GES Lead Scoring with David Moreira

Cool.

All right.

Well, maybe we should kick it off and get going here. And we've got the recording so the other attendees can receive a recording of it.

Sounds good.

So yeah, welcome everybody. I mean, whether you are from— whether you're a current Go-To-Market Engineer School student, I see a few of you, or a graduate, or a newsletter subscriber, or part of the Automate RevOps community, Glad you're here. So I run Go-To-Market Engineer School. And, you know, if you look at almost any go-to-market engineer project or RevOps project, what are we actually doing? We are enriching, we're filtering, and ultimately we're scoring. And then, you know, and a lot of our capstone projects involve, you know, our students are building right now involve scoring in them as well. But here's the thing. If the score itself is only as useful as sales is likely to believe it, buy into it, and use it. And anybody who's ever delivered, I think, a lead scoring project into a sales organization knows that this is, I think, the riskiest part. Will it be adopted? So I think we've all gotten that wrong at some point and where reps just quietly ignore it and, you know, you hear back channels if they don't believe it, you know, same way they could ignore MQL. So I'm especially excited about today's presenter and to learn his framework for getting this part right. David Moriah is the founder and go-to-market systems architect at Automate RevOps, where he builds signal-based demand systems for VC-backed B2B teams. And the kind of signals that capture, that get captured and a rep would actually use and the scores on awareness and readiness get incorporated, not just the activity. And so really excited to see his framework. David is a certified CLAY partner. He teaches business strategy at a university level and he's the co-author of The Demand Compass. So David, the floor is yours, and we'll take questions in about, you know, a half hour. Take it away.

Thank you. Thank you for the introduction, Jared. Um, let me present my screen. All right, here we are. So thanks for being here. Quick promise before we start: this is not a build-along. You will not leave with a template to copy. You will leave thinking like the architect of a lead scoring system that sales actually uses. Because most of them get rejected and it's almost never engineering's fault. So Jared already introduced me. So just some things to mention. We built the Demand Compass with Marcos. He's also in the meeting here. We named the framework after we implemented with different clients and started knowing that it actually works and it gives the CMO and the team some resource to actually know how they're progressing towards having a great system for sending leads to the sales team. So we build signal-based demand engines for technical buyers, mostly cyber dev tools, AI infra, And the system that I'm about to show you actually is the one we ship for clients. We started in 2022 as a cold email agency. It worked until it stopped working. So the lesson was blunt. Activity is not a system. So in like 2 years ago, we rebuilt the whole thing as infrastructure using Clay. That became automated RevOps at that time. We started coaching at that time. But then we started implementing as well. So two founders with a small team and cloud code, of course. No outside money, working with teams in the US, Israel, the UAE, and Europe. Just one quick thing before we dive in. These talks come from our book, The Demand Compass. It launches August 5th. So if you want to log in for the launch, just scan the QR and you can save a spot. And stay until the end because I have a surprise for this room and it's worth staying for. I'm gonna leave it 2 seconds more and I promise that you're gonna have the link as well later. Awesome, so let's move on. Here's the moment all of us know. You get the brief. You build a lead scoring system, you wire it up and the data start flowing, the score computes, it lands in the CRM. It's a clean build. But then you watch the sales ignore every lead they produce. Like everybody. We have been there and I think some of you have also been there. So this is an issue that is regular and that's why we focus on building the solution. But the experience goes like, the VP tells her reps, the leads are junk. And we have been there as well. They stop opening the queue and the thing you ship is now shelfware. You did the engineering right and the leads were not all bad. I mean, so what happened is that actually the problem was not the plumbing, it was a model you were handed. So the model is the MQL, and the MQL is one number pretending to be four. Do they fit? Do they know us? Are they already buying? Is this the moment? I have four different questions crushed into one single score. And the worst part is that it treats awareness and readiness as the same thing, and they are not. Someone can read 3 of your guides and never buy. Someone can be ready to buy right now and have never heard of you. One number cannot carry both, so it lies. And reps, they learn how to ignore liars. So here's the real failure mode. It's not accuracy. The reps had no reason to believe the score. There was no why attached to it. The first job of a scoring engine is not to be right, it is to be believed. A score everybody ignores is worth zero, however accurate it is. So the question I want you to design around is not, how do I score better? It is how do I build a score a skeptical rep trusts. So the fix starts by splitting the one number back into two that matters: awareness, which you develop slowly over time— I mean, brand awareness— and readiness, that is readiness to buy, you know, which you detect in a window. You put them on two axes and you have four quadrants. Each one gets exactly one play. So you have in-market, high readiness, low awareness— you need to move now because they are looking for a solution and they don't know you.

Nurture—

they know you but they are not buying, so keep developing and keep nurturing them until they start to buy, and they will think about your brand. Then you have sales ready, both high. You need to sell, you need to push them, push your team to reach out to them, start selling them. Then we have cold. There is the starting line. They don't know you, you don't know if they're buying. I mean, we have different plays for each of them. And this is the model that MQL flattened into one user's number. So two decisions, two design decisions hide inside that model. First, score what accounts do, not what they raise a hand for. A serious buyer does almost all of their evaluation before they ever fill in a form. They read your docs, they start the repository, they post a job that names your stack, they hit your pricing page 5 times in a week. The form fill is the last 5% of your journey and the noisiest. And second, score the account, not the person. The engineer reads the doc, the VP approves the spend, procurement hits pricing. So the person leaving the trail is rarely the person who signs. Reason at company level or you will be misreading every signal. Now, the part you came for, um, the architecture. I'm going to give you 7 rules. Same brief, 2 systems. At each rule, the easiest default builds the score Sales rejects. The other call builds the one that survives a real pipeline. For each one, the default, the call, and a real example. So let's start with the first one. Rule number 1: the Signal contract. The default: every tool dumps data in its own shape, and you glue it all together per source. You end up with a brittle pile that breaks every week. They call one normalized contract that every connector writes to. You throw connectors away— I mean, tools are expendable. You keep the contract. So Identity results on a waterfall. LinkedIn URL, for example, for people, then email, then domain. Every merge logged so bad roles get quarantined and never drop in silence. So as an example, one company shows up 3 ways: ACME Inc. in your enrichment tool. Agmeing. on a Gmail address in your CRM, agme.io on the job board. Without a contract, you score them 3 times and you route 3 reps into the same account. The waterfall merges them on the domain into one. This is the least glamorous slide and it decides whether everything else holds. Now, let's go to rule number 2. The deterministic score, AI at the edge. The default, and it's everywhere right now. Let an LLM read everything and say, looks like an A. It cannot show it works. It cannot be replayed. It answers differently on Thursday than on Monday. It looks like magic in a demo, but it's poison in production. The call— the scoring is plain deterministic arithmetic. So same signal should be same weight, should be same score every single time. So AI lives only on the edge. One hard question that needs reading, not counting. Is this account really building with AI agent? If that's the question, then you can drive an AI to actually go and fetch the data from the site, from the post, and tag them true or false. I mean, this account building agents based on what they're saying on their website. Was there a real breach here? I mean, a data breach, for example, and you give them data access to a news or to some piece of post to read it throughout and give you back a true or false match. Now the model returns one tag and one subscore. And you use it as an input in the score system. It's never the score. For example, same account, next quarter, we have the same signals and the rep sees the same 8 and you can replay every step. The LLM version scored 8 on Monday and 5 on Thursday. And nothing changed. So you need to build a trust for the reps. In order to build a trust, you need to be able to replicate how you score them. So the gap right there is what actually kills their trust. Rule number 3: Auditability is a product. So the default, you optimize the score You spend a lot of time into crafting different formulas and ways to have like a big process to actually score them right. But you hide the math. Decoal. Every score breaks down into named, dated signals in plain language plus tags. So the routing never says, "Hot." It says, "Why?" A rep opens the account and reads the case in 4 seconds, and the rejections that kill the MQL turns into action. So there is a second audience, and for us, it is not optional. In our case, we sell to security buyers. They run a vendor assessment on you. A black box fails that review on site, a score that decomposes into named dated signals, pass it. So for example, let's call this company Norwind. The new VP of sales, um, started 12 days ago. They visited our pricing page 3 times this week. Also, they check our docs twice. Readiness is 8, awareness is 4. So they are in market. A human can read that and believe that. Different is saying, "This lead is hot." And it's difficult to trust a system that cannot prove why it's going that way. Rule number 4: a signal is evidence with a half-life. So the default, a signal scored once counts forever. Your hot list fills with accounts that were hot last year. Now they call readiness decay, decay is fast. Awareness decay is slower because recognition lingers longer than urgency. So for example, we had a funding round from 14 months ago, still flagging an account as hot. With decay, that drops to near zero and leaves the queue. A run from last month counts near full in that case. So let every signal count forever and you don't have a Compass, you have a graveyard that glows. So we need to avoid doing that, and you can create your own decay system depending on the client, depending on each signal. Now, rule number 5: build capture and creation, both. Demand capture finds the signals that already exist and routes it. Demand creation builds the awareness that makes the signal appear in the first place. So the default by any intent tool, we have all been there, like hiring RB2B or some data that is collected, uh, reaction from your LinkedIn post. You wind up rooting, then complain that there is no intent in your market. Almost always the market is fine and you only build half the machine. So as an example, a client swore there was no intent in their niche. And we have all been there and see that. They had no awareness engine at all. They weren't creating content. They weren't posting blog posts, LinkedIn posts, anything. They weren't running any cold email campaign, anything. So we turned on the content and found they're posting to their ICP. A quarter later, the same brother started lighting up. Because now the accounts knew who they were. No intent in our market is usually a confession that only half the system exists, so you need to build both. Rule number 6: one play per quadrant and root with context. So send a full signal trail to whoever owns the account. They open the deal already knowing why it is worth a real conversation this week. And one hard rule on the message: the signal tells you which problem to open on. It never appears in the copy. Do not write, "We saw your job post," "We saw you visiting my page," um, lead with the problem they actually have. Naming the signal makes a stranger sound like they have been, uh, being stalked, and people pull back from that. I mean, nobody wants to feel like you have been monitored. So as an example, an in-market account goes to the account executive with a why now attached in a message. It could be a Slack message, it could be a note in the CRM, whatever they are used to. A new CISO, a breach in their sector, free visit to pay to, to the pricing page. The opener leads with the problem. It replies near 3 times more than a cold one. Sits around one. So let's, um, talk about the, the last rule, the rule number 7: the human in the loop and learn honestly. And that's something that, um, I heard a question before, uh, I think David from Barcelona, um The default, full auto and let the model learn from whatever clothes. So right now you can use AI to increase or improve the model every time. So first thing, you need to capture signals all over the time to keep having more data. But once you have the data, you cannot trust the model like the AI that built it or the AI to analyze it. You need to make the calls. So there are 3 parts. The wait, learn, but offline, propose, shadow test it, approved by a human. The scoring itself stays deterministic. You keep a small holdout though, like 5 to 10%. You route it regardless of the score. Even if they're cold, you route them to the elites, so to the reps. So that holdout is the only thing that proves the score predicts anything. Instead of you marking your own homework, that's what most of us are doing, and you keep a human on the moves that are hard to undo. Automate the rest. For example, um, you scored hot accounts convert at 25%, the holdout converts at 5%. That gap means the score actually works. If they convert the same, your scoring system predicts nothing, and only the holdout would ever tell you that. So you can architect all of this and still lose if you cannot sell it inside your company. So 3 moves I would recommend you: win the rep first. That means you need to listen to them, you need to be close to them, understand their needs. Second, auditability plus one small win. So you need to start making them feel, uh, that you're actually listening to them and providing them what they need, what they ask for. And never a year-long build. Give them a score they can read, an early win they can feel. Win the leadership team by showing closed deals as timelines. So if you can show them all the track record on why we reach out to that lead, how it came from cold to aware, and then they start showing us intent and we reach out to them, that's a lot of value that you can provide to, um, to the company, and they would see it. The month of signal before close. And awareness stops looking like a cost. Do not boil the ocean. Start with one sensor in 90 days or 2 or 3. I would start with web data anonymization in some companies that already have traffic. Watch the name accounts that were on their pricing page. While you stayed invisible to them. And this is the moment to stop being theoretical. So if you want to screenshot something during this presentation, this is the slide. This is it. The 7 rules on one page. One contract. Deterministic scoring with AI at the edge. Score that decompose decay. Capture and creation. One play per quadrant. Learn offline with a holdout and a human on the irreversible moves. Get these 7 right and you have system sales uses instead of reject. And I have a quick reminder, there's a surprise on the next slide. So please take a screenshot right now if you like. I'm gonna hold that another 5 seconds. So the surprise I promised you is, um, I want you all to have access, um, to the book before everybody else have. So the only request I have is post about this session on LinkedIn, tag me and Automate DevOps, and we'll send you a free copy of the Demand Compass, whereas we explain everything about the system I just mentioned, the rules, before it launches. And I actually ask as well a honest review on that. Not a nice one, honest one. Like, if you don't, and if you have not joined the launch yet, there is also the QR there to join the Luma event. Now let me take your questions.

Yeah, I would add for that post to also tag the Go to Market Engineering School as well that invited us here. Maybe we forgot to add that.

I'm sorry, yeah, you're right.

Thanks, Markus.

Tag both of us. If not, you are not getting the book. All right, let's chat.

Okay, we've got one question. Shubha, do you want to come off mute and read your question?

Yes, absolutely. David, first of all, thanks a lot. Can you hear me clearly?

Yeah.

Ah, perfect. Yeah, so I just had like 3 aspects that I was thinking about. The first one is that you mentioned about building intent. And as a GTM engineer, I'm just curious how you'd recommend one to build intent. And the second one was about selecting good signals. And there I was curious that for, for the respective startup product or offering, what's your recommended framework for finding the right signals to focus on? And one last question would be about showing the deals as timelines. And would you have some strategies that you'd recommend to reduce the timeline, to reduce the duration of the sales cycles?

Awesome, all those questions are, are great. Let's start with the first one. So building intent is mainly something that you need to, uh, first understand what's the market. Um, if you're targeting, let's say smaller e-commerce, probably you won't find them on LinkedIn. So you will need to build intent somewhere else. For example, using ads, using cold email, using different tools that they are used to interact with. So building intent is actually pushing things for them to see and click and interact. So once they start interacting with those assets, uh, you can capture the intent. Another good example would be blog post. I mean, if you're not building, um, any traffic on your site, uh, that's something that mainly most of the companies can do, like posting blog posts, starting running ads, running cold email campaigns, running outbound campaigns for LinkedIn. Um, we also comment on people's posts on LinkedIn. Um, there are a lot of moves that can drive intent for you, but if you're not doing anything, it's difficult to capture those. For example, it's not the same, a LinkedIn post about a general topic, like, um, I can post about getting married and I can get a lot of traction for that post. It's not the same as opposed— I'm saying, okay, I have 5 seats for the next batch of clients I'm taking the next month. And if people are reacting to that post, it's a high intent for readiness. For the second— for the first pose, it's zero for readiness. It might be a good one for scoring like awareness, but not for readiness. Um, what do you think about that? Do you have any— did I respond to that? Maybe Marcos or Jared want to also add something else.

Yeah, I mean, this is just— you're just saying like, how are you going to know if people are aware of you unless you're putting something out there in the universe, right? If you don't put something out there, people have nothing to react to. You don't know if they, um, like, that is what you're going to be tracking and feeding into the score. We got that right?

Yeah, yeah, yeah, totally.

So building the inbound motion actively Right, so also like the, the personal brand and the online presence in some sense.

Yeah, it's inbound but also outbound, um, both of them. I mean, for let's say cold email, you can track replies. If you're sending them an asset, you can also track if they're visiting your site, downloading that asset. If they're, in some cases, you know that you receive a cold email and you read the email, but you didn't respond, but you check out their domain. That's another thing that you can do. Track if they're actually checking out your domain, how much time they're spending in it, and so on. For the second question, speaking about selecting good signals, It depends also. Sadly, there are no like general, uh, answers for this type of questions, but it depends on, um, your industry, who you are targeting. Um, but in general, most of us have first-party data, uh, that can be good signals. Most of them are good signals for brand awareness, like If they're reacting to your post on LinkedIn, for example, you can track those using many tools. We generally use Apify for that. Then, for example, if you're tracking website traffic, there are many tools for that. You can use Clay, you can use Warmly, you can use, um, RioDev, you can use Vector. Clearbit. There are many tools that can help you do that. If you want to track, uh, for example, um, people following your competitors, that's something possible as well. You can scrape those competitors, and there are tools that can help you do that. For example, Boomerang. Marcos, you opened your mic. You want to add something else here?

I was going to ask you if you can go one slide back for the ones that couldn't screenshot the checklist, so then they have more time. Yeah, I agree with everything you say. So commenting is an interesting growth pillar within LinkedIn. And right now we are seeing the algorithm punishing the post's reach, and we are seeing an opportunity on comments. But there's always a challenge, especially when you automate, to make it like sound like human and not like AI. And also make sure that we are adding value with every interaction and not make it like, you know, like something that bothers and cannibalizes the platform as well. So that's one system that we normally work with Extrovert as a provider, and we keep a human in the loop that sort of, you know, takes care of the the quality of the comments. But then if you systematically do that, you could track the sentiment of the replies and generate positive signals out of that daily activity, right? So that could be another application of this concept.

Yeah, um, but yeah, you The way I would put it is, first of all, try to understand your ICP, their buying journey map. Once you have it mapped, all the possible actions they can take, you have a lot of ideas on how to capture good signals for your market. Um, regarding the third question, when one thinks about showing deals as timeline Forward strategies to reduce duration of sales cycles. First of all, reaching out to, to companies that are showing you buying readiness. Like, let me go back here. If you reach out to companies that are in this quadrant, like if they show you that they have high buying readiness the sales cycle will be shorter. Why? Because they are already buying. And it's, I mean, it seems obvious, but most of the companies are targeting equal companies that are not buying and companies that are buying. So of course, the speed will be increased on this quadrant, these two quadrants. The, the other strategy would be to ramp up the nurturing phase or the awareness, uh, of your brand, of that account. If they don't know you, they don't trust you, they will compare your brand with 5 different competitors, and it will take more time to close the deal. Now, if they know you and they trust you, they will select your company and they won't evaluate so much. I think we have all been here in a stage that at a random fact you receive an inbound lead and they trust you like 100% and they buy your services without even comparing it. We all have had that experience before. Now the, the question is, how do— can we replicate the process? And it's increasing brand awareness. Another strategy might be, um, I mean, depends on if you have the possibility to change their ICP, um, to reach out to companies that have shorter sales cycles. And this is obvious as well. Maybe you don't have that possibility. But if you have it, you can start researching the market and you will find some ICPs that their sales cycle is smaller and they have the same budget. For example, Um, if you're targeting an account, uh, that is VC-backed, they will have more eager to grow faster than a company that doesn't have a VC-backed because they're growing organically. So even though the company, if you compare it to both, might have the same revenue might have the same headcount, one of them will have more possibilities to grow faster and they will take faster decisions. So yeah, those are some angles. Maybe Marcos, Jared have already ideas, but I think that covers some ideas.

Well, sometimes, especially on like deep tech companies. You have the technical buyer. I mean, especially on complex sales with long deal cycles, you have the technical buyer and the economic buyer. So there are multiple different signals for each of these types of buyers, and you have to figure out the, the best path for conversion, right? Sometimes it's worth like starting out from the decision maker, the economic buyer, but somehow you, you need to relate the signals from the technical buyer with the economic buyer. So that's another challenge that this type of companies face, and this type of systems could help by monitoring different personas within the same accounts. Yeah, would be my take.

Nice. Thanks. Thanks for going deeper, actually. I'd say I really like the last point that Marcos mentioned about relating the signals of technical buyer with the economic buyer. And then also what David mentioned about building trust and nurturing and also reaching out to companies that might be VC-backed because there's like a higher motivation to react quickly. Um, yeah, thank you. Thanks, guys.

Thanks for the questions.

Other participants have any questions? While you're thinking one, I got one I want to throw at you. You know, if you're just starting out with a company and they haven't really done any lead scoring or they haven't done it successfully, what is that process that you run through with them to figure out what to track, what goes into the lead score? Imagine it's different for every company. So Is that you shadowing their top reps? Are you doing some fancy math and some machine learning? How do you solve the kind of cold start problem of putting a lead scoring algorithm together?

Yeah, that's a great question as well. The first step we take after we start working with a new client is to audit their CRM. So, uh, by auditing the CRM, you can reverse engineer their best deals, and by reversing that, you can create a proposal for them, like a hypothesis. These are the type of companies, these are the type of leads, uh, that close more often, that have higher conversion rates, um, that close bigger deals. Shorter, long, shorter cycles to close, and so on. Um, but then that's data. Um, if you don't have it, uh, you need to start from scratch. And sometimes you have clients that they don't have any information about the market, or they're starting into a new market, um, and Either way, you don't have data to support the decisions. So you need to make a trust loop and run by them a discovery process to actually understand what they are looking for, which type of companies. And we usually make a list of, let's say, 10 different ICPs. We score the ICPs based on how possible it is for them to close, like if they already know them from previous experiences, if they have like a network of contacts on that ICP and know their pain, how to relate with them, how to present something to them, if that ICP actually have the budget to spend on your solution, if they have a lot of competitors in that segment or not, if that market is growing or not. Based on those questions, we propose like an hypothesis of, let's say, let's reach out to these 3 different ICPs. Now, once we have like the description of the ICPs, we map the whole TAM and we check with them. So we map the TAM using Clay, using Prospio, using SalesNav. There are many tools that you can do to map all the possible accounts and leads. Um, we double-check with them, so if they have data, if they have experience, we run a batch through their reps to actually score them by their own. And if we match the score I mean, this is a qualified lead or it's not qualified, we move on. But if we don't match, we collect the feedback and keep improving it. And that's a process that it's, I mean, you do it once to start with, but then you need to do it again and again and again after, let's say, every month or every, quarter to actually have a sense of what's changing inside a company because they're meeting them, they know their needs, they know if the market have changed. Um, for example, last year, uh, there was like a huge trend of companies, um, subscribing to Clay and they need Clay trainers. But this year it's A lot of companies wanted to learn how to use Cloak Code. So if you keep targeting companies that might be fit for Clay, but they are not anymore, the scoring will be off. So the scoring should be live and be updated frequently. Yeah, I think that's my whole answer.

Okay, awesome. And then, you know, you're— the model that you propose, it can get somewhat sophisticated with decays on different signals and different kind of decay rates. So where are you actually computing that all? I mean, you're not doing that in Clay, right?

Yeah, that's another great question. We build our own system. We used to run it in K8s, but we built our own system with the K8s. Yeah.

Okay.

Because we have like a, back then we had like a, yeah, there is an enrichment that captured like recent date and you can rerun that one on each column every day. So you compare it like recent date, like today with the date that you capture that signal. And you can apply a decay mode, another formula column that will decrease the value of that signal if they are separated more and more and more, you know. So, but right now we're using it in our own system. Actually, let me show you this.

Oh, sorry.

It's not. This is a system that we have been building and deploying for some clients in the last few months. And in the system, we have the map for them to know. Let's say, uh, let's go to current date. This is mockup data, by the way. Um, you can know where your accounts are. Or where your leads are on the quadrants based on the heat map. But also you can know that they have been moving. Like this week, this lead moved from cold to sales ready because they visit your pricing page, they book a call, whatever, you know. So they show high intent, engage, and based on that, you should activate those leads. There are some leads as well that started last week in, let's say, sales ready, sales ready, and then they go back to nurture because the window of their readiness closed. And they raised funding like 6 months ago, and you said about threshold that If they raise funding in less than 6 months, you score them 6. If they score them, if else, it's 0. So the scoring model, it's something that we build in our system to do it. So we have like a decay mode for each, like for awareness and readiness. You can build it on Klaviyo, but in this case we build it in our own system. Um, once you have the decay clear, the system will be updating the records every day, and the ones that are not in market anymore will land in cold email, and you need to start reaching out to them again. The same happens with nurture. If you haven't nurtured your leads properly after a while they will lose track of who you are. Um, that's something important as well. So that's how we're, we are doing it, Jared.

Okay, fascinating. Um, and the actual inputs here, okay, you know, it decays over 6 months, or right, the, the fine-tuning of that, you just start out basically on a, on a hunch Or do you recommend starting out with some kind of sophisticated multivariate regression? I mean, where do you— and how do you kind of bootstrap this?

Yeah, that's also a great question. We propose a way to work this with our clients. So for example, if you're talking about the case, awareness is step That means that it's going to decay, but slower than readiness. Readiness is exponential. And it can score 10 the same day, but then, for example, express interest in the conversation, we decay to half the score in 2 days. That's something that you can get from a conversation with their sales team, uh, because they need to agree with the system. They don't— if they don't agree with it, they won't take the leads here. So if you send them like a lead that have a positive conversation with their events in their booth like a month ago, um, they, they will prioritize those leads. But they won't trust it as much as if you send them a lead that attend an event yesterday. So this is something that you need to build it with them. And once you have it built, um, the rule we set to this system, and in— and the rule I also proposed in the book, is to have a 5 to 10% of leads sending them to the reps even if they don't have any signal, and try to reach out to them and see if they get the same results with those leads and compared with the system. If they have the same results, the scoring is off. You need to change the scoring. Maybe the readiness score that you have been building is not good. Or maybe the awareness is not good. But anyways, the system should be like, should have like initial hypothesis. And you can build and improve it over time. You can also use AI to capture like changes based on the deals that you have been closing. Because once You implement the system, you can go back and say, let me show you. For example, if that account closed a deal with you, you can go back and see the timeline and say, okay, this company have shown us like one week ago that they were hiring for an AI position, whatever. They have received LinkedIn Ads impact, but it was 2 weeks ago. We don't score that. Uh, we score that they have attended to an event. So you can track back all the actions they did, but you need to start collecting the data first. So after you start collecting the data, it depends on the volume, depends on the type of company, um, but after a while you have enough data to you know, to string the score with the closed, uh, accounts.

I noticed back in the scoring, uh, screen of the prior one, you, you're tracking a lot of, uh, signals from LinkedIn, and I imagine that is, uh, yeah, I mean, how about other— but yeah, wow, amazing. And then, so, but, and this is across everybody in your company? Um, okay, for you guys maybe it's not that, that many people, but for your clients there could be tens or, or hundreds of people in the company. How, how do you capture all the LinkedIn activity across everybody in the, in the company and feed it into this?

Those are great questions. So, um, there are some activity that you can capture that is public, like reactions to your post, posting, reactions to your comments, comments. All the things are public and you can capture it through, for example, Apify, or there are other tools that can help you do that. But then there are some data that is not as easy to capture, like for example, conversations. In order to capture conversations, you can use GetSales, you can use HeyReach. Uh, you can— you should use a tool that can capture the data for you. The same for LinkedIn connections. I mean, HeyRich or GetSales or whatever LinkedIn automation tool, uh, will update your Facebook connection time. So you should connect your team into that tool.

Okay, so every— and that's a challenge that you want to include needs a GetSales and HeyReach account or user. Okay, got you. Well, look, I think we're about at time. I, I think this has been, uh, very interesting for me. I, I, I love this framework. Simple 2x2 always wins, uh, beats out anything complicated. I think you've done a masterful job here. Um, and, um, I guess people know how to reach out to you. If they've got more questions. And I, I hope that people do post and get the book. Um, any, any final thoughts from anyone?

You got to open your mic. I'm happy to, to hear you as well. Thank you for the, for the emojis.

All right, awesome. Well, uh, thanks guys, and, uh, Thank you, Jared, for hosting us. We'll stay close.

We'll send you a book as well.

Okay, I love it. I'll put it on my bookshelf and read it. I will read it.

That's the most important part.

That's the important part. All right, thanks everyone. Take care. Bye-bye.