The B2B Outbound System Behind 400+ Meetings/Month

Frederik Jakobsen: So in this video, I'm going to show you how we book four hundred plus meetings per month for our B2B clients. So essentially, I'll run you through everything that we've built and our entire process and our entire systems that actually, enables us to produce results for our clients. So this is a brief overview. So we have something called Campaign Factory, Mailbox Hubs, Domain Inventory Gia AI, Analytics, and Monitoring. So go through each one of them. So essentially the campaign, like factory looks at each or like essentially We've built out each of these steps. So the first step is actually defining the ICP and scoring each lead and each contact that we find with AI to actually see, is it a fit or is it not a fit. Then it's the list building parts where we use multiple providers for everything. So we have multiple different providers for the actually finding the contacts and multiple different systems, some of which are, like, custom-built as well where we can go out and scrape Google or scrape Google Maps, or scrape an industry directory or whatever it might be. And typically these databases are also the, the ones that performs the best because they're the hardest to find. The ones that are just built based on LinkedIn, so Apollo or ZoomInfo or whatever, they typically perform the worst because those people are the easiest to find, and thereby those are the ones that get the most amount of emails as well, or amount of outreach. So we essentially build it as a waterfall so that we find as many contacts and then as many emails for those contacts as possible, typically including like four to six different providers for each step. So first finding the contacts, then finding their emails, and then of course, we verify all of the emails. Right now we use MillionVerify to verify all of the emails first, and then all of the ones that come back as catch-all, which is essentially just a specific type of emails that are set up where it's hard to know whether they're actually gonna bounce when you reach out to them or not. So all of those catch-all emails, we then run through Bounceban, which is a catch-all verification tool where it can actually verify all of these catch-alls because most of them you can actually reach out to as well, and typically they get quite good reply rates as well because most people don't reach out to them. Then it's writing the actual scripts, so typically based on the list, so knowing exactly like who is the ICP, who is on this list, and using that to actually write the scripts and then writing all the AB variants and spintaxing that at scale. Which spintaxing means writing different variants of the same words or the same sentences. So no email that we send out looks exactly the same or like is exactly the same. And then it's the launch step of actually assigning the right domains and assigning the right mailboxes, the ones that are performing right now having the correct sending limits and controlling that across different campaigns and the different clients and all of these things. So the first step, we have plus 16 different providers both with contacts and emails at this point where we will then, choose the ones that makes the most sense because it doesn't make sense to scrape a directory or scrape Google Maps every time. That de-depends on business type. And then we use the ones to actually get like the, the highest coverage possible and And then it's the actual, like, script writing part, which, many people get wrong as, okay, then we just write, a template, and then that is the same one that's sent out f- to everyone. That's both problematic f- in terms of deliverability, 'cause then you're very easy to flag as, mass sender, and then you go to spam. But it's also... Typically, it's not, it doesn't actually feel relevant to the person you're reaching out to, depending on how good a job you've done with actually segmenting the list. 'Cause if you have a list of five hundred people and you know exactly who are those people and they have all the same demographics and all the same pain points and all of this, then you can write it more as like a template as well. But we've essentially built in, like a specific, like an entire directory of all of our templates and everything that is working best, broken down by the different industries that we've worked in over the years for the different clients. So anyth-anything from commercial solar to PE, to M&A, to financial services, et cetera, and different SaaS products. Because typically, it's broken down by industry, and then it's broken down by pain points. And then we can, go back to those templates and then use that, and we can use AI in a clever way to then reference those templates, and then we can use, those templates and what has performed in the past, and then create new versions to essentially try to beat what has performed in the past. That's essentially how we typically do A/B testing. Then it's the spin taxing, which I explained before. And then it's, of course, like the actually killing all the variants or turning off the variants that are not performing, so the, the ones not just that are not generating leads or not generat-- or generating fewer leads or fewer positive replies, but actually, drawing it back to meetings and drawing it back to like closed deals. 'Cause at the end of the day, it's relatively easy to set up a campaign where you can generate, send, like, where you don't have to send that many emails to generate a positive reply or a lead. But typically then it also converts a lot worse into actual closed deals, and that's that's the business. That's what we're trying to optimize for, to actually close as much revenue as possible. So this is an example of a campaign. This is a campaign for that, where I'm actually currently building for ourselves where we're reaching out to fractional CMOs or like heads of GTM, et cetera. Like all fractional titles, because typically they have, multiple clients, and we actually have a few of those clients already where we have really good partnerships with them. And then, they essentially, like we essentially work with some of their clients. So then it's like a referral-based approach that we reach out with. So in here, we're then able to create like the initial lead searches and what I actually wanted to show was the script writing part. So in here, we're able to actually write the scripts. And we have, this generate scripts with AI, where it actually takes us through like a flow, but asks more like in-depth questions about, okay, who is it that we're actually trying to reach, and which types of templates do you think will work well, and all of this, so that we get all of the knowledge of the person and all of the judgment of the, CSM or me that actually sits and creates this campaign. But at the same time, get the, like all of the things that AI is good at, which is just like in general writing and things like that, and the context and of course, doing it at scale as well. So here I can then choose all of these settings and choose, okay, what awareness stage are we trying to solve? Are they typically like problem aware or product aware or solution aware? And then we create like we essentially create the variants based on that. So that like then creates variants, and then we create follow-ups as well in here. And are able to do all of that in here. So then onto mailbox operations. So this is the part that is probably most difficult once you try to scale like Colima or, yeah, Colima in general because this is where you start to actually land in like land in spam, and you need to keep track of all of these email accounts and all of these domains that you have to set up so that you can send, only ten emails per email account instead of sending a thousand emails per email account because then it's very easy to flag that you're a mass sender, and then you'll land in spam. So actually then, we have more than ten thousand email accounts across our clients. So actually tracking each and every single one of those email accounts and their reply rates and their bounce rates and their like how many emails we have to send to generate a positive reply or to generate a meeting. Tracking all of that at scale and being able to then make decisions based on how that looks like you need to actually build out a system and that's what we've done over the years to actually be able to do that and make decisions on that at scale. And this is essentially like the workflow. So the first thing is we have set it up so that we can place orders directly through our ops app, which I also showed you just before, where we have-- like we can connect different email account providers like ZapMail or Maildoso or HyperType directly in there, and then we use their APIs to then have them create all of the email accounts because that part, if you do that manually, which we used to do, will take you a week to set up email accounts, like enough email accounts for just one campaign or for just one client. So like these providers are by far the best approach to do this right now. But you shouldn't rely on just one provider because it varies and some months one provider performs better than another provider. So you need to have it diversified at least if you're running at a, a decent scale. So the first thing is placing the order and actually creating the email accounts. Then they go into warm up for two to three weeks, where, we're essentially just sending emails back and forth, Between like email accounts that we can control, where it's then taking them out of spam if they land in spam, marking them as important and then replying back. And there's tools like Smart Lead or Warm Me or Instantly that have this built in where you can use that. Then we start sending and then comes the, then we might have, like we may send with them for a month or two months or 10 months sometimes. But then essentially the tracking part comes in where we look at each domain at scale and the performance of it, so the reply rate. So if the reply rate dips below 1%, which is okay, then you're landing in spam because that is including out of offices and not interested and everything else. And, or if the bounce rate goes crazy, that is that can be an indicator as well. Then the system essentially automatically flags that, and then they are put into what we call recovery, which is essentially where we send out more of these warm-up emails. And then we either half, so that's like what we call soft recovery. So then we half how many emails we send on the email account, like the campaign emails, so that it can like warm up and get the overall reputation back. Or we have hard, where we just drop it all the way to only one campaign email per day. And then like even more warm-up emails. And sometimes as well like they just, they're just completely burned. So if you haven't noticed this soon enough, or whatever it might be, you've sent out to a bad list where you had a lot of bounces, then then the decision is to just cancel those email accounts and buy new ones buy new domains as well. But yeah, then they're essentially in recovery for one or two weeks, and then they go back into sending, and that's essentially then a loop that we can keep running. So I'll actually just show you what that looks like in the Ops app now as well. Right. This is then the, the mailbox inventory where we're able to, se- select email accounts and then put them into, soft or hardcore recovery or revert a recovery, or we're able to change all the sending settings. And then all of this is connected into Smartlead and Instantly that we use for actual email sending and the warming up of the email accounts so that we can control everything from in here in bulk The And then the other overview is like the domains overview. So here it just aggregates by domain, so it only shows all the domains in here. Again, I've just filtered to only show the canceled ones. But in here, we can essentially see, like it breaks it down by... And this is like statistical that we've found that, like being able to judge this. So this judges it based on how many emails they've sent over the period that we're looking. So for example, the last 30 days, and then how many positive replies and how many replies, so the overall reply rate. And then it scores them, scores each email account. It's like the statistical thing. It's some guy called Wilson something, Wilson CI where we can then use that to actually determine, is this email account likely completely burned? Is it likely, like just needs to warm up or go back into recovery for a few weeks? Or is it actually performing really well, and then we should, like we could actually increase how many emails we sent for that email account. So that gives us the overview of that in here as well. So yeah, that's the domains inventory we've just went through. The next thing, so what we've also built out is what we call Gia AI. Gia AI is actually a, a person on our team who handles the inbox for all of our clients. So all of the replies that are coming in, she's essentially, replying back to and making sure that we get as many of them booked in as possible, or that we hand them off to our clients. And we essentially, like a few months back, built an AI, so that's why we call it Gia AI, that can do most of this for us so that we essentially have this set up on every single client by now, and it's able to follow up with all of these leads. So if they, reply back in with interest, but then don't book for whatever reason, it actually follows up with them like three, like on day three, day five, day seven, day 10, day 14, so that we, book as many as possible. It also has the context of all, and it's like prompted and set up based on all the context and all of the data that we have from running outbound at scale and booking meetings for the last three years. And the, the combination that's really key here is still having someone that can actually look it over and actually make sure that, we tweaked the AI and like the, the way that we do it is the first two weeks where we set it up for a new client. Gia essentially looks through every single reply and or every single draft. So the AI only drafts the reply, it doesn't actually send them. And then once we're at a point where the drafts are good enough then we turn it to full auto, and then it actually sends out all of the emails automatically. Which that is the, place where then it becomes beautiful because then it sends out these responses within five minutes of actually receiving the replies. So that means, the leads that reply back with interest, they actually get a response within the first five minutes as well, which we've seen increase book- booking rates by 30 to 40%, depending on the client and the industry. But that is pretty massive. And then it books it directly into the, the client's calendar as well. So here's what this looks like in the actual ops app in the Gia AI, where we actually-- like, where it actually drafts the responses and CCs in the correct people, and we also push it to an internal CRM. So we create a CRM and a client portal for every client, so they have access to these leads. And you have all of the information, and you can see the history and everything else. So this is where Gia actually sits and goes through all of this and makes sure that they are actually booked in and it follows up and everything else. So the next thing that we've found over the years to be extremely important is the actually having the insights as an, outsourced agency or whoever actually sits with the, the outreach or the outbound typically is not the person that also closes the deal. So actually having the insight and actually being able to see and try trace all the way back to the outreach effort which deals closed that is extremely key because it's very easy to optimize based on getting a lot of positive replies or getting a lot of interest. But that typ- like, those campaigns typically aren't the campaigns that then actually also turn into, closed deals and good clients. So the way that we've done this is, as I mentioned before, we have a CRM that's then connected to our client CRMs as well. So we can trace back every booking and every reply back to which ones, like, how do they end up? Do they end up in a, you know, not like lost pile, or do they end up in a not qualified pile, or do they actually close a deal so that we can, trace back, okay, this was the email and this was the specific lead, and then use that data to actually optimize our outreach. So not only optimizing the initial outreach, but also optimizing the like the way that we reply to them and rep- like the way that we actually book them in So essentially being able to know these 12 meetings closed $340,000 in revenue and not, okay, we just send off 50 leads and then we close our eyes and then we don't know what's going on the other side Yeah. And then there's, because we've actually, built out this internal system and because we have this platform and we've connected everything, so we track every single, reply and every single email that we send, and every single domain and email account, we're then able to also connect tools like Claude Code or Codex into these databases and actually draw back analytics and have AI look at these analytics or this analytic look at these reports at scale. The other day I ran a query because we have domains, some domains are .com, some domains are .info because .info are typically cheaper, some domains are .co. So I had AI actually run the analysis on which domains performed the best and were there any difference. Turned out there wasn't really a difference. I think the ones that performed the worst were the .info, but that wasn't even for all clients, so that depended on the client as well. So being able to draw an analytics like that gives you so many possibilities to improve on things that you were never able to improve before because you just simply didn't have the data to actually back it up. So it also comes down to, which variants wins like w- which specific split tests, which ways of building the lists, which sizes of companies, like all of these data points actually being able to optimize on them, so like actually on a daily basis going through the underperforming variants and using AI in a clever way to also do this. But, going through the underperforming variants and this depends on your volume as well because you also need to have sent some volume on each variant to actually be able to determine whether it performs or whether it doesn't perform. But actually, turning off the variants that are not working, creating new iterations based on the learnings and based on the variants that are performing. So you're constantly trying to beat the variants that are currently performing the best, rotating mailboxes, replacing domains and mailboxes that are not working anymore. Adjusting the ICP, so the ideal customer persona and the targeting like the filtering and the criteria based on what deals are closing and the positive replies that we get and the feedback that we get from our clients. Yeah. So this is essentially what outbound looks like at scale and what enables us to book more than 400 meetings and per month for our clients and having, booked more than 10,000 meetings for our clients over the past three years. So I hope this video was valuable and gave some insight into how we run outbound at scale, and hopefully you can draw some inspiration from that