5 Ways to Turn Your Cold Outreach From “Meh” to “Wow!” With AI

This ad is not shown to multipass and full ticket holders
React Summit US
React Summit US 2025
November 18 - 21, 2025
New York, US & Online
The biggest React conference in the US
Learn More
In partnership with Focus Reactive
Upcoming event
React Summit US 2025
React Summit US 2025
November 18 - 21, 2025. New York, US & Online
Learn more
Bookmark
Rate this content

Cold outreach is tougher than ever. Prospects are tuning out cold messages, channels like LinkedIn no longer perform, and buyers’ attention spans are shrinking fast. Five years ago, emails under 200 words got good replies; two years ago, it was 150 words, and today it's under 80 words. People are still open to cold outreach, but it must be concise, on point, and hyper-personalized. Before, there was always a trade-off between “good personalization” or “personalization at scale.” Then AI came along and turned it all upside down. Now, you can serve up hyper-personalization to the masses—just mix a little AI with API, and you've got the perfect recipe.

This talk has been presented at Productivity Conf - Practical AI in Marketing, check out the latest edition of this Tech Conference.

FAQ

Albato is a no-code integration and automation platform that helps users automate workflows and tedious tasks through API integrations.

Albato is headquartered in Lisbon, Portugal.

Albato has over 80 employees on board.

Albato's services are available to individual users, SMEs, SaaS companies, both B2B and B2C.

Albato offers over 800 out-of-the-box API integrations.

Albato Embedded is Albato's flagship product, specifically geared towards SaaS companies, allowing them to offer a marketplace of white-labeled, ready-to-use integrations.

Albato allows SaaS companies to embed a marketplace of 800+ white-labeled integrations, enabling them for their customers in just a few clicks, thus reducing the need for developing and maintaining integrations in-house.

Albato uses AI to personalize outreach at scale, improving the effectiveness of cold email campaigns through data-driven insights and tailored messaging.

Albato's outreach strategy relies on three fundamental pillars: high-quality prospect lists, impactful data points for prequalification, and personalized killer messages.

With the integration of AI, Albato achieved a reply rate of 6% and an interest rate of 3.5% in their outreach campaigns, showing significant improvement over previous strategies.

Leo Goldfarb
Leo Goldfarb
40 min
05 Dec, 2024

Comments

Sign in or register to post your comment.
Video Summary and Transcription
Hey, everybody. I'm Leo Goldfarb, a managing partner at albato.com. We're a no code integration and automation platform with 800+ out of the box API integrations. Our flagship product, albato embedded, is geared towards SaaS companies. Instead of burdening your team with developing integrations, you can embed our white labeled integrations in your SaaS. Our outreach is targeted mainly at B2B SaaS companies. We explore leveraging AI to boost outreach performance. Cold outreach has become increasingly hard, and some techniques that used to work are no longer effective. Let's dive right in and see the positive outcomes of our efforts, thanks to AI in our outreach. Even a reply like this is always better than no reply. We believe in making our messages meaningful and valuable to grab people's attention. There's a never-ending dilemma in cold outreach: personalized outreach or outreach at scale. With the help of AI, we can personalize at scale. Every outreach relies on high-quality prospect lists, impactful data points, and killer messages. The first pillar of outreach is high quality prospect lists. Gathering impactful data points helps prequalify leads. The spearhead of outreach is a killer message. The reply rate for this campaign was around 20%. After implementing AI in outreach, we focused on personalized and on-point messaging. We addressed the problem with integrations via Zapier and emphasized the need for native integrations. We leverage chat GPT to generate personalized messages and specific use cases for sense park customers. The campaign involves extensive personalization, including basic, buyer persona, and dynamic value propositions. We used an outreach dataset filled with AI-generated data points, including company details and integration use cases. We pulled customer reviews from G2 and used AI to identify the most common and critical problems with integrations. Chat GPT is good at determining review sentiment and identifying reviews mentioning integrations. However, it struggles with aggregating data and identifying hidden patterns. Chat TPT is better without setting hard boundaries, allowing it to be creative and provide convincing results. AI is effective at identifying isolated ICP criteria but struggles with complex composite ICPs. We attempted to use LinkedIn social activity as a conversation starter for sales, but the generated messages were off and dubious. To incorporate AI into our cold messages, we feed our prospect list to open AI via API. We train and tweak the language model through iterations until we're satisfied with the outputs. The AI-generated data points are then uploaded to our outreach automation tool, enabling hyper-personalized AI messages. We also offer white label, embeddable integrations for SaaS companies to provide seamless experiences within their platforms.

1. Introduction to Albato

Short description:

Hey, everybody. I'm Leo Goldfarb, a managing partner at albato.com. We're a no code integration and automation platform with 800+ out of the box API integrations. Our flagship product, albato embedded, is geared towards SaaS companies. Instead of burdening your team with developing integrations, you can embed our white labeled integrations in your SaaS.

Hey, everybody out there. My name is Leo Goldfarb. I'm a managing partner at albato.com, an all code integration platform. And yeah, I'm so happy that you're all listening to my presentation and hopefully you're going to enjoy it.

So before I talk to the chase just a few words about albato and who we are. So we're a no code integration and automation platform with 80 plus people on board and headquartered in Lisbon, Portugal, and we have basically all kinds of customers, individual users, SMEs, SaaS companies, both B2B and B2C. Automate their workflows and their tedious tasks through our no code API integrations and the figures that you now see on the slide should give you a pretty accurate idea of the scale of our business.

So we've been around for five plus years and as of today, we've got 800 plus out of the box API integrations that you and your customers can use from day one. And there are 100 K plus active customers on the platform. And also just another highlight that albato embedded is, so we're actually have like a portfolio of different products geared toward different businesses and albato embedded specifically is our flagship product geared mainly towards SaaS companies. So in a nutshell, if you're a SaaS company, instead of burdening your devs and your team with developing and maintaining dozens or hundreds of integrations, which is always a huge pain for everyone, you can embed a marketplace of 800 plus white labeled, ready-to-use integrations in your SaaS and enable them for your customers in just a few clicks. Yeah. So that's in a nutshell about who we are, just to give you a bit of a context.

2. Leveraging AI in Outreach

Short description:

Our outreach is targeted mainly at B2B SaaS companies. We explore leveraging AI to boost outreach performance. Cold outreach has become increasingly hard, and some techniques that used to work are no longer effective. Every lead is precious, and burning your audience with a crappy outreach can have long-term consequences. Let's dive right in and see the positive outcomes of our efforts, thanks to AI in our outreach.

So as you may have already guessed, our outreach, and this is what we're going to be talking about here today. It's targeted mainly at B2B SaaS companies, which is a, that's a super rough ICP, I would say, yeah, that's a very niche audience and it's, it's pretty hard to kind of put together the prospect list and reach out to all those B2B SaaS people, let alone get some replies to your cold emails.

So let's cut to the chase and basically what I'm, you know, what we're going to explore throughout my presentation is how our experience, our practical experience of leveraging AI to boost outreach performance and I guess no one would argue if I said that outreach has been increasingly hard to tackle in the last few years. Yeah. And it's actually getting worse. So I guess, you know, everybody out there listening to me can relate, especially if you're involved into running a cold outreach, you know, to a certain degree.

So people, people tune out cold messages more and more. Yeah. And basically every time you get a, you know, come up with another secret sauce, you got to reinvent the wheel. Yeah. And some outreach techniques that used to work just a while ago are no longer effective. Yeah. And for example, you know, maybe just a couple of years back you know, I'd say, Hey, let's, let's do this. Let's just run a massive outreach campaign. Yeah. Let's use this so-called spray and pray approach. Yeah. And just collect like the low hanging fruit. Yeah. Because anyway, we'd get a bunch of replies. We'd get a bunch of meetings, a bunch of sales.

Well, you know, I got some disconcerting news for you folks. Yeah. Because there's, there is no longer the low hanging fruit in the cold outreach. Yeah. So basically every fruit is very high hanging. Yeah. Today. And you got to be really, really thoughtful and diligent in how you run your cold outreach, especially if you have like a pretty limited addressable market like we do, because you don't want to just, you know, tune out and burn your audience with a crappy outreach. Yeah. Because once they flagged you as spam or once they unsubscribed, you can never reach out to them anymore. Yeah. So basically every lead is, is precious. Yeah. It's priceless. So yeah, let's dive right in. And here I'm kind of putting a card before the horse. So this is actually, you know, the outcome of all our efforts. Yeah. So here you can see just some examples of what, what our prospects, you know, are replying to our cold emails. And of course it's always great to see our prospects reacting positively to our cold emails. And basically well, I'd say the reason why it happened. Yeah. Because we had a lot of ups and downs with our cold outreach and now we're, you know, looking kind of at some kind of accomplishment that, that, that took us a while to get to. And it's all thanks to bringing AI to our outreach. Yeah. It's just a bit of fun. This is a real reply from one of our prospects. Yeah. I'll give you just a couple of seconds to read through this stuff.

3. Engaging Prospects with Meaningful Messages

Short description:

Even a reply like this is always better than no reply. We believe in making our messages meaningful and valuable to grab people's attention.

Yeah. Okay. I'm going to help you. This is quite funny because, you know, we ended up booking a meeting with this guy. Yeah. But always be ready that sometimes they can drop an f-bomb at you. Well, this one is even more convincing at one point. Yeah. Honesty is the best policy. Yeah. So basically, you know, a reply from a prospect, even a reply like this is always better than no reply. And you know what happened actually? That guy, the guy who dropped the f-bomb, this one. Yeah. He, like a couple of days after he got back, and he was eager to learn more about our product. Yeah. And I think we ended up scheduling a meeting with him. Why? Yeah. Because we believe that, you know, we've put in a lot of effort into making our messages meaningful and, you know, just having that value that make people, you know, pay attention and they can just ignore it. Yeah.

4. The Power of Personalization with AI

Short description:

There's a never-ending dilemma in cold outreach: personalized outreach or outreach at scale. With the help of AI, we can personalize at scale. Every outreach relies on high-quality prospect lists, impactful data points, and killer messages. Let's explore examples of good and bad messages and how AI can make a difference.

Yeah. As we're getting closer to, you know, our practical experience. So I guess everybody out there involved into cold outreach, you know, to a certain degree, everybody out there should be able to relate to this one.

So there's a, like a never ending dilemma that, you know, that all of us have faced many times. So you can either do personalized outreach or outreach at scale.

For us, that was the case before AI came into play. Today with the help of AI and also some AI integrations, we can really personalize at scale. Now I'm going to walk you through our practical experience. Okay. On a high level, every outreach sits on top of three fundamental pillars or cornerstones.

The first one is high quality prospect lists. That's like the baseline. Then basically gathering and pulling impactful data points that would help you to first prequalify and kind of cherry pick the best leads. And then of course, the spearhead of any outreach is a killer message that ends up being on your prospect inbox.

Yeah, let's explore some examples of good and bad messages and how AI can actually make a difference. This one is actually a very old message and that was more than a year ago. So at that point, we were not using AI in our cold outreach. Yeah. I'll give you just a couple of seconds to read through this message. So what do you think about it? Yeah, actually, you know, we kind of, we wanted to, we, we thought that would be a killer message, but actually we ended up just sending a yet another message, like very mediocre and not really impactful.

The problem with this message, of course, actually there's a bunch of problems here. Yeah. So I guess we made every single mistake that you make when you only start doing outreach. Yeah. First of all, that's a very vague subject line. Yeah. It doesn't really convey anything.

5. The Importance of Personalizing Messages

Short description:

The first pillar of outreach is high quality prospect lists. Gathering impactful data points helps prequalify leads. The spearhead of outreach is a killer message. Let's explore examples of good and bad messages and how AI can make a difference. An old message without AI was mediocre and under personalized. The subject line was vague and lacked problem-solving. The email lacked urgency and was too generic. The reply rate for this campaign was around 20%.

So the first one is high quality prospect lists. Yeah. That's like the baseline. Then basically gathering and pulling impactful data points that would help you to first prequalify and kind of cherry pick the best leads. And then of course, yeah, like the spearhead of any outreach is, is a killer message that, you know, that, that ends up being on your prospect inbox. Right.

And yeah, as you know, let's explore some examples of good and bad messages and how AI can actually make a difference. So this one, yeah, is actually very old message and that was, I guess, more than a year ago. So at that point, you know, we were not using AI in our cold outreach. Yeah. And I'll give you just a couple of seconds to read through this message. So what do you think about it? Yeah, actually, you know, we kind of, we wanted to, we, we thought. So that would be a killer message, but actually we ended up just sending a yet another message, like very mediocre. Yeah. And not really, not really impactful.

The problem with this message, of course, actually there's a bunch of problems here. Yeah. So I guess we made like, you know, every single mistake that, you know, like that you make that you make in a, you know, when you only start doing outreach. Yeah. So first of all, that's a very vague subject line. Yeah. It doesn't really convey anything. Yeah. There's, you know, there's some basic things that are missing, like there is no problem part here. So we're talking about some, you know, general trends about AI is now driving growth in the conversational industry. I guess you're looking to expand your AI capabilities. Then we kind of, you know, move on to, and we're trying to sort of pitch some feature that our prospect could add to their product, but probably the question that, you know, that Jeffrey is thinking about when he actually opens this email, if he would ever open this email with such a mediocre subject line, right? So what's in it for me? I mean, what, what kind of, what kind of problem, what kind of urgency those guys are trying to, you know, to flag, because there's nothing about it here. Yeah. And of course I, I guess the biggest problem here is that it's, it's just way too generic. Yeah. And you see, I highlighted that in the yellow for you. Yeah. There's, there's just a handful of we can call it like, you know, very basic personalization. Of course we're using the prospect name. We're using the, the prospect's company name, the company vertical. Yeah. So we know that that company Abbott. Yeah. So they, they are in the conversational industry. They're some kind of a chat bot, but that's, that's not enough. Yeah. So I would say that this email is heavily under personalized. Yeah. And that's the problem. So now let's do some guesswork. Do some guesswork. So what do you think their reply rate was in this campaign? I'll give you a second. Yeah, it was pretty low, so why for 20%? Yeah. Which is kind of okay.

6. The Power of Personalized Messaging

Short description:

The reply rate was below the market benchmark. After implementing AI in outreach, we focused on personalized and on-point messaging. We addressed the problem with integrations via Zapier and emphasized the need for native integrations. The email included three layers of personalization: basic, buyer persona, and use case.

Which is kind of okay. So we got some replies and some of them were positive, but it's way below the, uh, you know, the market benchmark. Right. And also the interest rate was, you know, less than a half of that. So by interest rate, we're actually referring to, uh, say the ratio of, uh, you know, positive interest replies to all replies. And this is also super low, you know, against the, the, the market benchmark at all.

And then let's fast forward to six months after. And this is when we kind of, uh, you know, hacked AI in outreach and started seeing first positive impact. So, uh, as you can see in this particular campaign, yeah, as we're selling, uh, FBI integrations. Yeah. So we kind of decided to piggyback on the fact that our prospect, specifically this company, Sandspark. Yeah. Uh, so, uh, we're piggybacking on the fact that, uh, they offer integration with their platform via Zapier. Uh, which is great. Yeah. But not as great as offering native integrations within your platform, which is exactly what we're selling here. So, uh, yeah, here you can see that it's, uh, you know, it's a very different approach. So we're really, uh, yeah, we made that outreach much more, uh, you know, personalized and on point.

So first of all, there's the problem part here, so it's, it's right here. Yeah. So all for integrations via Zapier is never enough. Yeah. And, uh, your customer sales reps and agents need to switch from Sandspark. That's also, here we're talking about the problem. Yeah. And some, some risks. Yeah. That, you know, your customers, they, you know, they might churn because of, um, you know, some problems with integrations and of course it's going to negatively impact your revenue. Yeah. At the end of the day. So, and here we're talking about the solution. Yeah. But I guess the most interesting part is, uh, where is the personalization here in this email? Yeah. Can you guess I'm going to give you a tip. So here we're using, uh, three layers of personalization. So, uh, yeah, we call them basic, which is, uh, you know, something like very, very essential, like the prospect name, the company name. Uh, then there's also personalization for buyer persona. So as we're outreaching to different people, uh, for example, business persona, like a founder or head of growth or CEO or product persona, like CPO or owner, there's also like a technical persona, like CTO. So we have, I'd say roughly six or seven personas that, that we're outreaching to. Yeah. And of course, uh, it's very, those things are very different. Yeah. For persona. So for example, to the founder, we're saying that, Hey, you know, you're losing your revenue. Yeah. And if you, you know, try out our solution, that's going to boost your retention and MRR. Whereas to the product guy, we would be, you know, uh, we would be talking about some kind of product adoption, uh, yeah, like a monthly usage and stuff like that. Yeah. And the third and the most impactful layer of personalization is we call it use case. Yeah.

7. Leveraging AI for Personalized Campaigns

Short description:

We leverage chat GPT to generate personalized messages and specific use cases for sense park customers. Through AI, we extract data points and craft emails without manual efforts. The campaign involves extensive personalization, including basic, buyer persona, and dynamic value propositions. We use prompts to gather information and automate the outreach process. The positive impact includes a 6% reply rate.

And it's generated through chat GPT and brought directly to our messages. So for example, uh, here where we're talking to that company, yeah. And we're, we're mentioning their customers. So we know that sense park is a platform for sales reps and CS agents. Yeah. And that was the data point that were pulled from chat GPT. Yeah. And also in the PS, you can see that we're giving a very specific use case of how sense parks customers could use, uh, integrations. Yeah. So sales reps can create dynamic videos. So we know that, that, that company kind of offers a tool for, uh, for AI for creating AI videos or dynamic videos. And again, this is something that, uh, that we pulled through AI. So we didn't have to, you know, craft every single email by hand. Uh, yeah. So this is, this is amazing. So, and these are the prompts that, that we used. Yeah. So for example, uh, for, to get this data point. Yeah. We asked Chad GPT, Hey, who are the most common users of sense park? Yeah. Company is dynamic, uh, variable. Yeah. That we're fitting to AI. And for this part, yeah. We use this prompt. So tell us what's the most common integration scenario for sense parts customers. Yeah. Yeah. And just, you know, to, uh, give you an idea of what this campaign looks under the hood. Yeah. It's almost look, looks like, you know, a chunk of code. Yeah. Because there's so much personalization here and some, you know, like branching and some conditional logic. But, uh, you know, to make it simple yet again, we're using basic personalization, the company name and, uh, the prospect name, the buyer persona is kind of static personalization. So just depending on who we're talking to, yeah, we're switching to. Different value prop and different, uh, problem, uh, problems. Yeah. That we're flagging. And this one, like I said, is the most, uh, complicated, but at the same time, the most effective. Yeah. So this is what we're pulling from AI. So role one, role two are our prospects customers. Yeah. Because we're talking about, you know, creating some value for our prospects customers. So it's important for us to actually pinpoint hold who those people are. Um, yeah. And, uh, yeah, of course we're using an outreach automation tool. So again, we're not doing it manually. And, uh, yeah, some brain exercise for you guys. So let's take a look at the positive impact. So the rep, the reply rating for this particular campaign was 6%.

8. Leveraging AI in Outreach Campaigns

Short description:

We used an outreach dataset filled with AI-generated data points, including company details and integration use cases. The possibilities for enriching your data set are limitless with chat GPT API.

And the interest rate was 3.5%. Yeah. Which is lips and bounce, I would say. And here's an, an example of, uh, the outreach dataset. Yeah. That, that we used for this particular campaign. So as you can see here, these are like the, you know, the basic data points like company name, first name, last name. Yeah. And all these columns are AI generated. For example. Uh, and here, yeah, you can see the prompts that we used. So we pulled the company vertical, the company integration use case. Uh, the company top two customers for one road to, uh, the company top two integrations. Um, and yeah, basically the sky's the limit, so you can send any prompt to chat, GPT, API, and just enrich your data set with, with great, uh, AI data points that, that you can use in your, um, in your outreach communications. Right.

9. Hyper Personalizing Outreach with AI

Short description:

We pulled customer reviews from G2 and used AI to identify the most common and critical problems with integrations. By personalizing the messaging to the buyer persona and providing valuable insights, we can help improve product adoption and reduce churn.

And another, um, another example of how we hyper personalize our outreach. Uh, this is still an ongoing experiment, uh, just FYI, but so far it looks quite promising. So what we've done here is we pulled our prospects, customer reviews from G2 and ran them by AI to dig up some valuable insights and patterns. So yeah, just read through this email with me. Yeah. So, hey, uh, prospect name. We saw that your company's customers were unhappy about your integrations. Yeah. And here's what people are mostly worried about, and this is what we actually, some insights that were pulled from G2 with the help of AI. Yeah. So people are talking about difficulty of sending out the QuickBooks integration and something else, the Stripe integration. So we're really flagging, uh, something very specific and this is true. Yeah. So there are, uh, there are real reviews on G2 that are talking about it. Yeah. And then this is, uh, again, the buyer persona personalization. So here I can, I can say that we're talking to the product, uh, buyer persona, because you see, this is the, um, you know, the problem language that we're using. So this leads to peer adoption and churn. Yeah. And the value prop for the product persona is your product will become much secure for users. And your product users will go up with, with our native integrations. And again, the, the use case layer. Yeah. Um, it's here and here. Yeah. And you see, I kind of traced this one because this is the most valuable part. Yeah. So, and basically what we've done here is like I said, we fed dozens or even hundreds of real customer reviews. To chat GPT. Yeah. And we asked chat GPT to kind of identify some patterns to analyze that, that large dataset and to pinpoint the, let's say the two top two common. And the most critical problems with, uh, quillers integrations. And this is the output that we got. Yeah. Um, and yeah, again, that's, uh, this one we've already covered. Yeah. So, uh, we asked like the most common integration scenario here. Yeah. And he used that in PS and that's another example. So basically here we're sending the, you know, the, the links to G2 reviews, just to kind of, you know, as a proof that, Hey, these are the links where people are complaining about your integrations. Yeah. And again, um, how we actually pull those links. Yeah. And yeah, it helped us to kind of identify, identify. Yeah. But those are negative reviews that the sentiment is negative and specifically about integrations. So, uh, yeah, that's pretty cool, I guess. And this is what a, the dataset looks like for that, uh, customer review campaign. Yeah. So as you can see, uh, you can see, you know, thumbs up and thumbs down.

10. Chat GPT Strengths and Weaknesses

Short description:

Chat GPT is good at determining review sentiment and identifying reviews mentioning integrations. However, it struggles with aggregating data and identifying hidden patterns. Manual validation and improvements are necessary. Overall, it shows promising results.

So what it actually means that, and that's, that's one of our, you know, very, very important takeaways of, you know, interacting with, with AI. Uh, so chat GPT is really good at doing some certain things while it really struggles with doing some other things. So it's good at, for example, um, uh, determining the review sentiment. Yeah. Whether it's negative or positive. Yeah. So it's pretty accurate. Also, we asked it to, Hey, can you please, um, identify reviews that are mentioning integration specifically? Yeah. This also worked pretty well. Yeah. But what it's not that good at, yeah, is basically aggregating data and identifying some, you know, like pat some hidden patterns. So for example, you're fitting it with, I don't know, a few thousands of reviews and you say, Hey, can you, uh, let's say come up with, uh, two, uh, negative insights and two positive insights. Uh, so just, you know, going through that large dataset, can you kind of, you know, pinpoints and patterns? Yeah. So that didn't work out real well. So some of the data was good. Yeah. But again, we had to do, we had to sense check it manually and we had to do actually, we had to actually do a lot of, you know, um, improvements. Yeah. So we have to, um, yeah. Validate the dataset. Manually. Yeah. But anyways, this looks very promising. I can tell you. Yeah.

11. Enhancing Outreach with Chat TPT

Short description:

Chat TPT is better without setting hard boundaries, allowing it to be creative and provide convincing results. Hyper-personalization is achieved through AI-generated data points. The no-show and spam rates have significantly decreased due to the implementation of AI in outreach campaigns.

And the last example, and that's another, uh, I think cool takeaway that you might find helpful that chat TPT is better when, you know, we kind of, you don't set like hard boundaries. So for example, you're not fitting it with like a dataset and say, Hey, can you dig into this dataset and identify patterns, but instead you kind of, you know, you give it like a full freedom to kind of come up with new stuff to be creative, you say, Hey, what are the top three integration issues in the vertical space that companies, customers may encounter? Yeah. Without, without giving it any like, uh, you know, incoming data and actually, you know, it's pretty good at coming up with some, maybe it's more generic. Yeah. But it's actually looks pretty convincing. Yeah. So, Hey, and this is what we say in our cold email, Hey, I'll list the typical issues that often become a deal breaker in your space and the sales automation space, all related to integrations. And these are like top three, very typical, you know, problems. Yeah. That, that customers are concerned about blah, blah, blah. So yeah, this one works really good. Yeah. I guess I explained, uh, this takeaway. Yeah.

And just to give you an idea of, you know, what, when we're, when, when I'm really, when I'm talking about hyper-personalization, yeah, so this is what I mean. So these, this is the list of AI variables that, that we use in our average email campaign. And so the, the, the black ones are like basic variables. Let's call them static. Yeah. But the ones, uh, highlighted in orange. Yeah. So these are the AI generated data points. And we have here roughly 15 AI generated data points. And that's only for one email campaign. Yeah. So like across four or five, uh, cold emails. So I guess that's, uh, you know, we can call it like, uh, you know, heavy usage of, of AI in the cold outreach. Yeah. But really, you know, proves to be, uh, quite effective. Yeah.

And another, um, another cool insight. Yeah. That really, you know, caught our attention is that our no-show rate in 2024 this year is 2.5%, yeah. Which is pretty low against, again, compared to the market averages. Yeah. And also the spam rate. So let's say every time a prospect just labels our cold emails as spam. Yeah. Which can really damage our domain reputation. And yeah, it's a lot of other problems that, you know, that it creates, but it's, it's almost health percent. Uh, and this is average across all our outreach campaigns this year. Yeah. And just again, to give you, uh, something to, you know, to compare these numbers with just, you know, a year ago, last summer, our no-show rate was above 60%. So, uh, yeah. Lifts and bounce, I would say. And we think that the reason, yeah. Why, you know, we really saw that improvement in our outreach performance is, uh, basically, um, yeah. Injecting AI into these three corners. Those, yeah. Working with our prospect lists, uh, pulling data points.

12. Leveraging AI for Prospect Pre-Qualification

Short description:

AI is effective at identifying isolated ICP criteria but struggles with complex composite ICPs. Another AI model may be more suitable for working with large datasets.

And of course, uh, you know, injecting all that AI data, uh, into our outreach messages. So, and just, uh, a few words about what did work. Yeah. Because it's always exciting to dive into something that failed. Yeah. So because we've been doing a lot of experiments lately and of course, you know, a lot of stuff that, you know, didn't work out as expected. So, uh, we try to use AI to pre-qualify our prospects. Yeah. Because we have a pretty complex and like multi-layered ICP. So essentially we have like 10 plus criteria. Yeah. That we need to validate to say that, Hey, you know, it's a, it's a good lead. It's a relevant lead. And of course it takes a lot of time. Yeah. So we try to leverage AI to sort of, you know, do the work for us. Yeah. But what, what really worked is that, uh, basically AI, uh, is good. I mean, specifically charge equity is good, um, at identifying whether leads match, you know, like an isolated ICP criteria. So for example, if you ask it, Hey, does this company, uh, you know, match the vertical XYZ or is it B2B or B2C or is this company a SaaS? Or, uh, do you think it could, you know, potentially require integrations? Yeah. So, uh, yeah, you will likely get a pretty accurate, uh, data point, but it really struggles to match against, like I said, a complex composite ICP. So for example, if you said, Hey, charge equity, here's my ICP. Yeah. And it is comprised of like, I dunno, 20, uh, criterion. Yeah. Just tell me whether, you know, these companies, ICP or not. Yeah. So then it's kind of, it starts to, you know, hallucinate and yeah, it won't give you a good data. Yeah. So that's one thing that you should be aware of. So again, and that kind of, you know, solidifies the, uh, the other conclusion that I brought up earlier that should be, might be good at working with some, you know, um, like isolated prompts and isolated data points, whereas if you try, if you're asking it to like, you know, aggregate to work with a large data set, uh, yeah. And do some kind of, you know, Mathematical work. Yeah. It's it's struggle. So maybe another AA model would be more, uh, suitable for that, for that kind of work. Yeah.

13. Using LinkedIn Social Activity for Sales

Short description:

We attempted to use LinkedIn social activity as a conversation starter for sales, but the generated messages were off and dubious. Personalization is more important than bad personalization. An example of a poorly crafted AI-generated message on LinkedIn is provided.

And another thing that we couldn't really pull off is, uh, you know, a lot of people are now talking, especially in the outreach space that, Hey, we got to use, you know, uh, the LinkedIn social activity. So we got to somehow reconcile what people are talking on LinkedIn, like their posts or their comments. We got to use that as a conversation starter, you know, for sales.

Yeah. And yeah, we gave it a try. So basically I'll give you a couple of examples. So for example, this guy, he was, he posted something about mental health. Yeah. So this is, you know, exactly what he, uh, what he wrote on his LinkedIn. Yeah. And we asked AI to generate sort of a bridging message. So kind of, Hey, give us like an, you know, an email opener that would, you know, help us sort of transition to a sales conversation.

Yeah. This is the outcome, like, Hey, your compassion for mental health. Uh, is great. Yeah. Like you would believe in the power of connection and this is why we're want to offer you some integrations. Yeah. To me, it looks a little bit, you know, off, uh, to say the least actually it looks ridiculous. Yeah. Or this guy, uh, you know, we're talking about like, uh, some kind of, uh, cybersecurity procedures like ISO 27 or one blah, blah, blah. Yeah. And this is what, um, tragedy came up with like, Hey, tackling ISO is like preparing for a tech marathon, just as you value cybersecurity, we value seamless integrations. Yeah. Um, yeah, that's pretty, I don't know. Yeah. Dubious, uh, I would say, yeah. Dubious. Uh, but on the other hand, she actually picked you was pretty good at actually, you know, giving us some tips and saying that, Hey, this social activity might be challenging for sales interactions, so you might, you know, like, don't use it. Yeah. Or while this one is kind of batter for sales engagement. Yeah. Because it's talking, you know, about business, about some. Uh, cybersecurity and stuff. So it's definitely got some potential. Yeah. But we, you know, we will, as I said, it's an ongoing experiment, so we will definitely continue to, you know, experiment and give it another, uh, shot. Yeah. But for now we, like I said, we decided not to go down that road.

Yeah. Because we believe in that lack of personalization is better than bad personalization. And just to give you a very funny example. So that's someone posted it on LinkedIn a while ago. So it is clearly, you know, crafted with AI and here's what they, you know, the sales rep is saying like, Hey, I was doing some research and found a picture in our Facebook. Yeah. And, uh, do you ever checked some SOMO clubs? And SOMO wrestling is a bit like battling a clunky CRM. Like, come on. What kind of, you know, bridge is that? Yeah. And, uh, I'm actually wrapping up. Yeah.

14. Incorporating AI into Cold Messages

Short description:

To incorporate AI into our cold messages, we feed our prospect list to open AI via API. We train and tweak the language model through iterations until we're satisfied with the outputs. The AI-generated data points are then uploaded to our outreach automation tool, enabling hyper-personalized AI messages. We also offer white label, embeddable integrations for SaaS companies to provide seamless experiences within their platforms.

So you're probably wondering how this whole thing where it's under the hood. So how we actually incorporate AI into our cold messages. Yeah. So I'm going to explain it to you in literally one minute. Yeah. So imagine we have a prospect list, like a Google sheet with, uh, with the companies and for example, G2 reviews or some other data points. Yeah. So then what we're, what we do is we feed it to open AI via API. Yeah. And of course we, we craft, uh, a prompt. Which, you know, it takes a pretty long time because essentially we're building like a, sort of a language model and we need to train it. Yeah. And it's not like, you know, it's not a one go. So we need to basically, yeah. Train it and tweak it. And that's a little bit of like going back and forth here. Yeah. Because, uh, we're fitting it, like we're getting out like an output sample here. And then we have a person, you know, who is like, sense checking the output and then providing feedback for tweaking the prompt, yeah. And this can like repeat, um, at least, I don't know, five, seven times before. Before we're happy with, you know, with the model and the outputs and the outputs. And then once we're happy, yeah. We kind of, uh, enrich our entire data set with the AI output. Yeah. And we end up having those AI generated data points that I showed you earlier. Yeah. And then we just upload it to our outreach automation tool. Yeah. Um, yeah. And that's it. And here you go. You have dozens of email campaigns running at the same time, sending thousands of, uh, you know, hyper personalized AI messages. Yeah. I'm just wrapping up again. I wanted to talk a little bit of our product. So, uh, yeah, because again, even with outreach, we're heavily dependent on FBA integrations and you were using our own product to actually integrate our, you know, entire stack, like the outreach automation platform, open AI and some other tools that we're using. So basically, uh, if you're a SaaS company, imagine that this blue sidebar is your platform. Yeah. Here. And like I said, what we are offering a white label, embeddable integrations that you can enable to your customers within your platform. And that's, uh, you know, the, that middle area that you see on the screen. Yeah. So your customers it's kind of can have sort of a zap here on steroids within your platform, fully white labeled seamless. So they wouldn't need to worry about, you know, getting another paid account or leaving your platform. So this all lives within your product. Yeah. And you can give your customers like a full-fledged white label marketplace with 800 plus, uh, one flick integrations. Yeah. So that's pretty much it. I hope you enjoyed it. Uh, and, uh, yeah, I hope that, uh, you know, some of the insights and some of the, uh, the, some of the experience that I shared with you today, uh, you know, will be helpful and will help you make the most of your, uh, cold outreach. And if you have any questions, uh, yeah. Please reach out to me at Leo at alberta.com and yeah, I will be happy to, to get in touch. Take care. Bye.