33 Minutes 
Mark Lerner:
Everybody, welcome to this episode of the Revamp podcast. My name is Mark Lerner, director of Growth Marketing here at Deal Hub, and I’m your host and joined today by DeMar. We had the opportunity to meet recently at an event in New York City, and we had some really interesting discussions. We wanted to bring those to our audience. But before we jump into some of that, why don’t you go ahead and introduce yourself to the folks at home, maybe tell ’em a little bit about what you’re doing and kind of how you got to where you’re at in your career.
Demar Amacker:
Yeah, absolutely. So, thanks for having me. Mark DeMar Amer, director of Revenue Operations at AccessiBe. As you just said, I’ve been with the company for about five months now, so really still kind of getting my feet with all of the things we have going on from a Go-to-market standpoint here. But in my role, I oversee our Rev ops team, which includes a Rev ops analyst, a Salesforce admin, and a Salesforce developer. So, it really grew up as a sales ops function, and we’re transitioning more into a rev ops function here at Vee. So exciting times. We just brought on a new CRO, so definitely an exciting time to be part of Vee, which is a web accessibility platform that really aims at making the web more accessible. So we’re knee-deep in quite a few projects right now, but looking forward to chatting more.
Mark Lerner:
Yeah, awesome. Yeah, so when we met the other day, there were a few topics that we got into, but one of them that I think we really got deep into was we were sitting with a group of people and I was asking everyone if, in RevOps they’ve yet to see actual use cases where AI has been valuable beyond kind of the parlor tricks that people get interested in, but really be able to use it at scale. And you and I were talking about some of the different things we have been doing, and so maybe we can dive back into that. From your perspective, as somebody in a Rev Ops role, it especially sounds like there’s a transition happening where you’re moving Rev Ops to take on a more prominent function. Does AI play a role in your work today? From your vantage point, where do you see its value for someone in a rev ops role?
Demar Amacker:
Yeah, I mean, I think AI is, obviously, the buzzword for at least Gen AI specifically. It’s been really the buzzword things like chat GPT really taking hold over the last year, year and a half. So, as an early adopter of Gen AI in my role, I’ve been trying to find the right application for different projects that I’ve worked on. Is it better used as a time-saving tool? Is it better used? Is there a way that you can start to get the wheels turning on things like sales enablement content? Is it a good way to sort of summarize things like calls and whatever? Obviously, every platform now, from a tool perspective has some lean into the AI world, whether it’s Gong or Salesforce or whatever tool you have in your stack that is leveraging AI in their native capacity. But I like to use the chat GPTs of the world, the Claudes of the world, to really help me force multiply myself, and my staff obviously uses it in that capacity as well.
So things are as simple as taking a Salesforce formula or a row-level formula that is applied to a report and making it more efficient, or maybe the syntax was incorrect when we originally designed it. Maybe just fixing it in that regard. It’s really, really great with error handling. So, if you just feed it something that you’re seeing, whether as a Salesforce error or from a syntax error in Excel, it can usually help you modify your query so that it becomes something that’s obviously more usable and obviously giving you the output that you’re looking for. So I use it to short-circuit the sort of code dictionary that I don’t have to have readily available in my brain as far as knowing the exact syntax sometimes.
Mark Lerner:
So you use the sidekick aspect of it often; it sounds like that, just,
Demar Amacker:
Yeah, when sales obviously, or whenever sales ops, Rev Ops are oftentimes operating very leanly, you sometimes have to learn on the go so that it does help to bridge some of the gaps of how would you do this? Or if you’ve already done the work to do a lot of the documentation, make this more consumable for my end user, make it something that turns it into a fun training or a fun exercise. So I know our sales enablement team has used it to create training dialogues and coaching out of transcripts and things like that out of Gong to give that AI assistance for our very busy sales leaders.
Mark Lerner:
Yeah, I think there are two types of, well, there’s probably more than two, but two that go through my mind of the way people use AI. One is, I think, the way you’re talking about, which is your assistant, that helps make you more productive and democratizes certain information. So if you’re not a dev person, you can get code questions answered and do some of that apex stuff that maybe you wouldn’t have been able to or somebody on the team wouldn’t have been able to. But then there’s also using it within automated workflow to do something for you so that you don’t have to hardcode if and/or have an AI in the loop when making those decisions. That’s the use case I find most exciting, but it’s also the one that’s most complicated because it’s very difficult to ensure that it’s doing the right.
Demar Amacker:
Thing. And it obviously takes a ton of testing and validation, and it is obviously prompt engineering. But one example that I have of that is, that the accessibility business model is very sort of 50-50 partner and sort of direct. So when our inbound SDR teams have to what we call vet a new inbound lead, it can oftentimes take somewhere upwards of a minute to five minutes to determine if this account is a good partner lead for us or is this account is more of a website owner direct lead. And what we’ve actually done is taken AI and applied a script that I had, Python script that I had chat GBT helped develop for me that essentially uses website domains that are entered into a lead form creation of an account to then call out to that service for chat GPT to essentially read the domain that’s been inputted, look at the website for things like keywords, things that we determine as this would be a partner or this would be a direct client, and then classify that with roughly 90 to 95% accuracy. So that last 5% of the job is really all we’re saving for the human intervention to come in and double-check; oh, is this a likely partner, or maybe this is a likely website owner? So it’s that fringe where we can’t make the full distinction and then it saves 90% of the time on the classification side so they can move on to things like the personalization of their emails, other more strategic sales outbound approaches that aren’t just, is this or is this that? So again, using AI to assist.
Mark Lerner:
And that particular use case, was it something like God, somebody’s got to figure out how we can do this better? Or was it just one day you’re like, Hey, why don’t we just try this AI thing? Was there somebody who said, we need to operationalize this, or was it just one day you got fed up, and we’re like, let’s see if we can do something about this?
Demar Amacker:
So it came to my desk, and we realized our SDRs are spending, where’s the 80 20 Pareto of where our reps spend the most of their time. And we found that it’s in this vetting process of simply.
Going to a website and determining what type of, because we do a lot of inbound leads, so for everyone that we get, we want to touch it and give it the right care, but sometimes the process of determining if it’s a real lead, if it’s going to go to a 4 0 4 site, things like that, obviously for every website that they’re going to, it just bogs them down, and it makes our speed lead slower. It obviously makes the user experience suffer when we can’t get back to them as quickly. So by simply using chat PD to evaluate this domain with publicly available information that’s right there on their website, we can pretty effectively classify these accounts coming in and then that’s a whole aspect of the vetting that’s already pre-one for them, which saves a ton of time for them. And then we can leverage that in things like our lead orchestration and assignment rules so that at least we’ve sort of done a lot of the legwork for them before they have to obviously do the rest of the work.
Mark Lerner:
So that the majority of their time was spent on this one task that was kind of a useless task or not useless, but it was busy work, and you were able to basically automate it and save a whole bunch of time, which is I think the perfect example of the value of ai. I think often people get so abstract about these huge things that they lose track of the simpler things that you could basically fix instead of outsourcing it to somebody to do that busy work or forcing your reps to kind of do that.
Demar Amacker:
And the other piece of it, is it effective, or obviously we had to prompt engineer it to get to a point where it was first started out at 65, 70% sort of likelihood. So okay, well we’re saving 70%. Could we rat that up all the way up to 90, 95%? That accuracy was something that was really viable and sought after for our sales leaders because the more effective the AI could become in reducing that time, the more efficient our reps could obviously be with it. So it’s that balance of if we were only able to effectively classify 50% of these using an AI script, maybe the earlier models could have done it, but obviously, it’s not as effective. But as we doctored the prompt, made it exactly what we wanted it to be, tested it, and had human intervention there to assist it, we were able to essentially do what a human could do with enough obviously adjusted inputs.
Mark Lerner:
So I wanted to get your input on this. I’ve been asking people a lot about this. I recently read an article, are you familiar with the company Klarna?
Demar Amacker:
Yeah,
Mark Lerner:
Yeah. So their CEO and their kind of people are unsure if he said this to create a buzz or if it’s legit that they’ve essentially deprecated the use of enterprise software like Workday and Salesforce, and they’re basically only using AI and knowledge graph database essentially that over time they basically were able to create this internal database that was able to make these connections. Essentially, people have, whether it’s customer success or sales, have a chat interface, and they can kind of get this information when they need it. He didn’t give that much detail, and it got me wondering, do you foresee a world in which all of these interfaces that people interact with, whether it’s the CRM or the CPQ or the billing system, whatever, that there won’t be those things anymore and will only be interacting with the chat interface, is that a world you think will ever come to be in the next 10 years?
Demar Amacker:
I mean, it’s certainly an interesting concept. I think the way that our brains work in structures means that there’s always going to be a need to put things in nice, neat piles. That’s why relational databases are so effective: they are able to easily colonize and measure things. So I think for things like knowledge bases and things like that, customer information, I am really curious to go to Salesforce’s world tour event next month because it’s all about agent force and all about AI-driven chatbots and where they can leverage existing customer data with your customers’ context for better user experience in place of obviously a human chatbot. So to me, that’s an interesting application for basically interfacing with an Amal sort of set of information. Obviously, it’s made easier when you can colonize it and organize it and sort of label it, tag it, and whatever. So I think you’re always going to need it to be something that’s at least somewhat orderly so that you can do things like board reporting and all of the things that relational data was bred out of. But I do think it’s a fascinating concept to be able to leverage the context of your existing data stack across multiple systems so you can tell a story or derive a narrative that is obviously more of a 360 view than just a single kind of picture or a profile on a database.
Mark Lerner:
For sure. And that’s the kind of thing that fascinates me. I don’t know if normal people are interested in that kind of thing, but I am very excited about it. But living in the world we live in now in which Rev Ops empowers go-to-market teams and often as kind of the product developer of those solutions that they’re working on, how do you think about, as you said, orderly giving the end users of those solutions, the visibility they need without the signal from the noise and how you create a workflow for them or enable workflow for them that’s efficient?
Demar Amacker:
I mean, I think it has to come back to something that is tangible for them because simply, I think that’s the thing that for nonpractitioners of gen AI tools, the thing that is sort of, I think, intimidating for them is what is my use case? Could I even apply this to my job? And I think that that’s the sort of learning curve that early adopters got in on pretty early is, well, it is sort of as good as the prompt that you derive, and if you can dream up a use case that you could apply it to. My wife’s a pastry chef, and we’ve experimented with things like menu design and dish creation, and I think that should be saved as she’ll appreciate more of a creative kind of process and development of obviously flavors and things like that. A chatbot can’t do that. But I do think that curiosity is really the key, being able to apply AI to your job and make it really a utility for your professional, personal, whatever application you have because at the end of the day, as users test the limits that drive the need for improving models. And it’s kind of the one situation where we’re as impactful to the design and future of it as the developers are because we’re constantly driving the learning of the model.
Mark Lerner:
It really does feel like a blue ocean in terms of what’s possible. And I think that’s part of the challenge. It’s the cold start problem. You’re looking at a blank screen and a blinking thing and you’re like, okay, now. However, if you go to an interface, you have everything there, and you can, so yeah, it’s a challenge. But stepping away from the AI stuff for now and talking just kind of the traditional tech stack, I think we’re in a time, we’re coming out of a period where, especially during the peak of the pandemic, when money was kind of falling from the sky go-to-market teams kind of had the pick of the litter about whatever tools they wanted, and things got pretty bloated. And I think we saw a period of consolidation and contraction where people were looking to consolidate their tech stacks. And I don’t know if we’ve normalized, but when looking at the tech stack in your organization or any organization that you’re working with, what is the filter with which you look about deciding to build or buy? Do we need a new solution? Can we optimize our existing tech stack and actually deprecate some solutions? What do you think about all of them?
Demar Amacker:
So, I think it really comes down to not losing sight of what you’re ultimately trying to accomplish at the end of the day. Because if you’re simply just getting tools for tool’s sake, you’re just trying to keep up with the Joneses, you’re never going to have success. I think that way because I’ve said in talks before, I don’t think that software is really a silver bullet kind of thing where you can just put it in, and it will solve all your problems. I don’t think anyone believes that today, the solution is the software. I think the solution is surrounded by all of the things that make it successful, like the planning and the preparation and obviously the key influencers in the project and obviously the good project management is important and all of those things when it comes to a successful implementation and rollout, adoption and reporting and transparency and accountability of obviously the end goal.
But I think if you have to not lose sight of why we get this thing, this software in the first place, and if we get this, what is the ripple effect of that, right? Is it going to create redundancies? Is it going to obviously create a situation where we can’t integrate with that other tool? So, does it play nicely with the rest of our stack? So I think all these are considerations that you have to take into account. I think ops is very well positioned to do this, but sometimes, we don’t get brought in on the evaluation until later in the process. Or maybe at the very end, when it’s kind of a final check of approval like this is going to touch Salesforce, does this work for your system? But it’s like Salesforce touches a lot more. So the spiderweb and this wheel spoke of what is impacted, which is very sensitive and very finicky.
So I think the earlier on Rev ops has brought in on the decision-making of the tool stack, the vision for what it’s going to ultimately add, and obviously the perspective that is added there that what is this potentially going to block or remove as well. Because I think a lot of times, people don’t think of the subtraction-by-addition effect of adding new software, but adding one more pile of data requires management and maintenance and obviously upkeep. So if you’re just adding a system that’s going to collect cobwebs and no one’s going to really manage that process, you’re obviously setting yourself up for failure. So, all those things to say, you have to have a clear set of outcomes and visions in mind from the very beginning. I think that starts with tool selection and vendor procurement.
Mark Lerner:
That was another conversation that we were having in the group. They were talking about tool selection and implementation, ensuring successive implementation, and how to ensure that there aren’t implementations that don’t go through. And so maybe we could double click a little bit on what you were talking about and from a rev ops perspective, assuming that you’re in the room and you are strategic rev ops and you’re given the opportunity to have a say, how is the vendor selection and implementation process made in a way that you can maybe not ensure, but as best as you can, make sure that you cross your i’s and dot your T’s so that you don’t get stuck in a broken implementation process?
Demar Amacker:
Yeah, I mean, if there was a silver bullet for this, I think every Rev ops team, every project manager team across the board would definitely be signing up for it. But I think it starts by having realistic expectations of oftentimes implementations have phased approaches of this is our minimum viable, this is our sort of phase one, phase two, this is sort of the pie in the sky vision for once we get this all rolled out. And I think that people lose sight of these milestones and stop to say and evaluate and say, okay, did we succeed in, we obviously talked about it in sales, a lot of getting to these sort of win milestones, and I think that’s important for project management and implementations as well, is that sort of inflection point of, okay, whether it’s in scoping or in vendor selection, I think staying lockstep with both the end users, the sort of managing leaders that are obviously going to be responsible for standing up the tool and obviously the admins that are going to be working through it day to day, making sure that all of them are clear cut on what the expectations for the tool are.
Something as simple as rolling out a new calendaring tool or a new dialer. Where does the data go? Is this going to cross wires or something that exists already? So, I think that gathering dependencies and requirements is critical to the very foundation of the project. And then using that as your compass throughout and making sure that people, as things maybe fall off the priority list and as vendors maybe get eliminated, making sure that you’re always centering back to, okay, does this solve what we set out to solve? Because if it doesn’t, then just simply sticking in a piece of software that’s a sort of round peg square hole doesn’t really, be a good investment of our money. So I think we get in these positions where it’s like, oh, we’ve committed time to this. We’ve sort of done the evaluation; we have to make a decision.
It’s like, well, if none simply standing patent and making a decision of no, because it’s not the right time or we don’t have the right resources is an okay thing as well because, at the end of the day, the wrong decision could set you back further than a no-decision. So I think that simply making the bold choice isn’t always the right one. Sometimes you do have to. Action is better than inaction, but I think you do have to be careful of setting off on an implementation that maybe there’s a changeover in leadership or things like, are we going to have a consistent, are we going to be set up for success throughout this next three months of this implementation? You might have leadership changes; you might have turnover within the implementation team. All these things could buck the success of it. And I think you have to have that foresight as the Rev Ops person or the project leader. What’s the worst-case scenario here? What does losing our project sponsor tomorrow mean for the project? Anyone who’s gone through enough of these recognizes that you can’t obviously predict for everything. But I think that in your mind if that is something that is a possibility or could be a derailer, how do you get through that if that is the case, right? Because it’s not the end of the world, but it is something that needs to be solved.
Mark Lerner:
So oftentimes when people are looking at bringing on a new tool, if that’s the case that they go to, they had started with some sort of internal solution, Google Sheets, whatever, tying this back to ai. Do you think that with these tools and the democratization of knowledge that they offer perhaps there are companies now can get further on those internal tools before they have to make a decision? They may squeak out a little bit more juice of that crazy spreadsheet before deciding they need to use Salesforce.
Demar Amacker:
I mean, I think so, right? The original sort of tools like Excel and obviously Google Sheets and those tools are really kind of bread and butter analyst type tool, analytical type tool. So if you can squeeze more out of that lemon, ultimately, you do have to do backflips in MacGyver sometimes as an operational leader operational professional because there’s a job to be done or an outcome that needs to be solved. However you get there, maybe that end user doesn’t care; they just want what they want, and they don’t know how hard it is to get there. And so if you can use tools like AI to make Excel easier to understand for you, or just have something like a Python script that does three steps instead of five automated, to me, that is where AI can come into play and make you look like a superstar when all you had to do is ask the right question to the chatbot, test it and validate it, and then give a product that would’ve taken potentially prior months and months of development and a resource outside of your purview.
So I think that’s kind of how this sort of partner direct tool came to be. I bet ChatGPT if I gave it the right prompt and built it up from what the original kind of, could it do this little thing? Okay, could it do this little thing in an automated fashion? Could it do this automated thing in an automated fashion and production-sized without having to feed it a chat script? Could I just call the API directly? So, as you mature your solution, I think using AI to make it even more sophisticated is a great application of AI.
Mark Lerner:
Awesome. Alright, DeMar, thank you so much for your time today. As we close things up, we’re kind of heading into the final quarter of the year. We have 2025. Looking at us down the barrel here from a rev ops perspective, how do you see 2025? I know this is very broad and asking you to be Oracle, but what are some of the big changes to look out for, I think, on the Rev ops side in 2025 that will impact folks in this space?
Demar Amacker:
You said you think it’s sort of normalized and settled down, but I don’t think we’ve reached the full maturity of the; I think there’ll be another big consolidation that will shock us sometime in 25. I think there’ll be more obviously exciting applications for AI. I’m really excited to see how the sort of leveling of the playing field when it comes to these different AI models goes because everyone is, I think the rising tide, lifting all the ships is obviously a good thing, but it also, it’s another year of experience using these tools. It’s another year of pushing the envelope for the early adopters. So I’m excited for obviously just the community of rev ops as well. We have great groups out there that are going to be going into some of them in their second year of having a conference. So I’m excited to see where these really amazing communities take the sort of ops landscape just from a community standpoint as well because I think there are a lot of really great leaders in this community, and they’re starting to finally get the recognition and public medium and space to be that force. So excited to see where my Rev Ops practitioners and leaders take the community next.
Mark Lerner:
Yeah, it’s interesting, right, because we’re putting so much focus on these generative AI models that are decidedly unhuman, but then there’s also this real interest in genuine human interaction that has a real value, maybe even more so now than before. And so that’s an interesting thing. Maybe the more we rely on interacting with AI, the more we need that human interaction. I think that’s a whole other conversation for another day.
Demar Amacker:
Yeah. Well, I mean, I’m sure I will definitely see you at the next event soon, so I can’t wait to catch up on that. But yeah, I’m excited to hear more about all of the exciting things that Deal Hub has come down the pipe as well.
Mark Lerner:
Yeah. All right, DeMar, thank you so much for taking the time to chat with us. Really appreciate it.
Demar Amacker:
Thanks.
