How a 20% Deflection Rate Transformed Our Support Strategy Against All Odds

OK. Uh, let me take a minute to just think through that quickly. Yeah. OK, so, uh, this was the time, uh, when I was working in, uh, at Macbellen and I was, uh, leading a team, uh. Uh, who was essentially working on, uh, driving the supportability, uh, effort within Dynamics and Power Platform Team. And what essentially it meant was, uh, uh, my team owned the front-end experience for support and as well as the agent experience as well. And the key KPIs, at least for the team was, uh, around one, reducing costs for the business and specifically support costs for the business. And second, uh, improving customer satisfaction. Uh, these were like, so from a business perspective, it was caused, from a customer perspective, it was largely around improving the customer satisfaction. So our, our goal was, uh, the, and this was, uh, uh, this was our first year in this, uh, workstream area when we were working on it. Uh, we had an aggressive goal around that by the end of the fiscal year, not only we should sort of, um, uh, build our support experience because that at that point, it didn't exist, but also, uh, uh, invest in any self-help experience so that we can achieve just through this experience, uh, uh, uh, almost a deflection rate, and I'll come to that one deflection rate. It means deflection rate of almost 20%, which is essentially 20% of the time when customers decide that they want to uh file a support ticket, how we can um leverage any automation or build a solution so that even before they submit a ticket, we can provide them self-help, which would result in essentially less support tickets for the business, ultimately cost-benefit for the business. And so this 20% saving would have resulted in, in a north of almost like uh $15 million or something just for the, for the entire fiscal year. Um, so this definitely, uh, uh, I'll, I'll be honest, like, even on day one when we were just trying to build the support experience from ground up. That, uh, none of us actually thought that this would be achievable. Uh, but, um, the, the critical moment for us, like when we sort of took the pivot in order to make it happen was, one, when we had built the initial support experience, uh, uh, I, as a leader, at least invested a lot of time in just sort of learning about how this specific challenge is being solved, um, Of improving self-help experience for customers in the rest of Microsoft. And if there was an opportunity for us to essentially leverage the technology that exists that people have already built and partner with them to accelerate our investments to reach that goal, uh, we don't need to sort of build from ground up. If there is something that we can learn from them, we'll leverage, we'll customize to our needs and essentially go ahead and deliver it. So as part of that journey, a couple of actions that I took, uh, along with my team, which is doing an assessment on what Office was doing, what Azure was doing, what are the different texts that they, uh, tech stack that they had that we could leverage from a self-help experience perspective on the back end service essentially. And is there. Some, what are the potential pain points in terms of what are the pros and cons of integrating going by Office or Azure, as well as what is the inclination of partnership with Office and Azure looking at their backlog because ultimately, if we were to take a dependency on them, we would need their support as well. Ultimately, we were able to close the sort of a partnership with Office Team. In which we took a component of their service and what exactly that was that based on the user uh description of the problem we were able to use a backend service which was uh uh a machine learning service built in partnership between Office and Microsoft Research Group, uh. It would take that as an input and provide the most relevant self-help from either from public documentation or from forums or communities. And so it, it was extremely intelligent terms of uh being able to parse the query and remove and just focus on the most important ones and render the most relevant results. And so we were able to do that partnership and, um, and we were not even aware like what kind of honestly benefit uh we, were we, would we be able to achieve our end goal through this, but we partnered with them in the initial phase itself, we got about 13% deflection rate, but we needed to get to 20%. Uh, so the 13% did happen roughly around, I would say 6 months or 6 months or 7 months down the line. And, uh, but still, we were, that itself was a big victory, to be honest, of which we were not even sort of thinking about it because uh that partnership and getting that tech uh definitely helped us accelerate to get to that point. But in order to uh close the last mile, the other thing that my team did was that, uh, they spun up a quick sort of, uh, uh, a solution which was around, um, it's, it's a bit technical and let me know if you want me to go into the detail, but we spun up a JSON-based solution wherein teams could come in, define certain rules, and author solutions which we were able to complement in addition to our partnership with Office to boost our, uh, sort of a self-help experience. Which took overall self-hel experience to 25%, and the outcome of this entire effort was that we actually overshoot our, uh, goals that we were estimating that we'll get benefit of $15 million saving. Ultimately, it was $18 million of saving in the first year itself at the end of the year, which was a big, big win, and ultimately, it was like, not only sort of, um, was individually for everyone who was working on it, it was very impactful, but from a customer perspective, That 25% of um Self-help success rate was a significant, significant win, which, and which was sort of validated through a lot of feedback from customers that uh that this all uh started asking for similar sort of a self-help experience across different parts of the product as well. And it was so, so helpful in terms of sort of achieving their goal. And we started measuring. The NPS of that experience, which ended up like in, in the north of, uh, uh, um, started with, I think, uh, on the scale of about initially the support experience was in 1 to 2 or something. It took north of 6 to 7 and it was just for that experience. So yeah, that was the outcome.

Yeah, so I think from a goal perspective, why it seemed unachievable, as I said, one was that the day this goal was sort of uh decided, day one, think of that as when that work stream was formed, and we didn't have even the support experience, like, so for uh forget about building any uh sort of a uh automated self-help experience. We didn't even have the support experience, so that was one. And second, whatever was sort of achieved through cost reduction, whatever we were able to accomplish in that one year, Office and Azure were on that journey for almost like 2 to 3 years. And so all of those, uh, what we had learned from other teams and where we were in that stage, it seemed completely unrealistic from the perspective of that we will be able to do it. Uh, neither we have the right resources, neither we have the right tech to even get there. And even from a tech perspective, what we had learned from Azure and Office seemed unrealistic to actually get to that point in the first year itself. Uh. Sorry, uh, Brandon, you're, uh, you're

I think the biggest decision that I took was, uh, one that when I was exploring about building, um, uh, Ground up versus partnering with, uh, other, uh, uh, other teams within Microsoft. I got a huge pushback from other stakeholders uh in the beginning that I think the way to accelerate the investment would be that you guys should think about, you should certainly learn from other parts of Microsoft, but probably build something ground up. Uh, because in the past they had sort of burned their hands in partnership with, uh, one of these teams, and they were not very supportive of it. But, uh, I think based on what, when I tried to learn what they had done and on the journey that they were there, I really pushed back on that what we need to do here, if we want to achieve our goals and if we really want to make an impact, we have to go ahead and partner with them because the longer run, when you think about it, when customers are reaching to Microsoft. They are not thinking about whether they're reaching out to Office or Dynamics or power platform. They're reaching out to Microsoft. So long term, probably having a consistent support experience across entire Microsoft, if we have to think about it, if we can reach there, that should be the goal. And if that's where we have to reach, the partnership between all these different, uh, cloud services is extremely critical, both in terms of front end and the back-end experiences. So this is the first step we can take right away. And so everyone was very skeptical about that in the first one. OK.