Building an AI Lead Scoring System That Boosted Conversion by 70%

0:00Chris Marin: So between January and basically the end of the first quarter, three months, we moved the conversion rate from 10 % overall across OBUs to 17%. So seven points of increased conversion. And when I say conversion, in our case, it's MQL to opportunity. Not even MQL to SAL. Not MQL to accepted lead. MQL all the way to opportunity open.

0:31Nick Panayi: Welcome back to the Health Tech Marketing Show, the award-winning podcast covering everything at the intersection of health tech and marketing. I'm your host, Adam Cherinas, the CEO of Health Launchpad. Before we dive in, thank you to you for tuning in. And if you haven't already done so, please follow or subscribe on whatever podcast app you use. We've got over 130 episodes in the library now covering just about everything in b2b healthcare technology marketing and also a quick thank you to our wonderful syndication partner healthcare now radio they are the 24 7 healthcare only radio network that brings this show to listeners all around the world through their amazing syndication network big thank you to roberta and carol for making this partnership Such a wonderful part of this show.

1:32Nick Panayi: Today's episode is the third in a series we're doing on AI and sales enablement. We started with the wonderful Jennifer Lazars from Medical Informatics Corp. And she talked about AI and sales enablement and all the change management issues that go into it. Covered a lot of ground in that episode. And then we most recently had Bruce Brandis from Whalehawk, who talked about agentic sales acceleration and the really intriguing solution that they brought to market. Today, we're tackling a problem that every health tech marketer knows pretty well, which is your marketing engine is working. You're generating leads.

2:24Nick Panayi: But how do you know which are the ones that are actually worth handing to sales? The ones that are going to help you actually build trust with sales. And this is particularly relevant if you work for a company that generates hundreds or even thousands of leads every quarter. And this is the situation that my guest today, Nick Panay, was facing. Nick is a technology marketing veteran who most recently led marketing at Enovalon. And if that name sounds familiar, you might have listened to the episode with Nick last year when he went in depth into what he led at Enovalon in AI enabling pretty much everything that they do in marketing.

3:15Nick Panayi: It's an amazing episode. You really should. You know, there's a link down below in the show notes so you can go back and listen to this. And one of the things that Nick touched on that he was working on back then, he brought to market this year with his partner in crime, the amazing Nick Marin. And what it is, is an AI-powered lead scoring system that took his team's MQL to opportunity conversion rate from 10 % to 17%. That's a 70 % improvement in lead conversion. And they did this by scoring multiple different signals instead of just one or two. and they needed AI to do that. So with that, let's get into it, meet Nick, and hear all about the agentic lead scoring system that he built.

4:11Nick Panayi: Nick, so great to see you and have you back on the show. For people who don't know you, tell us a little bit about yourself.

4:19Chris Marin: Thank you, Adam. I appreciate being here. It's always fantastic to talk with you. I'm Nick Panay. I've been in technology marketing for my entire career. I won't divulge exactly how many years that is, but it's been a while. So I've worked in a number of different industries, different size firms. My last job was with a company called Enovolon in the technology space in healthcare. And I'm excited to start a new company in a couple of weeks from now and hoping to bring a lot of the learnings of all these years to that new company. So, you know, it's been my life's work, marketing and technology.

4:55Chris Marin: That intersection excites me and that's where I'm happiest. No, they better buckle up. You bring a lot to the party.

5:01Nick Panayi: You know, we had you on the show last year and it's one of our most popular episodes. We talked for over an hour about how you and your team at Inovolom AI enabled marketing. And it really sort of, you know, it's touching almost all aspects of go-to-market, the go-to-market motion. It's really an interesting story. We'll put a link in the show notes for people to go back and listen to it. But there's one aspect of it which you touched on in that episode, and it focused on lead scoring in particular. And you've developed a solution around that. So talk about what are the issues with lead scoring?

5:43Nick Panayi: It's been around for a while. It's an imperfect science, right?

5:46Chris Marin: Exactly. It's the perfect area for us to think about how AI can help. And why is that? Two things. One, you just mentioned, it's the way lead scoring is done today is as old as marketing automation platforms have been around. It hasn't really changed much. And the premise is that you look at the person who came to your website or has engaged in some sort of activity through your marketing outreach. You apply a scoring mechanism to each one of those touches, and then you add up those points, let's call it, and then it crosses a threshold, and then you call it a marketing lead or a marketing qualified lead, and then you take it on from there.

6:35Chris Marin: interesting and it was sufficient up to a certain point in time, but it's not using the power of AI because it's a very single dimension. It's very much a single dimension and it doesn't take into account so many other variables that you can take into account before you score a lead. The reason we couldn't do it before AI is because we didn't have the ability to scale and move as fast as you want to move without the help of AI. So we'll talk a little bit about why that is. So that's number one. It hasn't really changed in years. Number two, it is by far the most challenging problem that a CMO has today and even in the past and going forward.

7:15Chris Marin: And that challenge is to build a strong, fast-growing, but high-quality pipeline for the sales organization. I always say this to my teams. It doesn't make me a very popular man in marketing, but it's true. Marketing is not successful unless sales is successful. So of all the things that we focus our activities and our resources on, not to focus on making that pipeline as strong as it can possibly be is not a very smart thing. So we said, hey, those two things combined make lead scoring a wonderful opportunity to show ROI. And at the end of the day, we get funded by the executive team and the CFO.

8:03Chris Marin: All of the other things that we do are interesting, content generation and all kinds of other things. We can move faster, we can do things better, but it doesn't directly translate to ROI and therefore it's harder to get funded. But if you do something like this and you show that you apply AI in a way that provides significant pipeline growth compared to what you had before, and you silence all the naysayers, you get that approved, you get going, that becomes your tip of the spear. And then from there, you can expand to other AI initiatives. So for us, it was a no-brainer to focus in on that. Not at the exclusion of other things, but the one thing that you can really bring forward to the C-suite and say, here is why AI can advance our costs in a significant way.

8:52Nick Panayi: So break it down a bit. What are the specific issues within lead scoring that you attacked with AI?

8:59Chris Marin: Yeah. When you are focusing on a single individual, you are looking at a small part of the elephant, but not the entire picture. The individual who is doing research on a website It is not necessarily the big picture. And sometimes it can lead you astray, right? For example, many of these individuals come in and they may line up the scoreboard. They may download five white papers and come to the website every other day. And you get super excited and you go, my God, this individual, this company must be really in the market. It turns out it may be a competitor doing research. It may be a student doing some sort of theses.

9:37Chris Marin: it may be somebody who's a potential future job seeker trying to learn as much as possible about the company so those leads kind of go along with the rest of the crowd especially if you don't have good mechanisms to screen further down the line so it creates a lot of false positives number one number two even if it is a legitimate member of buying committee it's only one aspect you're not seeing what the rest of the company is doing, you're not determining right away if the company is within your ICP. You're not determining if this individual is a persona that you care about. And you're not taking into account all kinds of other factors like, is this the kind of company that traditionally converts to pipeline?

10:24Chris Marin: Of course, all the other things may be true, and they may really want to do business with you badly, but they're a company that doesn't following your ICP. Therefore, it's not going to be a good product market fit or product customer fit. And then it ends up being a dud at the end. All of those things are problematic because it waters down the quality of leads that you pass to sales. And as a result, you start to little by little break down the trust, if you ever had it, between marketing and sales. So it has problems along multiple dimensions. That's what we set out to fix.

11:04Nick Panayi: Got it. Very well described. I think this notion of the trust between sales and marketing is, I mean, it's the biggest issue, right? So just for example, ABM, you can't implement ABM if there's not a good functioning relationship between sales and marketing and trust underpins that. And that obviously extends way beyond and ABM. So it's interesting the way that you characterize it of lead scoring is kind of where the rubber hits the road. All right. So that's the problem that you're trying to solve. What was the solution?

11:36Chris Marin: The solution is a AI powered platform. Unfortunately, one does not exist off the shelf. So we had to, as we did many times in my career, kind of innovate in this space by using other best-in-breed solutions underneath, but brought together to solve our particular challenge. So there's two parts. Let's describe the outcome that we went for. The outcome we went for was a multivariate platform that looks at not one, but 10 different signals. Each signal is scored individually, but then collectively bundled up with the other scores with the weights and the caps for each one of those variables, how you score it, happening automatically based on the success of that lead.

12:33Chris Marin: So what happens is we look at, as we said, traditionally the actions of an individual. Great. That's a part of that solution. But we also look at ICP alignment. And by the way, many companies don't have an official ICP. They go with what a salesperson says or something like that. We use the AI and an AI platform to develop an ICP model. So the system would bump up, take the company, bump it up against the ICP and give it a score. We would also look at company-wide behavior. We have a solution that tells us, even though we don't have the actual name on the individual, is how many other individuals from that company came to the website.

13:15Chris Marin: That gets a score. We looked at the formal graphics of those companies. We looked at also a lookalike model that we also built that says, even if all those things are true, and again, each one gets a score, even if all those things are true, is this a company that traditionally converts to a win. All of those, plus a few more, without getting into too much detail, all of those give you a combined score. But that combined score, when it finally goes through the process, whether you close it or not, it informs the model again, and it makes adjustments because you're not going to get it right off the bat.

13:53Chris Marin: So let's say, for example, you score a certain attribute in a certain way, when it comes all the way through, it figures out, hey, looking back, did I score that attribute correctly or did that really not factor into the success? And if it didn't factor into the success, you adjust it and you increment up those variables that did contribute to the success. So it is a self-improving mechanism. And it happens with a model that is based on Bayesian math that in essence allows you to have a very flexible dynamic scoring mechanism as opposed to a static scoring mechanism. That's the biggest part of it.

14:35Chris Marin: Then, and more important, is all of those things. Consider now 10 inputs, but those inputs have sub-elements. So consider all of that and try to convince the salesperson that they should take that lead on. You're not going to ever get to that discussion. You're going to have to give them like 50 fields of information say, hey, check this out, and this is why we're sending it to you. No salesperson in their right mind would ever do that because they'd be burning time they don't have. So instead, we take all of those, we take the defining characteristics of all of those signals, what was really what made this AA-type lead, A1, A2, A3 for prioritization, and then we develop an AI-generated synopsis that goes with the lead and lands in the CRM.

15:25Chris Marin: So they get the lead, but it says in English, in prose, in a paragraph, this individual is in this title, in a company that looks at companies we love, and he or she did all those things. And by the way, given their persona, when you reach out to them, the first thing you should say is this, because this is what they really care about based on their persona. So you net it out in a single paragraph. So you tell the salespeople, if you don't read anything else, read that paragraph, number one. Number two, I'm telling you with high confidence, this is very likely to convert because that index I told you before, the scoring, is in essence a score of likelihood to convert.

16:11Chris Marin: Now, if you do this right, and by the way, you manage the expectations that it's not going to be perfect off the bat. But if you do this right, and they start to see that what you send them are high quality leads, guess what happens? Their confidence goes up. When their confidence goes up, what do they do? They react to the leads we give them faster. We all know that when you don't react to leads right away, consider leads to be almost fish. When you don't take fish and cook it right away, it gets stinky, right? So that's the same with leads. In the past, when you don't have trust between sales and marketing, they kind of gradually attack those leads.

16:49Chris Marin: They wait until it's too long. So if you do this right, their confidence is up. They attack those leads faster. And if they attack those leads faster, they're more likely to convert it because there's recency to the conversation as opposed to reaching out to the lead a week later, you reach out in the same day or the next day. So it becomes a positive contributor to the success. So you're doing two things at the same time. You're giving them high quality leads with good information that they can act on. And when they act on it, they gain confidence. Now confidence gives you recency. And the combination of recency and quality is what gets you a much better conversion rate.

17:30Nick Panayi: Adam here. Let's take a quick break. So if you're listening to Nick describe how his team combined multiple signals into a single AI-generated lead score, you're probably thinking about your own funnel right now. Most marketing teams don't have a Chris Marin on staff, someone who can connect an LLM to your CRM, build persona models, and design a self-improving scoring system from scratch. And we know this gets complicated. And that's exactly where health launch pads can help. We have a solution called AI enablement. We help B2B health tech marketing teams put AI to work on the problems that actually move pipeline.

18:16Nick Panayi: This can include things like filtering out the noise of your lead flow, building your ICP and persona models into your CRM and doing it in a way where sales can actually take actions. And so So it's generating kind of plain language lead synopsis that gets sales to pick up the phone instead of ignoring another notification of an MQL they don't have much confidence in. So you don't need to hire an entire AI team to get started. We can actually help you with that. And you can learn more through the link down below or just check out healthlaunchpad.com and what we do and you'll see our AI enablement services.

18:57Nick Panayi: Now, back to Nick. So it all makes intuitive sense in concept. In practice, did it make an impact?

19:06Chris Marin: It made a very significant impact. So let me just explain that again. These things, especially when you're breaking your ground, are not easy to do. But we spend, number one, we have a digital AI guru in the company, Arinobalan, who is my partner in crime for many years, who was the lead architect for this. And his name is Chris Marin. Chris and I got together and said, what do we need to do to make this happen? So from that whiteboarding session to a platform we can go test, it took like three months. We wanted to test internally. So we ran two systems parallel. We ran the traditional system and the new system, and we made adjustments and corrections as we went.

19:50Chris Marin: I would go in on a daily basis, and Chris loves me for that. I would say, well, what about we add this, and what about we add that as a practice and add this? So it took three months' worth of work to get there. But once we got there and we launched it, and we launched it, by the way, not in secret, not in hiding, but out in the open, with a combination of training for sales and very high visibility with sales leadership. And I said very clearly, we invested a lot of time and effort. So your part now is to work these leads as quickly as possible. And another thing that we did is we created an email channel that not only were these leads passed from marketing, Marketo, to sales system, salesforce.com, but they would also send an email.

20:37Chris Marin: And that email CC'd their boss and me and a number of other people. So we shed light on it right away. So there was high visibility and a communication that was very, very strong. So between January and basically the end of the first quarter, three months, we moved the conversion rate from 10 % overall across OBUs to 17%. So seven points of increased conversion. And when I say conversion, in our case, it's MQL to opportunity. Not even MQL to SAL, not MQL to accepted lead. MQL all the way to opportunity open. Seven points.

21:17Nick Panayi: So that's a 70 % improvement.

21:19Chris Marin: Massive, right? So think about it. Again, ROI is a little bit easier to calculate than, hey, I'm doing blogs a lot faster, and I'm putting more blogs out. It's a lot easier to calculate. Those extra 70 % of those opportunities is your ROI. It's as simple as that. So now that we have the cookbook, so to speak, that can be repeated easier in other organizations, but it took a while to get there. Very impressive.

21:47Nick Panayi: And how do sales react to it? One of the things I particularly like about what you described is that you're not asking them to change behaviors because everything's happening in Salesforce. So you're not asking them to go off and check something else. But how did they, you know, does salespeople notice the impact? Do they comment on it qualitatively?

22:06Chris Marin: Yeah, of course, you know, they loved it. Number one, they understand how much work and effort it takes to get there. They understood that. And number two, this is what they've always wanted. You know, again, they get semi-impressed with marketing awards and how pretty the website looks and all these white papers you put out. They get more impressed if you affect their wallet. So of course, they're positively predisposed. Having said that, they're a tough crowd, as they should be. So in the beginning, you had to kind of go prove it out. They would look at only one data point and say, why am I getting this?

22:40Chris Marin: Why am I getting this? This person is too low a level. But then we had the right answer in that we knew that those questions are going to come. We develop a dashboard where marketing has access to this dashboard, and we can go in and tell them the entire story of the lead. Sure, this person is only an administrator, but check this out. Five other people from the company have been researching. They look like they're part of the buying committee. The company is the right ICP. And their lookalike model suggests that this is the kind of company that could close very, very quickly. That's why you got it.

23:16Chris Marin: So we have a dashboard that you can go in and get the entire detail, but more so, and this goes into attribution, we're able to look at all the different touch points, not just the last touch point. That's one of the challenges in marketing is people hyper-focus on the last touch point and they say, why am I getting this? I'll give you a good example. There was one person, actually repeated many times, but the first time it happened, that registered for a webinar but did not participate. And the salesperson was like, why am I getting this lead? They didn't participate in the webinar. And we're like, That's exactly why you're getting the lead.

23:49Chris Marin: They made the effort to register and didn't participate. Then we sent them an email with the replay and they opened it. Oh, by the way, they did three other things before. And again, they're in the right ICP and everything else. So if you look at the entire picture, of course, they get, oh, now I get it. Thank you. I'll jump right on it. The other thing that was super, super valuable was that email. Because when that email would go out, one of my first things I would do every morning is look at those emails and send a little positive reinforcement and say, hey, we're looking forward to you jumping on this.

24:22Chris Marin: This looks really, really good from our perspective. And again, spotlight is on it. Everybody's on that email. So unless you have a death wish, you're going to go and reply to it because it's like the whole organization is looking and it's looking very, very hot. So, and oh, by the way, because we've copied your supervisor, if you don't jump on it or say you're out on vacation, the supervisor says, oh, he's out. Let me take care of it. It gives you a nice redundancy as well.

24:50Nick Panayi: The other thing I really like about this is the notion of being able to look back on a deal. So we hear this in the, it came up in the HealthTech Marketing Network and we actually had three CMOs, VPs of marketing, talk about how they create a deal story. And every single one of them had had a lot of work just to create one single deal story that they could show the board. And when they did, incredibly effective. I can see how that would make it much easier to look back on a deal that went from MQL to opportunity and all of the different touch points. So what was it that got them to move to that point?

25:29Nick Panayi: So I think there's a lot of applicability to that. Yeah.

25:32Chris Marin: And to your point, very good point. That was an enhancement that I didn't get a chance to make because I left the company. I know Chris would be all over that. In terms of the data is there, it's a simple extraction of that data. If somebody asks, it says, here is almost like a histogram of all of the activities, but it's also good to reward positive behavior. So in the beginning, I did that manually because I knew how important that is. I took the time if a salesperson jumped on it right away. And by the way, they would say, thank you, Nick. I appreciate it. I'm jumping on it right away. And the next day because they got super excited, said, you know what?

26:11Chris Marin: It was a real deal. We opened up the opportunity. I took those and I'm like, great. And then I would take all the touch points and say, here's what happened. They came to the website, webinar, downloaded. Then I sent it to this amazing salesperson who reacted to it right away. And then we have an opportunity in three, five days of activity and we have an opportunity. Please do more of that. So the fact that you have that data is not only good in a defensive way to basically go explain why you did something. It's also good for positive reinforcement. And people are like, wow, I didn't know we have the entire history of that need.

26:48Chris Marin: And I'm like, yeah, we do. To your point, you can build that as part of the solution, press one button and show a histogram of everything. That could be magic, right?

26:57Nick Panayi: Yeah, absolutely. Yeah. Okay, Nick, if a CMO listening to this takes away one idea and wants to get started with it, like get started next week with it, what would that be?

27:11Chris Marin: Get started right away, because even though you may not have 10 triggers, you'll have two, you'll have three. It's better than one. Even if you have two, it's better than one. Start to create a system like this. It is, again, not for the faint of heart. It's not something that I or any general marketer can go build because it requires technical expertise. It requires somebody who understands the marketing stack and the whole lead flow, but also understands AI tools. because what I didn't explain in all of that is behind all of the steps, there's an AI element. For example, putting together the synopsis, the story of the lead, that is AI generated, right?

27:53Chris Marin: You're not going to have a marketer sit there and craft prose every single time. Somebody needs to know how to connect an LLM to that. Somebody needs to go teach the LLM what areas to go look at and synthesize these prose. another area we used LLM was persona alignment okay for you to know if that particular title in that function for that individual what the alignment is to persona you need to have pretty well established persona in a place where you can actually act on it not like somebody says oh this is generally a persona it needs to be in a format then the system can go connect title and function with persona and look for alignment and apply a score to that.

28:37Chris Marin: That's AI driven work. Actually, building those persona was AI work. Building the ICP was AI work. In that particular case, we used an actual outside vendor. But you got to know all these things to be able to connect them. Once you build that Bayesian model, flexible, interactive scoring system, then it scales from there. You don't build anymore. Then you're just making tweaks and adjustments, but primarily done by the system itself because the Bayesian model itself is designed to be flexible. It's designed to be self-improving over time. So once you build it, it will improve itself over time.

29:16Chris Marin: And again, that's the benefit of having AI. So get a Chris is what I always say. A Chris is equivalent of somebody who is a strong AI digital marketer and a strong executive advocate and go start building this. The advantages are significant.

29:34Nick Panayi: I could see how you could at least do a limited sort of manual pilot using Claude or ChatGPT, just pulling in the different points of data. And so you could create a Claude project and do it manually. Now it's a pain in the neck, somebody's got to actually do it, But at least you can do that to see if it's worth then investing the time to go. Because if you don't have a Chris, you're going to have to hire somebody or hire a firm to help you do this. But I can see how you can pilot it. So it was great. Nick, thank you. This was great. I appreciate you going into the level of detail. And good luck in the next gig.

30:15Nick Panayi: Decided to hear more about it.

30:17Chris Marin: Thank you so much. And the last word is exactly what you said. You can do a proof of concept relatively easily by getting just a couple of variables and starting with that. If you need any help, I'm sure you know people and I know people who can help others do this. Reach out to us and we can hook you up with the right people.

30:36Nick Panayi: I will put your LinkedIn information in the show notes down below. Fantastic. So people can follow you and bombard you with connections. I would love that.

30:44Chris Marin: I would love that.

30:45Nick Panayi: Great. Thank you so much. I love having Nick on the show. he's so passionate about the impact AI can make. And it's truly amazing what he's accomplished in the last year or so with his team. So I've got five key takeaways from this show. So number one, traditional lead scoring is single dimensional and hasn't changed for years. AI lets you combine multiple signals, individual behavior, company-wide activity, ICP alignment, firmographics, and lookalike modeling into one dynamic score. Takeaway two, a Bayesian self-improving model beats a static point system because it keeps adjusting itself based on which signals actually predicted a win.

31:40Nick Panayi: not just which ones you guess would matter at the start. Takeaway number three, don't just hand sales a number. Nick's team paired each score with a plain language AI generated synopsis explaining why the lead matters and how to open the conversation, which is what actually gets reps to act fast. Takeaway three, visibility built trust. Copying sales leadership on every qualified lead created accountability. And that trust is the same ingredient that makes ABM and broader pipeline programs work when they work at all. And the last takeaway is you don't need all 10 signals on day one. Start with two or three triggers using whatever LLM that you're using right now to prove out the model and let it expand from there.

32:34Nick Panayi: If you want to connect with Nick, His LinkedIn contact is down below. And we've also left a link to the earlier episode with Nick on how he and AI enabled the marketing team and the marketing system at Inovalon. And if you want to learn more about our AI enablement services, there's a link to that down below as well. Now, that's three episodes into our AI and sales enablement series. so Bruce showed us how agentic AI can accelerate the sales motion itself. Jennifer Lazar showed us what it takes to manage the change internally when you bring AI into clinical and commercial workflows and Nick just showed us how to make sure the leads feeding that whole system are actually worth the sales team's time in the first place.

33:27Nick Panayi: Next time on the Health Tech Marketing Show we're going to continue with this series and meet Tim Hughes who's going to talk about AI and LinkedIn for sales enablement. You won't want to miss this. It's really a great episode. See you on the next episode of the Health Tech Marketing Show.