How a Product Manager Pivoted Under Pressure to Achieve a 95% Success Rate

Yup, um, just a clarifying question, is it, is it really have to be about new data or it can be about changing the business process requirements?

new data. Yeah, yeah, never happens to me. No, I'm just kidding, um. So this, this, this I would say um happened. In 2021. Um, so one of the products I'm managing now is, um, we call it AI coach, but what it's really is about, um, Taking some of the burden away from the retail store managers on coaching their employees for performance improvement and goal achievement and trying to use data and analytics and, and later on we have a vision to use some, you know, um, computer vision and other things as well in the store. Uh, so, so I started this project, um, for our retail employees and our goal was that we will. Improve, um, our, our like so right now I would say 80 85% of our employees meet their quota target. The goal for this project was we'll take that to 95%. So 90 help our 95% of employees to meet their quota target. That was the goal for the, for the coaching product, 90%, so 9095 + 5%, yes, uh, uh. So we started working on this. We are working on some You know, hypothesis of what causes somebody to miss their quota. Can we, you know, look at some data to validate those things? I would say we spent 2 months almost looking at the data validation and some, you know, talking to reps, talking to managers, uh, looking at the existing performance data, looking at the coaching tools and looking at the output of those coaching sessions also, just to see, you know, how we're going to approach this, this thing. And my team start building some, you know, basic modeling around first, um, let's figure out, um, you know, what are the leading indicators that somebody is, is not on track, right, that that's for to meet their quota target, um. During that time, um, business made a decision that we will move away from individual quota. Um, and we will have a quota at the store level, and the reason for that I can understand is, um, in order if you want to set up a quota for some employee, it, it gets into detail the details, but I think it's important detail. It's like you have to manage the shifts that way that everybody gets a fair shift, all right, so then you lose the flexibility of scheduling. And all, and because we were not having enough labor, I would like everybody else, we were struggling to hire people. The management decided that we'll move away from dual quota. We can have more flexibility in scheduling, uh, and we'll have a quota at the store level, um, but, um, we still wanted to help our employees kind of. You know, uh, help the store to get the quota. So the problem shifted from employees to store, but store doesn't have anybody. Store only has employees. So for us was how can we, how can we help, um, you know, with the new business problem and continue on what we're doing. And I think the answer was that rather than trying to meet the quota for an employee, let's figure out. What would be their potential performance, right? It, it, it doesn't need to be somebody establishes for them, but it's something we can predict confidently using the data. So the, the thing was that if you are, you know, just stuck with the bad shifts, let's say during the afternoon when nobody shows up in the stores, um, your, your performance. You know, optimal performance will be different than somebody who's in the morning. So we kind of build another layer in before, uh, what we wanted to do. We build a layer to say that we predict quota for you, right? It's not quota, but it's like pseudo quota, and then we kind of continue on the, on the path to say how can we help you meet that particular target. So it was kind of uh change the direction, um, I would say required us to say rather than trying to To meet one particular goal, let's come up with a process to establish a reasonable goal first and then continue in the process of meeting the goal.

Yeah. So, so. The first thing was like when we, when I heard that, you know, they're they're moving on the quota, I knew that we can't, I mean the project we were starting with can continue with the same goal. The goal has changed. The goal has shifted from that rather than an employee meeting their quota, let's help a store meeting their quota. But then I kind of thought about like, you know, there's nothing I can, I can provide a lot of insight at the store level. But there's no action I can take at this so level, right? So your information is as good as what people can take action on. So I, I can say, OK, we have to stay at the employee level, but let's see how we can shift the objective and shift our approach that the employee level thing rolls up and kind of, you know, uh, aggregates at the the store level. So that was like just a thought process in terms of actual data looking at and, and so we, we looked at the operational data to say if. Somebody is like, you know, so we started, we started kind of predicting the performance by shifts and then say, you know, if look at the schedule and if you're on this shift, that's your optimal performance. So we didn't, we initially we were going with an employee, um, fixed like having Brandon in mind, like, you know, what is the quota for Brandon and what's how we can help him meet that. Then we said, OK, what's the quota for an afternoon shift? And if Brandon gets afternoon shift. What would be his. So I think that was kind of a little bit of, um, you know, building uh additional data set as well as kind of just introducing a new variable in the equation.

So, so the biggest obstacle was that that business was like not giving us any additional time. They still wanted to kind of, you know, keep the same year-end objective, uh, for in terms of performance improvement, but just like shifted from employee to store. So I had to kind of, um, you know. One, look for creative ways um to see how we can build this, you know, additional modeling capabilities, and, and, and second, I had to just, you know, take it back to my, my leadership and say, look, business has shifted direction. We understand where they're going, we support where they're going. Uh, but this requires more work for us, and, and we will need your support and, and some additional resources. So I was able to get like, you know, additional data scientists to build this additional layer, um, and that's how I was able to do. It was, um, it was not a tough sell, but because I think we had a very good data to back us up and able to justify and also articulate, you know, our approach, in a very clear way. OK