Making sense of sales data

0:06Dr Simon Hefti:Welcome to Mastering Sales and Negotiations, the podcast for ambitious professionals who want to sharpen their skills, transform their conversations and deliver lasting commercial impact. In today's episode, we're joined by Dr Simon Hefti, co-founder and chairman of D1, a leading Swiss data and AI consultancy based in Zurich. Simon combines deep technical expertise with the rare ability to translate complex data challenges into better, more practical business decisions. Together, we explore a frustration many sales leaders will recognize, capturing huge amounts of CRM data, but still struggling to use it to improve performance.

0:45Dr Simon Hefti:We unpack why that gap exists, whether the issue sits with the data itself or how organizations think about it, and what needs to change for sales teams who want to drive better outcomes. We also look at the role of AI and where it can help surface patterns and bring clarity and why it still depends on having accurate up-to-date data at its core. So whether you're leading a team or working on the front line, this episode offers a grounded practical view on the best steps you can take to ensure your data becomes more meaningful and more useful in the moments that matter. So let's get started.

1:27Dr Simon Hefti:Good morning, Simon. Welcome to our podcast. Good morning, Rachel. Good morning, Robin. Very happy to be here. Let's go in with the first question then, Simon. So from physics PhD to co-finding a data consultancy, what does a physicist see when they look at how businesses use data? Oh, that's a nice one. So in our training, we learned to look for the underlying mechanisms, systems, the first principles as they are called, you know, gravity, electricity, these kinds of things. And I think that's what we bring into the picture when we look at business data. We try to understand what's beneath that, what makes it look like it does look, trying to find the rules or the laws behind it.

2:20That's probably the deterministic factor. I just want to add that I realized in my life that I could bring two passions together. So I did physics at the beginning, which I loved. Then I went into software engineering, which I loved as well. And all of a sudden I said, hey, you can combine the two and actually have the fun of the both. And that's how I came into the business. your company d1 that you founded some years ago says that it looks around data-based value creation just unpack that for us what what how do you differentiate between those organizations who've got data and most organizations have lots of it and then being able to create value from the data that they have?

3:11Yes, it's actionability, right? So learning things you can then use to influence the course of what's going on or what's going to come. So in a very simple economic view, the value creation comes from the four basic principles. You increase your revenues for sales, that's probably the most important goal. You decrease costs, that would then maybe the CFO be driving. You decrease risk, that would be the third one, meaning you detect things, patterns, which are going to happen early on while you still have time to mitigate them and react to them. And then there is a fourth one, which is ambitious and also difficult, and that's finding new ways of income.

4:00We have seen the switch to subscription-based business, which is a typical such a move that you turn a service into a subscription and then you have a new revenue source. But basically, in all those four areas, it's always actionability. So data in itself, it's like if you look at the weather report and you see it's going to rain, but you still don't take the umbrella, right? Then you have the data and you haven't done anything about it. if you have the umbrella ready for when it's raining and you can open it, then you'll get some value out of that. And I think that's the differentiation. I think within our world, we see quite a lot of sales leaders who go out without an umbrella.

4:43So I'm intrigued to understand why sales organizations still struggle. And I think I'm making a bold and unfounded assertion there, But certainly anecdotally, and when we look at how many people are using AI within sales or using data correctly, I spoke to a number of sales leaders just after ChatGPT was first launched. So in that period, about six months after that had happened, I spoke to a number of sales leaders at various conferences and events, and all of them have got CRM systems, and none of them think the data is accurate. So that seemed to me to be a bit of a disconnect. What is it about using data that sales teams, sales leaders find difficult?

5:34I think we have two pieces here. So one is, you just mentioned the accuracy of the data, the availability of the data, maybe also the reasonness of the data. So that's one piece to look into. And the other piece, the keyword for me is danceability. Let me explain what I mean by that. A few years back, we had a project where we looked at the charts, you know, how songs behave in the radio stations, how well they are received. It was 50 years of the Swiss charts, radio charts, so we looked at that data. And if you look at the song, it's really boring, you know. So you see it go, a typical thing is it goes up to the first rank and then stays there for a while and then comes down again.

6:25And you can see this pattern, but you learn nothing from it. It's just you look at it and say, yeah, that's what I experienced and that's it. And then we started to add properties about the songs into the picture. And one was danceability. And then all of a sudden the story becomes clearer because you say, aha, now I have an upbeat song which makes people dance, which goes to rank one and they can make a story out of that. And I believe for sales, this is a bit the same picture, right? If you just have, let's say, the chart rank, where am I in a simple metric? You don't get really speaking data.

7:06You need to search for additional input for features, we call them, which then make the data explainable. and and that idea of of making the data explainable i suppose what what what are organizations doing with that data most of which lives either in a crm system or some other kind of customer record process but it also is data which is coming from the market and it's coming from year-on-year finance figures and and those things what what are organizations doing where they're really making that danceable and therefore actionable? What are they doing right? What are the organizations that make the most of it doing right?

7:54Maybe picking two examples from projects you've worked on. One is a typical metric is looking at the win rate of the sales team or the individual sales people. And what you then try to do in terms of explainability is go to the explanation to the driver which lead up to that win rate. So you try to find from the data you have, what makes it? Is it a specific product specification? Is it a specific go-to-client approach or go-to-market approach? Is it the specific behavior of an individual which drives that factor? And you drive to a frequent tool for that is clustering. So we try to identify identify groups of people which do something similarly and match, correlate that with their success in their business.

8:48So that's one approach to look at that. And the conclusion then is very close to what you are doing is right, the best can teach the others what they do well and then they can pick this up. So you can basically upskill the organization as a whole. So just to go to the other extreme, one of our clients calculates for the existing customer base, but you always have the upsell potential there, calculates on the order of 150 ,000 such measures like that. It's not a dance ability, obviously, but measures like this on a monthly basis. So basically, they have a very detailed picture of what is important for a specific customer.

9:37And then, of course, can tailor campaigns specifically to the needs or to the expectations of these people.

9:47Dr Simon Hefti:Do you think then, Simon, that there is a person obviously doing that? there is a person looking for the dance ability, looking for the story, who's going to translate that story then. That's the piece that's missing because lots of sales organizations are swimming in data. They've got lots of data, but maybe not that person, that clever piece of technology that can translate all that data into the story that they need. Absolutely. So just expecting that a tool delivers that for you, you'll probably get disappointed. a lot of times for exactly the reason you're just mentioning. There are two aspects to this.

10:30One is your customer base, also your market is very specific to you and your product. So it's not easily generalizable. So whatever product you use, you need to make sure that you tailor it for the situation you have. This is unique. These features I mentioned earlier, distance ability really is different from client to client and not easily transportable from one to the other. The mechanics are the same, but the actual weights, the actual values are different. The second point is what you mentioned about connecting the stories. So really trying investigating. It's a bit of a detective work, right?

11:14So it's an inspector close all kind of, well, maybe a different person. That kind of approach where you really try to figure out what is going on, try to explain, as you say, the story in the data. Is there a danger? I mean, if you've got 150 ,000 data points that you're looking at per month, then there must be a real chance that some sales leaders, some organizations, some businesses actually look at that data and there's an element necessarily of confirmation bias. So they instantly say, well, okay, how does this support the story that I want to tell rather than how do I create a story from the data?

12:01How do you get around that idea of manipulating? that there's a phrase by a Scottish sociologist and philosopher from the 19th century, and he said he uses statistics like a drunken man uses a lamppost for support rather than illumination. How do you get people to use it for illumination purposes with the data that you have available? Now, that's a very important point you're raising. It has two sides to it. One side is what we call validation, right? So once you have figured out, let's say you find low spenders have a tendency to buy the product you offer in that segment. It's a very simple relationship which probably you have found anyhow.

12:56Let's now take the other point of view. Let's try to shoot holes into that theory. Let's try to figure out whether this really works or not. And this validation or backtesting we call it because we use the data from past sales which happened, past transactions, to justify the reasoning. And that's an important piece of work. And actually back to the physics question earlier, that's something which is also part of that training, quite rigorous testing of the hypothesis you formulate. And that this This applies here as well. The other piece to this is salespeople typically need an explanation. So if you go and tell somebody, hey, contact Simon, he is interested in your products, the reaction typically will be, I know my clients, I know who is important, don't tell me what to do.

13:56And so in order to overcome this, you need to accompany the information with a good explanation which resonates with the people. If you don't have that piece, if you don't have the explanation coupled to the advice, people tend to ignore it.

14:13Dr Simon Hefti:I think, yeah, you've got to have the explanation and the right data, but you've also got to be able to articulate it well. So you've got to be able to have that great conversation, haven't you? You've also got to be able to be human to human, tell that story in a great way. And I think that's where our behavioral stuff comes in because you can then look at that and measure that and apply that and i think that that idea of having an intuition i think i think most salespeople and that you know we we would argue quite strongly with with the idea that salespeople are born not made but this idea of it being an intuitive thing it's a spare of the moment thing It's that idea of being able to get the vibes within a room and a relationship right and having everything going on in here.

15:05You know, a robot cannot tell me what I've learned about my customers over years and years and years. Is that pushback something which organizations have got strategies to deal with? And if so, what are those strategies that they use to ensure that people are prepared to look at the data, have an open mind about it, and maybe be surprised by it? The pushback is real. And I think as a whole, we are in the transition. You know, we are somewhere on that journey. Companies are, some are a bit further ahead. Others are just starting. But that transition is going on. The strategy really goes around two elements.

15:56One element is explaining, well, no, I use the same term again as we had before, but framing what we are trying to do. And in many cases, what you get back, and that's maybe counterintuitive, is more time for exactly what you describe. As a salesperson today, you are burdened with administrative, in many cases, burdened with administrative work. And that part of the work can be reduced through automation, leaving you more time for what you just described. I really believe that this intuition, this personal relationship is important. And if this framing is understood that the technology or the project which uses the technology, helps you to gain time for playing to your skills, then the perception changes.

16:53And then we have many companies do quite extensive upskilling programs we have currently. Are we involved in a program which is called 100x100, which means it's 100 events, four-hour events with 100 people each, which receive a training for how to make better use of the AI tools around. So the upskilling piece is definitely, or the learning piece, maybe better frame, is definitely an important element. And this automation of the data goes back to the question we had earlier about data which is not accurate in the system. I believe we can set up systems. We do set up systems which help curate that data and detect changes.

17:47A simple example is an email bounces, right? So you have campaigns, you send out emails, a couple of them bounce, and that's a signal you can take back to your CRM and no human needs to do that. You can automate that. Yeah, and I think that idea of making time to do the things that only humans can do at the moment and using AI in that way is very positive. And I think people need that space, need to understand how collecting the data, should there be a kind of hand-o-matic part of that, a manual part of collecting that data. if there is a process that involves them doing that then they need to see some value in how that data is used to them and i think that that process is often missing but i want to pick up on on on where the data is coming from so so clearly by definition data is historical right so we're looking backwards we're looking at what's happened within an organization and and particularly if you're looking at win rates or you're looking at revenues or you're looking at profit per unit or whatever it is that you are works for your organization and your setup you're looking at things that have already happened and then making a prediction on that basis about what will come next and you know we've all backed the favorite in a horse race and it's fallen at the first fence So how do we set up some more leading indicators rather than lagging indicators that salespeople can look at, crunch the data, and get real-time insight, which will help them move from that leading indicator to the next leading indicator, whatever that will be?

19:35Is there a way of unpacking it in such a way that we can do that? I actually believe this is the direction we are going into. that not so much the rear view mirror, as you just explained, is the main data point, but understanding what's happening now. So one important element to that is that the data cycle you have fits to the sales cycle you have for your sales. So if you have an 18-month sales cycle, then maybe a monthly update is frequent enough. If you have a very shorter scale, like days or weeks, then you also want the data to be on that speed. So I think one important element towards the leading indicators is that you have data on the timeframe you need based on the sales cycle.

20:29sales cycle. The other thing which strikes me as important is people often work in the aggregate, an average sales time, an average deal volume, and so forth. I like this Ford versus Ferrari movie. And what I take from that is if you drive a sports car, you really need to feel the road. So you need to, if there is a bump, you need to feel that and be able to take action from it. If it's cautioned, if you have a big sit-on and you just glide over the road, but you don't feel what's really going on out there, then you'll miss a lot of information. And I think the leading indicator comes from the fact that you feel that small stone, which was the bump in the road, the small irritation, and you have that available in a way that you can react to that.

21:29And that ability to adjust to that data in real time, to be able to see what's happening. So, I mean, we would, you know, because we see organizations, and, you know, this is a generalization, and like all generalizations, it's wrong, but fundamentally we see quite a lot of organizations who will actively count, or rather they accord importance to the things that they can count easily rather than count the things that are really important. And so when we get into those leading indicators, very often it's about activity. How many visits have you done? How many meetings have you done? How many calls have you happened?

22:09Our research shows that the number of calls is not that significant, albeit that the data may point to a magic number in the aggregate in the average sales conversation but what we find is what happens inside that meeting that conversation that interaction is massively important about whether or not things progress or stay still or go backwards in some cases so is there a way of gathering data about what's happening in the conversation analyzing that and giving feedback about what's working, what's not working, what's making progress, how are things advancing, and is there a way of us capturing that data and being able to utilize that real time?

22:56Is that something that is now feasible, or would it be distracting? What would be the way of thinking around that kind of ability to analyze? Absolutely feasible. Not many organizations I know do it, actually. But in terms of approach, definitely. This reminds me, you know, that what you're saying resonates with me. It reminds me of John Gottman, who did this analysis for couples, right? So basically, what he introduced is, sit with me for 15 minutes, I'll take a recording of how you discuss a difficult topic. And I'll tell you whether you stay together or you get a divorce, right? That's basically the...

23:37And he has a tremendous success rate. He tracks that over time. And what he does is he splits the recording into, I believe, 15 second segments and assigns scores in, let's say, 10 dimensions for the both persons. And then, so it converts the video or the recording into data points And from that can make a prediction. So basically this approach is exactly what you're discussing as well. And the technicality to do that in an automated way, the key word is sentiment analysis, which is one aspect of that. So you have a text or a voice recording and you infer the sentiment of the author of that. But of course you can do that in many dimensions.

24:27You don't do just emotions, you also do is it constructive, destructive, is it a rejection or an objection and so forth. So the automated way of recording that is technically absolutely feasible. You'll probably use Teams meetings and have the transcript on. So you see how this technology functions in principle. And that same can be used for, I need to say, could be used for sales recordings. But as I said, not many organizations do that. And why is that? Because everybody fears about privacy right violation, right? So then we come into the regulation part of the story, which is completely understandable.

25:17but actually the relevant information can be processed and extracted without any inference to the actual person. So it can be very well handled on a technical level, but the sentiment is now you're intruding my privacy sphere. And I get that, and I think that there has to be a governance process in place about permissions, about how you're going to use that information. And interestingly, we've got tools now which will analyze transcripts against our behavioral model and enable people to gain some feedback about what they could do less of, what they could do more of, what they need to practice, etc.

26:00So as a kind of formative process to help people build the capability to get better and better and better, it's a useful tool. I think that idea of anonymizing the customer. I mean, we literally put in customer and seller when we create a transcript. We don't tend to put people's names in and then feed that into the system. So I think that way of making that work is really important. But let's broaden it out. So we've got the ability to look at win rates and data from past sales behavior. We've got the increasing ability now to look at leading indicators, whatever those leading indicators may be, and they could be down to the behavioral level within individual conversations with individual customers.

26:52If you were asking a sales director to think more interestingly, more actionable ways about the data that they have available, what are the five sort of numbers that you would ask them to look at? What are the kind of real things that from your experience make a significant difference? Just confirming what you said, in our business, a single email can kill a project. So behavior is really an absolute key. And so we pay close attention to that. And I think you're completely right. That's an important KPI. So in terms of KPI overall, I think you want to see the funnel anyhow, right? So you have basics and this will not change.

27:48You will want to see how much opportunities, prospects do I have, opportunities to have, how many of them get to a proposal, how many of them do I convert. That stays. If you're looking into signals which go beyond that, what you're really interested in is the mood of the client changing in some way. Is this interest increasing or is it decreasing with the interactions? And that's, remember the old marketing automation models work exactly by this mechanism, right? the more interest you detect from a client through feedback questions, through need for additional information, maybe even through a price discussions, you know, go through two different phases, the more you attribute interest on the client side to that.

28:42And if you see that decreasing, you know now something else is going on. And decrease can be long long spells of silence. You lose the interaction. So we mentioned counting activities before. In many times, what we realize is looking at how are they spread out in time, how closely one after the other are they, helps a lot to understand what's going on.

29:13Dr Simon Hefti:People have got so much data now. they obviously know they can create the story they've got behavioral skills what stops them then contacting customers or speaking to customers and it tips over from being really interesting into something that's a bit creepy like you know too much you're not targeting this very well and I'm starting to feel a little bit uncomfortable with it it you know could can that can we get to that point where it it does become a little bit like that and it turns people off a bit more. Absolutely. And I think we need to be very, very cautious about this to happen. That's an effect which I believe will then dramatically decrease the interest of the client.

29:59So if you detect that, you're on the verge of losing the client in that phase. I believe one element is what you have available and what your insights or actionable information you have derived from that. And then comes a filter where you reflect, should I now use that? How should I use that? So what you're saying is, I find it super important that you could say ethically use that information or in any way measured use that information for being wary of creating exactly that effect. And we see that ourselves, right? We are clients as well, and we receive information, and what I receive on LinkedIn as requests, which are clearly automated, is just outrageous.

30:57Dr Simon Hefti:Yeah, they're automated, and they always refer to something very specific. So they're sort of general and specific, I always find, when you receive them. As head of innovation, you'll be interested in, I doubt it. One of the things that I see quite frequently, one of the things that I've said for a long time, is sales organizations, because they're vendors of whatever technology platforms and what tech stack they have available have driven them down this route, have been convinced by and bought into a number of things which uses AI, in my opinion, in ever slightly the wrong way, I think part of what you've just talked about is there that the generation of outbound emails is one of the things that a lot of sales enablement platforms make simple.

31:52And in actual fact, it just allows you to annoy 100 ,000 people a week instead of 1 ,000 people a week. Well done you. I mean, that idea of just firing this stuff out, which just goes delete, delete, delete, you know, or block, block, block in my case. So you get those things that come out. Are there other areas where sales organizations have kind of missed the point, where vendors have kind of convinced them to have a chatbot here or a customer-facing chatbot over there on your website, and it's just kind of not quite backed up with the things that need to support it to make it more effective and to make it work.