Bill Bentley | Finding the Tipping Point
The past few episodes have talked a lot about the importance of mindsets in engaging employees and aligning with customers. Today we are going deeper into how data works in the feedback loop that make the job of sales easier.
Before you click to the next email in your inbox, here are a couple of important takeaways:
· There is frequently a tipping point in the qualification of a prospect – above it, near certainty it is a good opportunity, below it there is little chance you will land it
· Opinions can be quantified
Sharing this episode with your process improvement or sales operations people will help them look at how they can use data more effectively for your team.
Bill Bentley is a data guy. He is an engineer who became a general manager, then helped salespeople learn to sell to senior managers. He’s led an engineering firm and a software company then started a Six Sigma training company. He brings a well-rounded perspective to the challenges of sales.
Listen now
Mentioned in This Episode: www.value-train.com
Michael Webb: B2B sales and marketing works to find the highest quality prospects, reach decision-makers and sell value. Operational excellence uses data and systems thinking to make changes that cause improvement and eliminate waste. My name is Michael Webb and this is the Sales Process Excellence podcast. In the next 30 to 40 minutes, we’re going to destroy the myth that these two groups conflict and show you how to bring both strategies together to create more wealth for your company and your customers.
Hello, this is Michael Webb and I am excited to introduce you today to someone I have known for many years and who has brought an amazing insight not just to business but to measurement problems and statistical analysis. Bill Bentley, welcome here.
Bill Bentley: Thanks Mike. Glad to be here.
Michael Webb: You just have this most fascinating background. I want to kind of walk people through it. You got an engineering degree at Rensselaer Polytechnic and then you began as an automation engineer for Procter and Gamble and Frito-Lay, right?
Bill Bentley: Right. I was at Procter and Gamble for quite a while. Then went to Frito-Lay from there.
Michael Webb: As a manager at Frito-Lay, right?
Bill Bentley: Right.
Michael Webb: And then when we met, you had been hired by Rockwell Automation. Tell us about why they hired you.
Bill Bentley: They hired me because I had just lost a job during a downturn. I was in charge of all the industrial automation and electrical engineering at Nabisco and when KKR bought them, they kind of dissolved the company, broke it all up, and Rockwell was a big supplier to us and they called and offered me a job in their sales force to help the sales teams sell to people like me or people who had jobs like the one that I had just left. So it was very interesting. My first reaction to that job offer was actually, “No, I’m not interested in that. I spent my whole life avoiding salespeople. Why would I want to be one?” But after a few weeks, realizing that I was unlikely to get any other work, I called back and changed my mind and that turned out to be a very interesting job and lot of fun and I enjoyed it.
Michael Webb: And likewise, I was hired by them in around, trying to remember, I think it was ’81 or ’82, not at the level where you are at. I wasn’t an executive, but I was a sales manager and I was brought in to help their large vaunted sales force of degreed electrical engineers to make this shift. The president of the company, his name will come to me in a moment, was trying to turn Rockwell from a manufacturer of industrial controls sold in brown boxes through industrial distribution, limit switches, push buttons, PLCs, stuff like that. And he wanted, instead of selling the pieces in parts, he wanted to be able to sell integrated systems with all the margin covering the labor and all that, he wanted them to be able to sell solutions. And so I was in the field sales force in the St Louis branch and I met you as one of the guys at corporate who was helping us.
Rockwell had made it’s success in the automotive industry. It was a way to automate and better control systems of relays. And now they were trying to move from that on off digital kind of manufacturing environment, into a more higher level control system environment, especially with continuous controls. So that was a fascinating, fascinating time. We worked on, and I think you were involved in, we made the first proposal to a large brewery, Anheuser Busch in St Louis, to use digital automation controls for what had always been controlled in the past by a distributed control system, which was a continuous analog factory control. That was a $20 million project that we bid, we proposed, you remember working on that Bill?
Bill Bentley: I do remember it and I remember the sales guys make a whole lot of money, maybe I should do that rather than be an engineer.
Michael Webb: Yes indeed. Yes indeed. So from Rockwell, after that, you were a general manager, senior executive at least one engineering firm, right?
Bill Bentley: I was, I was general manager of a contract engineering firm in Tennessee and we were actually a captive engineering group for a Procter and Gamble plant, a very large Procter and Gamble manufacturing plant. So I had about a hundred engineers and designers and programmers working for me at that plant.
Michael Webb: And then when that tapped out, a lot of us had some difficulties after 9/11, the economy kind of went South and then there was the financial crisis in 2008, so you founded a successful Six Sigma training company for yourself. Tell us about that.
Bill Bentley: After that stint, I came to Atlanta to be the CEO of a software company and that’s the company that folded when 9/11 hit. So I worked again. And I did start my own consulting business and training business and that business lasted for 18 years. It’s still going at a low level, but 18 years is the longest I’ve ever spent in any company. And it was my own company. So I was happily self-employed and I loved it. It was quite interesting. It started as a community service project to teach people who are out of work, how to do Six Sigma and ended up being a national training and consulting business.
Michael Webb: Which is a fantastic outcome. Not many people would be able to make all of these transitions in their careers and then a few years ago I was surprised, shocked and interested to see what you did next. Tell us what you did.
Bill Bentley: After 18 years of doing what I was doing, I started to get itchy feet, it started to become routine and I could see that business going down hill slowly. So I actually went back to grad school. I’ve got a new degree, a masters in applied statistics. I picked a bunch of other courses online and declared myself a data scientist and now I had the credentials to back it up. Went looking for a job and I took a full time job as a data scientist for a very large company here in Atlanta.
Michael Webb: And that’s fantastic. I don’t know, just because you had a passion, you really like the way, the analytical part and the creativity involved in the mathematics of statistics, if I understand that right.
Bill Bentley: I did. Right. And what drove me to do that and choose that is sort of my retirement job, I guess you could call it because the stuff that was the most interesting to me was the analytics stuff, working with numbers and teasing information out of them. A lot of which I did for you when you and I worked, did projects with your clients, I chose to pick that as my next field.
Michael Webb: Right. Super and so very cool. So let’s get into that. I want to describe the projects. There were several of them that we worked on. I would get hired, this was after or at least around the time, right before and after, Sales and Marketing the Six Sigma Way was published and I would get hired by an executive, senior executive of a company, sometimes a big company, sometimes a small company. The first thing we would need to do is to figure out, actually like the second thing we would need to do but is to figure out, okay, what makes a qualified prospect, let’s define what that is. And I started trying to do that using Likert scale questions, with a one to five scale. And so I took the sales force through some facilitated exercise to define these questions and had them gather data according to it.
And so they would have a question. And for example, a coach network. A coach for a given sales opportunity, a coach is somebody inside the customer’s business who wins if you win, there’s something good for them if you win and they are in a position to help you win, they can tell you what people are thinking, they can tell you how to position yourself so that it better communicates your value. And so in any given account, you might have, well I don’t know who in the business is a coach and that would return a zero score or at the other end of the scale, the decision maker, I know them and I know their coach. So that would give you a five and then you can identify observable things in the middle. Several different ones. I know someone, but I don’t know if they a coach or I do have a coach, but they’re not in a position of influence. I do have several coaches, but I’m not sure if they can influence it very well or they can influence it.
Anyway, all these different gradiations and I remember what you told me when I was going to bring you data using a Likert scale. What did you tell me?
Bill Bentley: I told you that in those cases, Likert scale data is useless. That what it really is, is people’s opinions which are described in words and which we, for various purposes, convert them to numbers and just like you can’t find the average of a cow, a pig and a horse, you can’t find the average of a one and two and three when they represent opinions instead of actual measured data.
Michael Webb: Right. And so I talked you into proceeding ahead anyway. And the results tell us, I mean I thought the results were spectacular. Tell us what you saw in the data.
Bill Bentley: You had an advantage, while we worked on this together, we were able to do something that most people who work with Likert scale data can’t do. And that is you had the ability to train the people who were doing the scoring. So it was the same people over and over again and the sales people who were scoring clients. And we got the data for our model, looking at the things they had sold in the past year, I think. So you were able to give them some basis for choosing whether it’s one, two, three, four, five and you have to train them. So because you could train them, the quality of that data was much better than it would have been had you, for instance, been collecting data through a blind survey and sending surveys out to somebody who would only answer it once and that’s the worst kind of Likert data.
So you had the best kind of the Likert data. And the result was that we were able to make a statistical model that predicted the probability that that company would buy from you, given the answers to certain questions. And it was a very good a model with very good high reliability and had good ability to predict. So I was as surprised as you were that it came off that well.
Michael Webb: Well I wasn’t as surprised. I knew that there was some value in that data. I didn’t know what was going to look alike. But way back in the beginning when I first went independent from my sales training job, my first client was a fellow from some work I had done with the sales training company and he was a regional sales vice president for Marriott Corporation. This was before I started working with you on these statistics and I got them to … I didn’t know exactly what I was doing, I was helping them operationally define, after a fashion, what made a high quality account for them because they were selling repeating relationships. And they came up with, my goodness, they went crazy in this exercise and it took a day and a half, most of which was on the phone with their teams because they had, I forget how many, 20 or 30 sales people and they came up with 37 questions and they collected the data and I had another statistician do it and he was able to find some profound tendencies in the data that was actually quite helpful to Marriott.
I remember my wife was very nervous about me going independent and when I got the $20,000 check to pay for this engagement, I made her take it to the bank so that she would say, “Yep, you can do this.” And so it was several years later before we landed this business, when the first one was with a small software company called Replicon and they had about nine salespeople, I think. And the sales manager was very skeptical. But when the data, the results started coming in and he started realizing how accurate this was as a forecast indicator, he was quite impressed. Then there was another client about a year later, this was right after Sales and Marketing the Six Sigma Way was published, a company called Maquet and they had I think 36 salespeople at the time, M-A-Q-U-E-T. They made a very high end, very sophisticated set of equipment for healthcare, for operating rooms especially.
One of them was a respirator, the machine that breathes for you if you can’t breathe. And so we got them to define a little bit better this time, the observable characteristics of their sales opportunities and we gathered a couple of hundred, the number 240 rings to mind, to my mind of instances of sales opportunities across the United States over a couple of year period of time. Actually, we let them go historically to gather that data. One of the first things that came out of it, I don’t know if you remember, we did a pareto chart of the salespeople by close ratio and it was pretty even, but the person who had the lowest close ratio, you remember what you told me about him?
Bill Bentley: Yeah, I think, as I recall, I said, “If he worked for me, I’d fire him.”
Michael Webb: Yeah, that’s right. And I didn’t know who the fellow was, hadn’t met him, but we told the client we’re going to be open about this stuff. And we put that chart up there with the bar chart, with the lowest close ratio and the sales people’s names were on it. And I remember there was silence in the room and then there was some sniggling and shuffling of feet and a little bit of tittering and chuckling and I’m just going, “So tell me about this. What are you thinking?” It turned out that the guy with the lowest close ratio was a salesperson who had the highest performance, a sales guy by a large margin and one of the reasons was that he really truly bought into the mission of the CRM system. So if anything smelled like an opportunity, he would put it in that system.
Most of the other sales guys, we’re not that way. The sales VP had engineered it so you could not get a quote unless you entered it into the software. And most of the other people simply entered in a deal once they had to quote it, then they would be forced to enter it into the system. But the top sales guy was drinking the Koolaid and he would enter all of his deals and so he had the lowest close ratio. And I thought, “Wow, is that eye-opening about measurement systems in sales forces?”
Bill Bentley: It was pretty interesting and I do remember that. I think I discovered at that time, it was that when I looked at the data that each of the salesman had collected, I was able to determine that two of them had faked the answers.
Michael Webb: Yes. You said these two guys are not in the same universe as the others and you told me who they were and the day before you told me that, I had been on the phone with the sales VP and he was telling me about some circumstances where they’d had to change their comp plan and some of the sales people were really, really upset about this. And it was those two guys. So they were trying to queer the data and you could tell. I thought that was another astonishing outcome of measuring the data this way. And then a story I’ve told many times, some of the questions were way more predictive than others. I think there was like eight or nine, remember that?
Bill Bentley: In terms of what we call in the data science business, which are statistically relevant predictors, we only had something like eight or nine out of the 30 or so things that you had them collect data on. So in effect, they could not bother collecting data on the others. And I think at the time while I said you can forget collecting that data, but the head of that company didn’t want to do that because he didn’t want the sales people to know which of the factors were the ones that mattered.
Michael Webb: And then there was another crucial insight that came from this data, remember it?
Bill Bentley: From that we were able to make a graph in effect, which showed the probability that you would close the deal given the values of the data that you had. So it told them which jobs to go after and which ones to stay away from. Don’t waste your time on those. And if they had some extra time and if the company would let them actually look at this data, it would show them why the score was low, which factors were driving the probability low and if those were factors that they had some influence over, if they could do something about it and it gave them some things to do. So your extra time, go after these low probability deals. But the model told them what to do in chasing these low probability deals to make them better.
Michael Webb: Yes, it was all traceable. It was all traceable down, so they could tell and I called it the tipping point after that because it wasn’t a gradual increase in probability. Those are pretty sharp increase in the range of the scores. Say the lowest score on the qualification assessment was 30 or something, 28 or something and the highest possible score might’ve been 120. Well there’s some range, perhaps between 62 and 81 where below that was almost no chance of closing, above it was almost a hundred percent chance of closing. And so as you said, salespeople could walk away. But what they ended up doing was looking at it and saying, “Well wait a second. There was a lot of these questions where I really don’t know what the answer is.” And unwittingly that caused them to go when they were working with those accounts to make sure that they asked those questions.
So without realizing it, they were aligning their sales behaviors. We didn’t have to tell them to do that. The qualification criteria was credible to them. And so they used it like a recipe. I need some of these and I really need to increase that one. I need some more of the … I need to know the answer to that question. So we ended up helping them see that if they can’t increase the score, then they have permission to walk away. And in some cases you had to kind of force them to walk away. This is like an x-ray machine looking at your sales funnel, because it tells you where things are broken, where the parts are missing, where we need a new x-ray and it’s very, very productive. There was another thing that we were able to get out of that data. Do you remember what it was?
Bill Bentley: Well I do. But some of your questions, which everybody expected that the higher the score, the greater the likelihood that you would land the client. When I did the modeling, the answer came out to be exactly the opposite. The higher the score, the lower the probability that you would land this customer.
Michael Webb: Yes. And an example of that was that same one I mentioned earlier about your coach network and boy we were scratching our heads for awhile trying to figure that out. But because this is set up as closed loop, we’re getting data back from the use of this instrument, we were able to learn what it was. We interviewed some of the salespeople, it only took two or three interviews to realize what was going on. Maquet’s strategy had been to hire, degreed, not degreed, but experienced respiratory therapists who managed respiratory therapy departments, not professional salespeople. And so they were respiratory therapists first. And to them a coach, they were just interpreting it as, “Oh, this must be my friends back at the hospital where I used to work.”
And so because they were misinterpreting the meaning of the question, the data was coming back wrong. And so we worked with them. And there are actually two or three questions. I think there were three like this and we had to go back and rework the questions with the sales people’s help and then as a result of that, we produced a new instrument and moving forward the forecast accuracy of that, as I recall, it was like 94, maybe even higher then 95%, you remember what it was? Really high.
Bill Bentley: I don’t remember the number, but I remember believing for this kind of a process, it was an amazingly high number.
Michael Webb: Yes, and the president of the company told me several months later that it was incredibly effective for the sales force. They had a lot of respect for this data in this qualification criteria. And so when they would get to a prospect and they couldn’t figure out how to increase the score, go back and ask the questions, then they were supposed to walk away. Sometimes you had to kind of encourage them to walk away, but it gave them permission, so it’s like an x-ray inside their funnel. I may have said that before, to see what’s going to happen and what’s not. This is very profound and because it tells the sales force A, are we going to make our numbers or not so they have a little time to compensate if there’s not enough good leads in the funnel.
Another thing it does is it allows you to measure the impact. Suppose they decided, other data showed them that they needed some sales training of some kind to be able to get customers to talk to them more effectively during sales meetings or to better be able to present value propositions or build more coaching relationships. Well you would be able to see the effect of the sales training in the quality scores that the salespeople were producing. And so now you have the potential to calculate ROI on what has been called soft skills training. And it’s really not because there’s cause and effect relationships as professional salespeople know. It’s not like a machine, but there’s definitely causal effects going in there, happening in there and now we’re getting at it by use of this operational definition approach. Have you seen stuff like that happen in other places or implications of this in other places Bill?
Bill Bentley: I’ve done a lot of models and I have seen other instances where models produce results that were unexpected. Nothing in this system like I working with you, not in the sales business. Doing analytical modeling in sales is pretty unusual. You have a pretty unique business. But in other kinds of businesses it’s normal practice and they tend to be more mechanical and more physical and more process oriented. But even there we have found where variables don’t have the effect on the process that we thought it had, or the effect is limited to a narrow range, just like you were describing with the tipping point. Sometimes the effect is limited to a narrow range. And sometimes the effect is only there when other effects are there.
So we provide interaction and that’s very difficult for a person to recognize that certain circumstances, this activity that they perform or this value that comes out of a sensor is going to have a positive effect. And on the other times it has no effect at all or a negative effect. And that’s because some other part of the process is doing something to mask it. And that can come out in the modeling process and give them great insight into what’s happening.
Michael Webb: You and I shared an affinity in this recognition of the scientific method and how effective it was. And actually, it’s difficult. You have to learn how to do it, controlling your own mental processes to be able to do it effectively. And that’s something everybody has to learn for themselves. But I had grown up in sales and constantly seeing these situations where in a manufacturing environment they would apply a technique like this and things would improve when they did it right, had data instead of just somebody’s idea. But in sales, everything was subjective. It was just, work harder. So I knew there had to be a way to do this and these projects that you and I worked on prove that. And if you step back and you look at the so-called data that is available by the hundreds of thousands of gigabytes now in CRM systems, and yet none of the salespeople are defining their terms even in the same way because executives don’t realize they need to develop that discipline throughout their whole organization, not just on the sales team, marketing team, sales team, service team.
And then you look at the fact that the percentage of B2B sales forces that are sales people, that are making their quota every year has been declining. It’s eight years in a row now, as I understand, it’s just barely above 50% of sales make their quota. This is from CSO Insights. They do this survey every year of like 2000 companies around the world. And that’s what they’re seeing. And I mean that is a systemic issue. There’s something in the way management is managing sales and marketing that is causing this because all those companies are trying everything they can to improve their sales productivity and it ain’t happening. And so I believe that this bringing this scientific mindset to be more reflective about what are the observable characteristics that cause a sales opportunity or a customer account or a distribution channel account or a salesperson for that matter.
What are the observable characteristics that make them high quality or low quality, that creates a theory and you’re articulating that theory and you’re articulating by observed observation, this operational definition, that is data that can only help you diagnose what is actually happening in your organization and the experiments that you and I were fortunate enough to do together definitely prove that so I thank you for that. I couldn’t have done it. I’m a major in math but not like you Bill.
Bill Bentley: It just takes time and sweat, that’s all.
Michael Webb: I’m sure there are other people out there, maybe some other stats geeks like you, who might want to know how to do this or chat with you. How can they get a hold of you Bill or are you open to that? I mean you are retired now.
Bill Bentley: No, I’ve started a brand new career. I have a full time data science job now.
Michael Webb: Okay. But I mean it’s a retirement job. I mean are you okay if people reach out to you to chat about this stuff or something else?
Bill Bentley: I am. My value chain business is continuing and I’m still doing training, I do it virtually. I’m still open to doing consulting work. I’m just limited in when I can do it. I can’t do it in the middle of the week when I have my other job to do but I do look for it and I’m interested in doing it. I’m working on a project right now on a muddling project, that has a lot of the same similarities that you’re saying muddling projects had and that is that they’re using Likert scale data to make decisions, only unlike on your project where you had the ability to coach and teach the salesman on how to answer these questions. These companies are having the questions answered by strangers in effect, who are only going to do it once. So they had no ability to help them answer the question and they’re making big decisions based on what I considered to be very iffy, flaky data.
They’re not the only company to do it. It turns out there are a lot of companies that are doing this. I don’t know if I’ll be the one to blow the whistle on it and tell them to all stop. But we’ll see.
Michael Webb: Well, and likewise if you find somebody who wants to know how do we do this with our people, motivate them to be more anal and more precise as a team. And if you want help with that, you can send them my way. But this is a great story. It’s a wonderful discovery. I was really tickled to be part of it. So how can people get a hold of you, Bill?
Bill Bentley: The best way to do it is to go to my website at value-train.com and all my contact information is there.
Michael Webb: So we’ll include a link to that. I’ll include an example on the show notes page of one of the statistical charts. I remember one in particular that showed the borderlines of the tipping point for one of these clients so that the stats geeks can look at it and kind of see how we did it. And boy, this was a fascinating story and I really appreciate it. And maybe there’s another topic that you and I can talk about again. Would you be willing to come back to the show?
Bill Bentley: Sure. I’d be glad to, this was fun.
Michael Webb: Well thank you very much. And until next time, take care.
Bill Bentley: Thanks Mike, talk to you later. Bye. Bye.
