And we were able to really make that program engraved in how we operate. So coaching became more meaningful. The ability to level up on the BPO end and the internal team end was really, really meaningful.

From a pure automation and Siena work perspective, I think the ability to self-serve and tweak as we go was a game changer. Also the ability to work with the deployment team that understood so well and were invested in our success. They understood our expectations so well, knew the ins and outs of our queue, but also were super invested in our realistic growth and realistic progress.

Why Siena needs an owner on the team

Hadas (9:23) A caveat to that is, I think our ability to succeed and triple automation rate and do all these things in such a short period of time, at the heart of it, was the understanding that we need to have a dedicated person internally for that. Britt on my team is our Siena and AI implementation and enablement guru. Investing in her, and basically clearing her time to allow her to learn and also do all the things, and building the journeys, and investing in that is what made us succeed.

It’s such a common thing to be like, “Oh, you just need to give it some documents and it will happen,” or, “You just need to deploy, and then the tool manages itself.” And that’s such a wrong assumption, because the success of an AI tool really depends on what you put into it. So if you invest so much, you get the results. And I think realizing that at the beginning of the journey and dedicating Britt to that work, and also, once out of deployment, we revisited her role: okay, what does this look like longer term, and how should you continue playing this dedicated role? You’re now an owner of a channel, almost, or an owner of a tier, right? I have a manager on my team for our BPO, and I have a manager for our internal specialists. Siena needs an owner as well, because it’s playing as its own tier and channel.

Lisa (11:33) The success of agents depends on the people and the deployment. We are at this point where technology is in a really good spot. Models are really amazing. I haven’t seen a hallucination in, I think, 12 months, to be honest. So it’s really about the people, and I really like how you invested and realized that she needs that time and space to first understand, because the platform can be very complex. There’s so many things that you can do. And then make it successful. And I like how you described her transition. I think we are all going towards becoming AI orchestrators, and we’re going to work with our little team of AI agents. So you do need to master these tools, because that’s how you’ll get to the right outcomes. How much time did she put into working with our team?

Hadas (12:33) It was a full-time job. What was great was that we already had things in place. We didn’t need to make up our brand voice. We had a very distinct brand voice that we had to tweak for the purpose of Siena and building our agent, but these are things we didn’t need to build from scratch. We had a very, very specific vision on how we want the agent to respond, how our brand shows up in these moments. So that wasn’t something we had to build. Utilizing existing materials and our philosophy, and bringing it to the day-to-day work, was ultimately what we had to do. And then building the journeys. It was a full-time job. I think it was two or three weeks that it took us to get to 55% automation rate. So obviously, setting a tight deadline and committing to it from both ends, right?

Lisa (13:50) Was there anything else that you think really made that happen?

Hadas (13:54) I think your team, the Siena team. We had weekly sessions that Britt was leading with them, and they were very productive. It was step by step by step, working towards a very clear path that we outlined together. So that support, and the ability to self-serve, the tool being so intuitive, utilizing Ask Siena and making that a big part of our implementation phase as well.

Lisa (14:41) Can you share a little bit about how you’re using Ask Siena?

Hadas (14:42) We’re using it for anything, if it’s to build a new journey, to tweak existing journeys based on business needs, to pull different reports or different looks on something. Siena is ingesting all of our tickets. There’s so much insight that we can extract from that. From a building perspective, Britt is utilizing Siena all the time. From an insights perspective, I think we’re still trying to figure out how it can support us day-to-day, but my goal is that every person on the team, or at least every manager on the team, utilizes it for different business needs.

You mentioned Alyssa’s work. She was working on some proactive work and trying to pull insights around cancellation, and it’s easy to just ask Siena to understand when people are reaching out to us for a certain thing. So there is just so much information that it holds. I’m excited to think about more ways to engrave it in how we operate.

Lisa (16:04) Not sure if you’ve already integrated Slack, but if you integrate it with Siena, then anyone on your team can just tag Siena on Slack and ask questions, so they don’t need to log in. So we can look into that to connect.

Hadas (16:17) I’m excited. That’s really cool.

Refunds at 3 a.m. and the speed of service

Lisa (16:20) Another thing that I know you are doing, that sometimes is quite scary for other customers, is you have let Siena handle refunds up to, I think, $300. How are you thinking about these more sensitive edge cases, like refunds, or these more sensitive transactional scenarios?

Hadas (16:42) Siena is an agent, right? We are giving it the same authority that we’re giving our BPO team, and that’s just realizing that it’s an agent on your team that needs to be treated equally in many ways, right? So our offshore team is getting the same agency and ability to do these types of refunds. Why would Siena not? The benefit of doing that is way higher than the risk of doing the wrong refund. I think it’s

Lisa (17:21) Yeah, you’re right.

Hadas (17:22) the ability to be of service to our customers.

Lisa (17:26) And so fast.

Hadas (17:27) So fast, at 3 a.m.

Lisa (17:29) And do that in 17 seconds.

Hadas (17:32) Yeah. There was a mom that reached out that had realized that she never canceled her subscription when her baby graduated from formula. And she reached out at 3 a.m., because she was up, obviously. And in 17 seconds, Siena issued a refund. It didn’t just serve as, okay, this was really fast and really incredible and on point, but it also served as a surprise-and-delight moment for that customer, because she did not expect someone to follow up within 17 seconds. And I think that allows us to meet our customers where they are, fully, and when they need that support.

For many years, I’ve been thinking about quality of service and what that entails, and I think speed is a huge part of that. A customer that gets a response instantly, while they still think about it, and not just, “Oh, I sent that message and forgot about it,” and you’re able to resolve it and not add to their mental load, is huge. Especially now: our patience is running thin. At least I’m very impatient. I need things solved on the spot. And our attention span is so low, right?

One voice from AI to human

Lisa (19:10) This is amazing. Another thing that I know you really care a lot about is the human-to-AI collaboration and the handoff. How do you design the handoff?

Hadas (19:22) It’s very seamless for us, because we are not treating the AI agent as an AI agent, if that makes sense. We have invested a lot to make sure that there isn’t a gap in tone. So our AI agent needs to sound exactly like our human agent. That makes the transition way smoother. So it’s ultimately investing in that tone and voice of Siena, and knowing that they would respond the way a human agent would, and then once you transition, it’s not a clunky experience.

Lisa (20:08) That makes sense. And in terms of channels, what channels do you have Siena on?

Hadas (20:13) Right now, we are on email, SMS, and now social. SMS and email were critical for us to launch with. They are basically at 50/50: both of them are taking the same piece of the volume pie. And social is a service channel for us, so we’re treating it like any other channel, in DMs. We started implementing it pretty recently.

Measuring effort and holding AI to the same standard

Lisa (20:51) Beyond the automation rate, what are your other success metrics? What are you looking at on a weekly or monthly basis?

Hadas (21:02) We are revisiting what those should look like. I think as the center of gravity for CX is shifting, it’s really important to evolve those. But the CXP, what’s the

Lisa (21:16) Customer experience score?

Hadas (21:17) Yes, the customer experience score is something we just started looking at, with Siena helping us. I think it’s a really important score. Before that, we weren’t able to extract sentiment so easily.

Lisa (21:39) Effort, sentiment, satisfaction.

Hadas (21:41) Customer effort score.

Lisa (21:42) More than just CSAT.

Hadas (21:44) Yeah. I don’t care much about CSAT, because nobody really fills those. The response rate is so low for both CSAT and NPS that I think those scores are becoming really... The CXP is something we look more at. The effort score is interesting, because now there’s the ability to measure how hard the customer had to work in order to get the response they needed. And that’s meaningful. I think response time will always be top priority, but I think the KPIs are shifting in many ways, and it’s cool to see how Siena is able to support us in reimagining KPIs.

Lisa (22:31) Even myself, I don’t really rate. Maybe you would rate if you are unhappy. But if you’re happy, or the experience was fine, you’re not going to, so you miss out on so much data. Speaking of QA, I’m curious if humans and AI are held to the same standards. Can you share some of the scorecards that you’ve created there?

Hadas (22:56) They are held to the same standards. We have different standards across tiers, so different expectations from our tier one, which is mostly our BPO and Siena, and tier two and tier three, which are internal. There is an element of empathy, and answering to the point. And accuracy, especially when it’s transactional, accuracy is so important. Not adding too much

Lisa (23:32) Filler words?

Hadas (23:34) Exactly, the filler.

Lisa (23:36) The fluff.

Hadas (23:39) Being straight to the point, that is super important. In the more transactional moments, it’s critical, but also: how do you show up in those transactional moments in a really empathetic, caring way? So I think that’s a really interesting thing that we’re looking at for Siena and our external team. For internal, it’s a bit different, because there’s more meat to work with, more complexity. But I think the human part, the ability to invest in investigating the details, is so important to us.

Empathy is investigation

Lisa (24:26) Yeah, that’s amazing. And how has your team changed in terms of responsibilities, since AI has been taking half of the volume now?

Hadas (24:40) It’s still a work in progress, and it’s an evolution that we’re working on and will continue to work on. But I think, A, we are able to hit our SLAs in a very comfortable way across tiers. So, massive improvement there. Our human agents are able to be faster and focus on tickets that matter more, and don’t have the distraction of the WISMOs, of the cancellations, of all these manual things that take them away from the quality of the ticket. And we are able to do that faster. So SLAs have improved. It’s incredible to see our performance beyond SLA, especially for the BPO team.

The team is able to focus more on proactive outreach. So really understanding the friction points, understanding where we can show up for customers. If it’s customers with duplicate subscriptions, we now have the mental capacity on the team, and also the time, to dedicate effort towards that. We’re doing a lot of that, and the impact of these initiatives is incredible: to see how customers feel cared for, that somebody actually took the mental load of realizing that they have two subscriptions and instead just told them, “We canceled one.” Parents are in the thick of it.

Lisa (26:27) There’s so much going on. They probably so much appreciated it.

Hadas (26:32) It’s incredible to see the feedback that we’re getting, of just being able to show up for them in these moments. So it may seem small. And it’s also not something a business would do, like, “Hey, cancel your subscription because you have two.”

Lisa (26:51) Most never.

Hadas (26:52) Yeah. But it’s something that we really invest in. And the ability to do that is because Siena is taking a lot of the load, and we’re able to invest more in the more complex tickets. Not just spend a minute, but really go deep. There are moments where we need to find really specific solutions for a situation. I think empathy is not tone. Empathy is truly being an investigator: being able to compile all the information, all the pieces, which are in many places, not just in the thread, to understand what your customer is experiencing, and then figure out a solution that is for that specific case. The ability to do that without rushing, without the need to immediately close a ticket, is really meaningful.

And also utilizing even our BPO team for different things. We need to log complaints in a very regimented process. It needs to be very buttoned up, and our internal team used to do it. Now our BPO team has capacity, so they’re doing it, and now our internal team is actually able to focus on those complaints and really invest in recovery, and what that looks like for each customer. So there’s a big shuffle of responsibilities, and it will continue.

Lisa (28:31) The fact that now you have context, the memory, the intelligence, the customer intelligence so easily, is basically what sits behind that empathy that you’re talking about: putting yourselves in your customer’s shoes, knowing the customer better than anyone else. That is such a game changer that AI provides. That’s why we’re also excited for your inbox.

Hadas (28:59) I’m very excited.

Advice for brands choosing AI, and what’s next

Lisa (29:01) What would you tell another brand, whether it’s their CX leader or founding team, that is looking to transform their customer experience with AI, but maybe they’re also building a pretty sensitive and regulated brand? What would you advise them?

Hadas (29:30) A tool that really gets it. What stood out with Siena for us is you have built an empathetic agent since day one. Your ability to see what a consumer brand that cares deeply needs in these moments, and over time, is to me what really stands out as the differentiator. It’s not just a tool, it’s also a philosophy, and sharing that philosophy is really important, because then you know that the tool will evolve with you and those values will be kept at scale. So I think it’s really important to figure out what your needs are and how to bring it to life over time. It’s not just, okay, this tool is cheaper and this tool is more expensive. For an AI tool to work for you, it needs to do way more than just respond to your tickets. So I think the ability to evolve is critical.

Think about solving long term and not short term. Okay, you have an inbox that is unmanageable now. Sure, you can bring in an AI agent, but how do you build something to last and to have a long-lasting impact on your organization? That is the framework I would prefer. And also be really clear on what you want out of it, and what alignment you need to have with your human team, and what you want the human team to be able to do in order for that tool to enable that. Because we talked about contact rate, and me wanting to increase contact rate, as a philosophy. The reason why I’m able to even say that is because I’m very confident in the infrastructure I have, because we have Siena. And that enables us to have that confidence, and to be like, okay, you know what, we can have way more tickets coming in, and we can handle it.

And the constant shuffle is something that I’m not going to stop doing. Okay, we have Siena, it’s working, three months in. The shuffle will need to continue happening as the business evolves, as the team evolves, as the volume evolves. But looking at it holistically, and choosing a tool that has that holistic point of view as well, is really important.

Lisa (32:33) Well, I really appreciate the input here. I really love what you just said. I feel like when you think about AI, you think everything’s becoming more robotic, very transactional, you lose the human touch. But I feel like through AI and intelligence, we’re actually making the internet more human, because you can create these really unique experiences with all the benefits that we just talked about. And we are seeing this every single day, through moments like the 3 a.m. story that you shared. My last question is: where do you want to take the customer experience at Bobbie in the next 12 months?

Hadas (33:15) Thinking about audiences is really top of mind for me. How to diversify that, and how...

Lisa (33:21) Segmentation?

Hadas (33:24) Yes. And also, how can we cater to Spanish-speaking audiences, and how can we support that at scale? I don’t think we have scratched the surface of that yet. Thinking about services in general, and deepening my team’s expertise from an educational perspective, true feeding expertise, is something we think a lot about and constantly want to develop. We have a human team, and my goal is to only grow it. How can we continue adding value to our customers through education, through feeding support? What that looks like is really cool to envision, and developing that expertise together.

Voice of the customer is becoming a big pillar for us, and something I want to continue growing, because our ability to utilize insights and data to support the business and to inform business decisions is so critical for us to continue growing, staying true to our DNA, and staying close to the customer. So, connecting the dots better in the next year, and informing growth, informing innovation, and supporting the business, is a really cool role to play.

Lisa (35:01) I was talking to Amanda, who is a CX manager at Halfdays, and she described CX as the heartbeat of the company. So that’s basically what you’re investing in as well.

Hadas (35:13) Yeah.

Lisa (35:14) Which is great.

Hadas (35:15) Yeah, definitely.

Lisa (35:16) Thank you so much.

Hadas (35:16) Thank you.