In-Ear Insights: How to Use Generative AI in Sales

In this episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris tackle the challenge of using generative AI in sales, both B2B and B2C. Discover how generative AI helps you analyze customer data and identify pain points to improve your selling approach. Learn how this technology can bridge the gap between sales, marketing, and product development for a more collaborative and effective strategy. Katie and Chris emphasize the importance of understanding customer needs and aligning your products or services accordingly, highlighting the limitations of AI when it comes to human judgment and decision-making.

Watch the video here:

https://youtu.be/gUMFPlMxdX0

Can’t see anything? Watch it on YouTube here.

Listen to the audio here:

Download the MP3 audio here.

[podcastsponsor]

Machine-Generated Transcript

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.

Christopher Penn – 00:00
In this week’s In Ear Insights, let’s talk about generative AI for sales, both B2B and B2C. How do we use these tools to do arguably one of the most important functions within a company, which is to help bring on new customers? Katie, when you think about sales as a profession, as a discipline, as a function within a company, what do you see as the major obstacles, period, that sales runs into? If you were like, “Oh, if I had a magic wand, I would fix this.”

Katie Robbert – 00:37
That’s such a big question because it’s going to be different for every organization. I guess the common theme, the common deficit I see is collaboration with other teams. When I worked at an organization that had a dedicated sales team, there was a disconnect between the product team and the sales team, and the marketing team and the sales team. Product and marketing worked really close together, and then we would bring sales into it, and somehow sales remained on their own island and then would complain that we didn’t have the latest information: “We don’t know this; we don’t know that. Who’s our target audience? Who are we going after? Has it changed?”

Katie Robbert – 01:30
I guess the thing is data sharing and a communication breakdown. From my perspective, sales is the last piece of the puzzle. They’re the last touch. When you’re bringing on a new client, first, you have all of these other pieces in the customer journey. You have your content; you have marketing campaigns. You have maybe some interaction, like webinars, emails, and live shows, but sales isn’t in it yet. Sales is the one who comes in at the end to close it. I think that there’s a disconnect in that process.

Katie Robbert – 02:14
I don’t know if that’s the answer you were looking for, but I think that if I had a magic wand to fix—from my perspective—what’s wrong with sales, it’s that lack of collaboration, which leads to the lack of the most up-to-date information about prospects, customers, the product itself, and what you’re selling.

Christopher Penn – 02:38
When I worked at a tech company, the biggest complaint that sales had was that the product they were selling was twice the price and half the quality of the competitors, which was an accurate assessment. It was an overpriced, underperforming piece of software. Their closing rates for selling were about 1%. Out of every 100 opportunities, they closed one of them, and that kind of rolled downhill because marketing was like, “Okay, so we just need more leads. We need 3,000 leads a month.” You’re going to run out of companies to sell to really soon because there are not enough companies willing to pay exorbitant fees for underperforming software.

Katie Robbert – 03:24
In that case, it sounds like there was a disconnect between what sales was hearing and what the product was doing. Again, to me, that comes down to a communication deficit. Maybe things like having an ideal customer profile that everyone has access to that they can work off of might be a good solution. I don’t think having more meetings so that people can talk to each other is a good solution because it’s falling on deaf ears. It’s just—that’s not working. More meetings are not the solution.

Christopher Penn – 04:05
Going back really far, Bob Stone’s 1968 direct marketing framework is a really good model for understanding selling in general. It’s three layers: list, offer, creative, which is a broad way of saying, “Are you selling to the right people? Are you selling to the right people something they want,” which is the offer, “and then, how are you doing the selling?” One of the biggest mistakes I see in generative AI with sales—or in general, with generative AI and sales, specifically—is that people are devoting all their time and effort to the creative portion, like, “How do we update our sales scripts? How do we make more compelling sales emails?” That’s the last part of the process.

Christopher Penn – 04:49
If you’re selling to the wrong people or trying to sell people something they don’t want, no amount of clever marketing and clever selling scripts is going to change the fact that people are not going to buy from you. When we’re talking about generative AI in sales, you’ll be much better served figuring out, are you selling to the right people, and are you selling them something they want?

Katie Robbert – 05:13
It sounds so simple, and yet—well, it is simple.

Christopher Penn – 05:17
It’s not easy. That’s the—that’s the challenge.

Katie Robbert – 05:20
Got it. So how do we get there? How do we find out if we’re selling to the right people? For a company like us—we’re small—we can be agile, we can adjust our services, we can adjust our target audiences. If you have an enterprise-size company with a deep bench of salespeople who are on commission trying to—you know—get the thing done, close it, they have quotas; they have targets. Change management is really tricky because it’s a cultural change. Telling people, “Okay, now your targets are different. Now you have to do something else” is going to be—for lack of a better term—a hard sell because they’re like, “Well, my paycheck is dependent on me making these numbers. Now I have a different audience and different numbers.”

Katie Robbert – 06:15
“How does that work?” So how do we first solve for who we are selling to, especially if we have it wrong in the first place? Then how do we adjust our tactics?

Christopher Penn – 06:27
It’s kind of a chicken and the egg, right? Product and sales are chicken and egg. You have a product. You have to find—either find people who want that, or you find people, then figure out what products to sell them as. Product-market fit, which is a discipline unto itself. The most straightforward way to figure that out is to look at what you have right now for sale and—surprise—put it through the 5P Framework. What does your product do? What is its purpose? What are your product or service’s purpose? If you don’t know what your—

Christopher Penn – 07:05
—if you have one of those things that is famously called a solution in search of a problem, you should be updating the LinkedIn profile because there’s no amount of sales training that is going to fix a solution that no one needs because they don’t have that problem.

Katie Robbert – 07:23
But, Chris, they just don’t know they have the problem yet. I’m going to make them aware that they have the problem, and then they’re going to buy the thing.

Christopher Penn – 07:32
Which works for creating a new market if you have ten years and $100 million to invest in it because that’s what it takes to build a brand-new market that is sustainable.

A really good example of this: HubSpot took ten years and $100 million of marketing to create inbound marketing, to create this discipline that did not exist before, to identify the problem, to convince people it was a problem, and then to get people to buy their product. That is a totally valid strategy if you have the money to do it because then you own that market. No one else owns inbound marketing the way HubSpot owns inbound marketing in the same way that—no one owned the—no one owned the smartphone category the way Apple did in its first couple of years when the iPhone first came out.

Christopher Penn – 08:23
But they spent years and giant piles of money. That’s really tricky for any company that isn’t well-funded by a VC with an enormous amount of patience, or an existing company that is willing to burn money for a whole bunch of time.

Meta has been doing that for five years now, trying to convince everyone that VR and the metaverse are a thing. By your reaction, you could see how well that effort is going, in particular, even though they’re still working on it. It’s just going to take another five years and hundreds of millions of dollars to convince people that this thing is a solution. This is a solution, but it’s a solution to a problem most people don’t have, which is, “I want to be in a virtual world because it’s nicer than the real world.”

Katie Robbert – 09:18
Yeah, no, that is not a problem I have. I think that would create more problems for me—first and foremost, vertigo. I don’t want—I have enough of that. I don’t need more of that.

Your point is that a lot of companies are working backward. They have a solution in search of a problem. When you start with the purpose, it is “What is the problem we’re trying to solve?” not “What is the problem we’re trying to solve with our services?” It’s “What is the problem we’re trying to solve,” period. That comes from understanding your audience, but what are their pain points?

Chris, you like to say, “What keeps you up at night?” What is the thing that—I’m trying to remember all of the different ways that you’ve put it because, in this context, it makes sense—”What is the thing that will either get you promoted or fired? What is the thing you’re measured on? What is the thing that is the most troublesome in your week? What is the thing that takes you the longest amount of time to do?” Those are pain points. That’s what drives constructing and detailing out your purpose. You, as the solution seller—your purpose should 100% be reflective of the pain points of your audience. If you don’t know what the pain points of your audience are, that’s where you need to start.

Christopher Penn – 10:45
Exactly. That is something that generative AI can actually help with if you have the data. One of the things that we do for clients—and in general, because I think it’s fun—is build personas, build profiles of customers based on actual data. Let’s say you were selling smoke shifters for a barbecue—just a ridiculous example. It’s a fan. That’s what—that’s all it is.

How would you build that persona? You would go and figure out who barbecues, where they talk about barbecuing, and what they talk about. What are the things that are going wrong? Put the product out of mind—like you just said—put the fact that you’ve got this fan out of mind, and look at the problems people have.

Christopher Penn – 11:40
When you’re barbecuing, particularly with charcoal, one of the problems is it takes a really long time to get the coals up to temperature. As long as you’re doing it properly—you can, as I have done, just pour two bottles of lighter fluid on the barbecue and create a massive fireball, and that does get things started—however, some people say it does leave food with a sort of a fuel taste. You can use all these different chimneys and stuff like that, but that’s the pain point. “I love the taste of charcoal barbecuing, but I don’t have 45 minutes to wait when I just have to get dinner on the table during the weeknight.”

Christopher Penn – 12:19
“Yeah, I can do it on the weekend, but I don’t have 45 minutes to wait during the weekdays when I’ve been at the office for 12 hours,” and so on and so forth.

So how do I solve that problem? Doing that assessment, if you were to go onto Reddit, use social monitoring software, go into barbecuing forums, or talk to people—you know—who are barbecuing—that amazing thing marketers almost never do, which is talking to customers—will highlight a lot of feedback. You can then take into a generative AI system and say, “Summarize and outline the major things people complain about when barbecuing.” Once you’ve got that data distilled down, then you could say, “Do we solve any of these problems?”

You may have a smoke shifter; no one has a problem with smoke blowing in their face. That’s because your grill is generally outside; you just move. You don’t need to buy an $89 gadget.

Katie Robbert – 13:16
Yeah, you can just literally step aside.

Christopher Penn – 13:18
Exactly. But if you have a device that can move a lot of air, and it turns out that people are having a very hard time getting their coals to start to a proper temperature, can your device be used to accelerate how quickly those coals heat up? The answer is yes. That’s why people use hairdryers on barbecues.

Yes, you can use a hot-air hairdryer on a barbecue to dramatically accelerate how fast the coals—a leaf blower works well, too.

Katie Robbert – 13:44
I will not derail this episode, but I have so many questions.

Christopher Penn – 13:51
Through that market research, through building that ideal customer profile and digesting in all of that data with generative AI, you might say, “Okay, we’ve got a product. It doesn’t solve the problem we say it solves, but it does solve this other problem. Maybe we should try changing our selling approach to address the real pain points people have.” We basically pivot the product itself.

Katie Robbert – 14:17
One of the things I want to highlight that you’ve been talking about but haven’t explicitly said is that this is all qualitative data. This is all—you know—unstructured, conversational data. The reason why companies struggle with this information is because it’s hard to analyze. Historically, it’s been hard to analyze because you can have so much of it. We’ve done machine-learning techniques, like topic modeling, to find, like, “What are the common themes?” But not every company has access to do that kind of machine-learning analysis. Fast forward to today, and companies have access to tools like Google Gemini or ChatGPT, LLaMa, Claude—pick a tool—and summarization is one of the main categories of use cases of generative AI.

Katie Robbert – 15:16
That’s what you’re talking about, Chris. You’re saying, “Give me all of this text, give me all of this conversational data, market research that companies have struggled to make sense of because there’s so much of it, and summarize it down to a few pain points so that I can say, ‘Do we solve for that or not?'”

Christopher Penn – 15:36
Exactly. We are talking about leveraging—and this is a point that you make frequently, Katie—data is not just numbers in a spreadsheet. Data is any information you have lying around. You have probably more data than you know what to do with. Your customer service inbox is filled with it. Your call center is filled with it. Your website comments box is filled with it. Anywhere that a customer can talk to you is data; it is customer-facing data. If you’re not making use of it, you’re basically ignoring a big pile of gold and then running around the yard with a shovel, going, “Why can’t I find anything valuable?” It’s like, “Well, you already have a pile of gold; you’re just ignoring it.”

Katie Robbert – 16:18
So I feel like—we’re still sort of on the purpose, which I think is a great place to be focused on. Why aren’t companies using social listening tools or anything like that to say—to really understand their pain points? Are those tools not doing a deep enough analysis?

Christopher Penn – 16:45
The tools don’t do analysis generally, other than summarizing top-level data. They certainly don’t do any qualitative research processing.

A big part of AI enablement is to say, “Hey, what tools in your martech stack do you already have? What data do they output? Do you have access to a large language model with a sufficiently large memory to be able to process that data and boil it down?”

For example, with one of our clients, we get thousands and thousands of website survey responses every month. Prior to the advent of language models, no one looked at that data because there was just too much of it. Once a language model came around that had enough of a context window, we could feed it all to it and say, “Summarize this.” Now you have action—

Christopher Penn – 17:37
—now you have—you can take action on that.

More than anything, when it comes to generative AI for sales, do the people know that these capabilities exist? You have a CRM—HubSpot or Salesforce, or choose a CRM of your choice. You have customer interactions logged in that you have: called sales notes from your salespeople, I hope.

What’s going on with that data? Are you using it? You have one-on-ones with your sales manager and your individual salespeople, and those one-on-ones should be frequent, should be like weekly one-on-ones where you’re saying, “Hey, Katie, what was good this week? What are you blocked on? What deals are you working on? Why is that deal taking so long to move?” What are you doing with that information?

Christopher Penn – 18:34
With permission and consent, you could be recording all that and then feeding all those transcripts every week into a tool to say, “Here are the sales, the pain points that our sales team is having.” You might notice the seasonality, like, “Hey, I can’t get ahold of everyone. It’s the week of July 1. I can’t get ahold of anyone.” It’s like, “Well, yeah, that’s because everyone’s on vacation.”

Or you might hear things like, “Yeah, for some reason, people are saying, ‘We just can’t get budget approvals right now.'” Okay, then we—now you have useful information from the front lines saying, “This is an issue.”

Christopher Penn – 19:12
Again, this comes down to the people not knowing what capabilities are available. If they don’t know those capabilities, they can’t build a process around it, and they can’t use the platform to make that happen.

The purpose is pretty clear: you want to improve your sales, you want to improve your closing rates, you want to make more money. But the people need to know what capabilities exist, how to set up processes for that, and what platforms to use to get that performance.

Katie Robbert – 19:48
I like where you’re going with this because what you’re maybe knowingly or unknowingly doing is you’re setting up different versions of the 5Ps. We started with, “What is the problem we’re trying to solve for our customers?” You’ve also talked about, “What are the capabilities of our existing toolset?” Each of those questions can have its own set of 5P requirements or 5P answers.

Let’s say you start with, “Okay, the problem that we’re—we are—we’re trying to understand the problem that we’re trying to solve for our customers.” We say, “Here are the people. The process we’re going through is this and that, and here are the platforms.”

Katie Robbert – 20:32
But you get stuck at the platform and say, “Well, we have a CRM; we have social listening tools. We have this; we have that. Do those things solve—answer those questions?” You can go back and spin off a whole different set of 5P requirements to understand, “Can our CRM answer that question? Does our social listening tool have the capabilities?” Once you get that information, come back to your original set of 5Ps. Then you can augment and say, “We will need our CRM, and we’ll need our social listening, and we’re going to need to acquire some other tool, or our process is going to need to include gathering this additional information.”

Katie Robbert – 21:18
That’s one of the ways—in this context—to use the 5Ps in a mo—to get to a bunch of different answers, always coming back to, “What is the problem we’re trying to solve for our customers? What are their pain points?” Once you have a sense of how you’re going to get there—Chris, to your point—then you can say, “Do we have services or products that solve those problems?”

Christopher Penn – 21:45
Exactly. Generative AI is a super useful tool in this particular example because you can say, “Here’s all the data I have, and here’s my purpose; here’s what I’m trying to do.” You could say, “Do you have enough information here to draw those conclusions, or do you need more data? If so, what additional data do you need to draw those conclusions?”

You say, “Tell me what the key components of an ideal customer profile are,” and it’ll spit out demographics, needs and pain points, goals, motivations, etc. “Here’s the data I have. What am I missing that would help complete a solid, ideal customer profile?” It will say, “Hey, you don’t have purchase history, you don’t have this.” It’s like, “Okay, well, where do I get that?”

Christopher Penn – 22:36
Then you can say, “Okay, well, I need purchase history from my CRM, or I need interview data from focus groups and one-on-ones,” whatever.

The thing is that generation outlines—”It’d be nice if you had this.” You can use—you can almost use it as a tool for helping you do requirements gathering. It says, “Here’s what’s missing. Here’s—here’s the ideal state.” That’s what those systems are really good at; they’re really good at the big picture: “Here’s what would be great to have.”

Then you go back and say, “Well, here’s what I’ve got.” It’s like, “Well, you can do a version of this, but it’s not going to be as good if you had this.” Then you could say to your tech leaders, to your sales leaders, “We need this to solve for this problem.”

Christopher Penn – 23:19
“Why isn’t sales selling enough? It’s because everyone hates the product.” Or the product is too expensive, or there’s not enough of a need. Traditional sales training will say, “There’s no such thing as too expensive. If the need is urgent enough, people will find the budget,” within reason, perhaps.

But if you knew what the actual pain points were from the data processed by generative AI, then you would have less of that conversation because you would know, “This is the level of pain that someone is in; it’s not enough to justify the purchase of our product.”

Katie Robbert – 24:00
What’s interesting is we’re talking about direct from customer pain points to the selling process, but marketing is in between the two. It means that marketing also doesn’t have the right information for messaging. It also doesn’t have the right information for their campaigns because they are spending budget and resources and effort and all of these things coming up with campaigns that are not generating leads that sales can’t follow up on.

To your point, you’re saying, “We need to get 3,000 leads, and then sales—based on their historic closing percentage—they can just close the thing.” Not if nobody—marketing, product, sales, whoever, customer support—not if they’re addressing the wrong thing. It really comes back to, “Do you know the pain points of your audience? Do you know who your audience is?”

Katie Robbert – 24:55
Chris, you’re talking about—what did you call it? A smoke shifter? You may think that your audience is—you know—people who just grill and barbecue outside, but what if there’s—you know—a secondary audience that would stick—depending on the size of it—that smoke shifter inside a fireplace in their home to keep the air circulating and push the smoke up the chimney versus into the living room? That’s an untapped audience who has a real problem that you may not know you could solve for because the research wasn’t done because there’s just this assumption that, “Well, no, it’s people who charcoal grill. That’s our audience. That’s it?”

Christopher Penn – 25:39
It’s—it’s the—like the example I used to give with our predictive analytics talks. My Little Pony is purchased by 26-to-40-year-old men in much greater quantities than 8-to-14-year-old girls. If you were relying solely on your assumptions, you would be missing a lucrative audience. You would be missing an audience that has, like, 10,000x the spending power of what you think your real audience is.

Again, that comes from assumptions and not listening to the customers, not using the customer data that you already have. If you were pulling out purchase data, if you were looking at social media data—”Huh, there are a bunch of dudes who are buying, like, 28 copies of Rainbow whatever. Why is that happening? Should we be marketing into this trend? Should we be selling into this trend?”

You make an excellent point that marketing needs this data as much as sales does, if not more.

I ran into this—I run into this all the time. There’s a game I used to play, Diablo, by Blizzard Entertainment. Diablo Four came out, and they are spending crazy dollars marketing and retargeting me, except that I have a Mac. They didn’t make a Mac version. You are wasting money, and you have my data because I also play World of Warcraft. You have my data. You know what kind of computer I play on. Why are you showing me ads for a product I cannot buy? Like, I mean, I could buy it, but I’m not—I won’t be able to play it.

Again, the customer data is there; no one has connected it to marketing.

Christopher Penn – 27:11
No one has connected it to advertising, and certainly no one is connecting it to sales to say, “Maybe not—let’s not waste $125 a week trying to market this product to a person who can’t buy it.”

Katie Robbert – 27:24
Again, it sounds so simple, but it’s not easy because people—people—despite having the data, still make decisions based on what they think is—they’re calling it “gut instinct,” or they—they call it “anecdotal data.” Why? Because it’s faster. It doesn’t mean it’s correct, but it’s faster because they can get to action faster. Their line of thought is, “Well, if I can make a decision, and I can do a thing, then I have a 50/50 shot of it going correctly. I’ll take those odds because that’s better than spending time that is going to slow me down, and I can’t take action on it right this second.” That’s a problem.

Christopher Penn – 28:17
It’s a huge problem when you think about the classic way of qualifying sales leads, particularly in B2B, but this is universal. This came from IBM back in the seventies. They called it BANT: budget, authority, need, and timeframe. Those are the four factors that determine whether a sale is likely to happen. If someone doesn’t have the budget, they—you’re not going to get the sale. If someone can’t make a decision, you’re not going to get the sale. If someone has no need, you’re not going to get the sale. If someone has no clear timeframe, you’re not going to get the sale.

The problem that has been outlined in sales over the years since then is that, very often, you don’t have all the data for those characteristics. One of the things that generative AI solves—partially—is it can partially solve need and timeframe if you have customer data from all your social listening and data gathered.

You can get an assessment like, “Okay, when people are talking about this general problem, how important is the need? Do people say, ‘Yeah, I keep getting smoke in my eyes, but—you know—whatever, I’ll just move to the side?'” You can see there’s not much of a need there. If people are talking about solving the problem, like, “Oh yeah—you know—it took me nine months to get this product, but whatever”—you can see there’s not a level of urgency there.

Where BANT has gotten into trouble is that sales folks made assumptions to move faster—like you said—to—like, “Oh, this person doesn’t have the authority. Well, maybe they do, but we don’t know that.” You’re assuming because their job title is “marketing coordinator” that they don’t have the authority. Perhaps they do. Perhaps they have the ear of the CMO, and because they’re the ones putting together the shortlist for the CMO—you don’t know that.

Generative AI can partially solve that problem with the caveat: you can’t solve all of it. You can’t make those inferences. Again, don’t think it’s a magic wand. If you’re trying to use generative AI for sales, it has very good uses for summarization, but by no means is it going to do the selling for you.

Katie Robbert – 30:17
So I think what I’m hearing you say to start to—like, pull it all together. If we’re talking about how to use AI in sales and what problems it can solve, the challenge that a lot of companies historically have run into is having a lot of qualitative and unstructured data that they couldn’t necessarily analyze because of the volume, because of the capabilities. Now that’s something that generative AI can help with so that you can really dig into the pain points of your customers so that your marketing can be more focused, and your sales team can be selling things that people actually want and want to buy, that there’s a real need for, that there’s an urgency for. That comes from understanding your customer and your customer data.

Katie Robbert – 31:03
What it can’t do is if you have a product in search of a solution—I mean, if you have a solution in search of a problem, it’s not going to fix that for you. That’s very much a human judgment where you have to say, “All right, so we don’t have anything that addresses the pain points of who we think our target audience is. So we either need to do one of two things. One, we need to adjust our expectations of who our audience is or, two, we need to adjust the services and the products that we offer—or a combination of both.”

Christopher Penn – 31:38
Exactly. When it comes to that selling flavor—list, offer, creative—don’t get distracted using generative AI to just endlessly focus on creatives, like rearranging the deck chairs on the Titanic. Focus on the highest priority things first. Are you selling to the right people? Do you have something those people want? Then worry about, “Is your sales script or your telephone script or your email nurturing program in good condition?”

A less-effective sales script, but for a product that people desperately want—you’re not going to have any problems. If you’ve got a problem product nobody wants, the most finely tuned sales script is going to sell onesies and twosies at best.

Make sure that your priorities are in order when it comes to using generative AI in sales.

If you’ve got some stories that you want to share about how you’re using generative AI in sales, pop on by our free Slack group. Go to TrustInsights.ai/analyticsformarketers, where you and over 3,000 other marketers are asking and answering each other’s questions every week about data and analytics, AI, and—wherever it is you watch or listen to the show, if there’s a channel you’d rather have it on instead—go to TrustInsights.ai/tipodcast, and you can find us where most podcasts are served. Thanks for tuning in, and we’ll talk to you next time.

Need help with your marketing AI and analytics?

Machine-Generated Transcript

Leave a Reply Cancel reply

Subscribe to our Weekly Newsletter

Pin It on Pinterest