Transcription
Last year, in 2024, we sent between 1.5 million and 2 million cold emails on behalf of our 40 to 50 customers at any given time. We're going to go over 25 lessons that we learned in the past year and hopefully, we can make them super fast so they can be actionable for all of you. Let's go.
Keep your inbox volume low on each inbox and per domain. Right now, we're sending about 30 emails per inbox. For those of you who are unfamiliar with the email infrastructure of cold email, you can't actually use a platform like HubSpot and send out a bunch of emails; they're going to ban you. So, what we do is we use multiple domains, Google, Outlook inboxes, whatever might be, and we keep the volume low on each domain. So, as far as Google's concerned, we've only sent 30 emails, but we scale horizontally. You can use instantly.a to do this smartly. They both have done-for-you email setups. This is just the common practice that we're using right now to stay underneath the spam filters for our inbox setups.
Target 40 to 60% open rates on your cold email campaigns. Now, for those of you who might not know, ever since the Apple iOS update, any person who has their email account tied to the Mail app on their Apple iPhone, the open flag triggers automatically. So, can we actually trust open rates if you're tracking them? No, but they can be a good signal if you're landing in the inbox or not, because emails in the spam folder will not trigger the open tracker. And so, if you are tracking open rates, we don't, and we can talk about that in another section. You want to track between 40 and 60%. Anything really below 30% means you're having an email deliverability issue. If you're not using custom domain tracking, you really need to turn that on for your domains. And if you are already using that, I would set up more domains and inboxes so that you can start fresh, warm them up for 3 weeks, and get your campaign started again.
When it comes to troubleshooting domains for deliverability, don't spend a lot of time seed testing and doing deliverability testing. Just look at your overall reply rate or the open rate if you're tracking it. And if it's below 1% of an overall reply rate, not counting out-of-office emails, don't waste your time doing spam tests or GlockApps tests or any of those other things. The end result is always going to be you're going to have to set up new domains, rewrite your copywriting, and then start the new campaign all over again so that you cannot get marked as spam.
Stop sending email campaigns with more than three emails in a sequence. With all of our data across all of our customers, the best-performing campaign of our email campaigns was always email number one. Then we got diminishing returns from email number two and email number three. The only time I've ever seen a caveat to this, shout out to Patrick from Uptic, is when he intentionally was doing something funny between email 1 and email 2. Across all of our campaigns, email 1 is the most important. If you keep sending email four, five, six, seven, eight, you're most likely just going to be annoying the person and you're going to get marked as spam. Send three emails and when you go through that list, then just reuse the list. You were about to email them nine times anyway; it's just all the same thing. You can refresh your data in Clay, get better things to say, and take new learnings into the new campaigns. Keep those campaigns at three emails max.
You should be reusing your TAM list every quarter, if not even longer. Let's say you have an email campaign and you have 50,000 people in your target addressable market. We're sending a three-email sequence to all of them. When you first started that sequence, what are the odds that when you first emailed this person two weeks later, that enough has changed in their business that they're actually going to be feeling different about your offer? By putting time in between when you first sequence somebody and then when you sequence them again, more time has changed in their business where now priorities have changed, the environment has changed, something about their business has changed that now they might be more interested in your offer. This is always pivotable. When we do math and you don't have an infinite target addressable market, people come back to us and they say, "How many emails should we be sending per day?" First, try to get that ratio of the amount of leads that you get and as many positive responses as you get as low as possible. Then, once you've done 10 tests or so and you've gotten it as low as you possibly can, do your math and figure out how many emails do I have to send so that I would have to reuse my list every 3 months, giving enough time in the person's business to be able to change or in their role so that when you sequence them, you're actually sequencing a person with different priorities. Trust me, they're not going to remember your cold email. None of you remember a cold email you received 20 minutes ago. It's going to be fine; you just need the timing of the business to change.
Don't send an email until you get these things right first. In 2024, 2025, tools like Clay give us an infinite amount of filters that we can create criteria for list building. What I see too much of is a spray-and-pray tactic where you say, "Okay, I sell to the banking industry, and it's anybody with 20 to 500 employees," and then I sell to anybody with a Director of Marketing title and above. The problem with this is the person who is the Director of Marketing at a 20-person bank is a completely different person than a Director of Marketing at a 500-person bank. And so, what you want to do is use your targeting and look at your current customers and the people you currently work with and think about what are the things that they find interesting about your product or service and where are those breakpoints? They'll happen at different places, but often times it's not going to fit into a direct 1 to 10 employees, 25 to 50 employees, whatever that might be. Take a look at where you think those breakpoints are and do your best to use filters like the employee headcount filter, the geography filter, the title filter in Clay. You can even dig into what people actually say is their focus in their job because of their job experience. Use that data in your outreach as well and mine it for keywords. Be much more succinct and careful with what you're doing with your filtering so that you can actually make your messaging resonate, and you don't even have to use AI, but we'll get to AI later.
Leverage Clay to build a golden ICP by waterfalling data points and triggers. First off, a golden ICP. Let's take an accounting company for example. Accounting companies can service anybody, but wouldn't it make sense that it would be better for them to reach out to a new company because a new company might not already have an accountant or they're growing into the size that they're going to need a dedicated accountant? So that's better. Then we take it a step further and we say, "Okay, we want new companies that raised money because the investors are going to need the books to be extremely clean and they're going to need you to have some kind of real bookkeeping and process in there." Okay, then if we add another one on top of this and we say, "The company is new, they've raised money, and the CEO has never been a CEO before." We have a very different message that we could send across all of these things. Clay is the only tool in the marketplace that allows you to run all of these enrichments and change your messaging when all three of them are true, when two of them are true, when only one of them is true, or when none of them are true, and you still want to be reaching out to those people.
When we waterfall this, what we could do is we could set up your highest priority signal first. The biggest thing is we want to know that the company is new. The fact that they're funded and the fact that the CEO's never been a CEO before is just gravy on top. I'm not going to reach out to anybody unless the company was founded in the last 2 years. So, when we're setting up our Clay table, we'll run that first so we don't waste integrations on everything else. Then we can run integrations and waterfall, and only if they are recently founded, we can check if they're recently funded, and then we can run that integration. In this case, I think it doesn't really matter that they raised money to see if the CEO has never been a CEO before, so I would run that without an exclusion. But you could only run an enrichment to see if they've never been a CEO before. But in this way, all along this tier, you could change your messaging. "Hey, congrats! I noticed you recently founded the company. How's it been going?" This is terrible copy, but you know, "Do you need accounting services?" And then you change it. "Oh, you're funded? You must have investors. Now that the company's getting off the ground, they want to see better accounting analytics." Two, "Wow, congrats! I see in your lengthened experience, you've never publicly said you've been a CEO before. Must also be a wild ride because you raised money with the company just getting off the ground in the last 2 years. Do you need accounting?"
Most email experiments are worthless. The amount of times that I see people split-testing 11 different cold email campaigns, and then sometimes I see these posts, and I pay these guys because I want to know, "What are you doing that I might not be doing?" And then we go under the hood, and what they're really testing is, "Would you like to have a chat next week?" versus "Would you happen to be interested in this?" That makes no difference in your campaign objectively. If they're interested in your offer, it doesn't matter what you're saying on the CTA there, and that's not a real test. When you're doing testing, you want to be actually testing a hypothesis about your market that's repeated. Does it matter that you're reaching out to a Director of Finance and speaking about this feature or another feature? Does it matter that you're reaching out to a CFO versus a VP of Finance instead? Does it matter that the company is hiring people? Does it matter that you talk about this feature with this case study, or do people not even care about the case study?
Anytime we're thinking about campaign experimentation, we're always thinking about the first principles of our copywriting strategy, which mainly come down to the five offers in the world: How do you help people save time, make money, or save money? And then the other two is how do you raise their status and how do you help them live longer? Then, if you create copywriting around those things, split-test those against different list topics as well, too. And then off those big three offers, you should also be testing the "so what." We save time so you can do this other thing. We help you make money so you can do this. We help you save money so you can do this. And so, simply put, just again, when you're running your campaign experimentation, put an actual reason behind the thesis, not just changing around the words. Change up the offer and how you're messaging the offer. Change the "so what," and then also change how you're communicating your value proposition, your pain points, your case study, and those kinds of things.
Here's how your outbound campaign should change if you have a small TAM. If you're a company with a TAM of say, less than 20,000 people that you can easily find on LinkedIn, you're going to have to approach your campaigns much differently than if you have a very large TAM. For these kinds of campaigns, we highly recommend omnichannel messaging, and not just omnichannel that software companies talk about where they're just doing cold email and LinkedIn; that's just bi-channel. Omnichannel, we're talking about cold email, cold calling, LinkedIn, and direct mail. What we have found when we did some work with a private equity company in the past, one of the portfolio companies had a small TAM, and they said, "Okay, we want to set up an experiment. Our TAM is small. We want to see, should we hit these people in all of the channels and what cadence should we hit them in?" So, we first tried cold emailing people the first day, and then we tried cold calling them three days later, and then we sent them a direct mail a week later, and then we emailed them again, and then we LinkedIn messaged them, and all the things kind of threaded together. And then in another experiment, we just cold-called everybody all in one go, and then we cold-emailed everybody in one go, and then everyone that was left, we did a LinkedIn message, and then everyone that was left after that that still worth reaching out to, we did a direct mail campaign. You got a higher response rate when you did the threading and you did the email and then you cold-called them the next day and all these things, but the response rate was marginal in comparison to the omnichannel messaging of just doing cold calling all at the same time, cold emailing all at the same time, and then going down the line.
Now, the reason I think this happens is because I don't really think that people care that they got a call from you one day and then they got an email from you the next day and then they got a LinkedIn message. If anything, I think that might annoy more people. There's just people who respond to cold calls, there's people who respond to email, there's people who respond to direct messaging, and there's people who respond to direct mail, and those are all different kinds of people. When you have a small TAM, you need to make sure that you're hitting them on all of these different channels because you need to make as much use of your TAM as you possibly can. The amount of horsepower SAS software that you have to pay for, just the thought process to perfectly sequence people, email, LinkedIn, cold call, whatever that might be, in my past, I've never seen that actually be worth the effort because of all the stress that's going to cause you to try to get it perfectly right. When you're on the channel marketing these people, do one channel, exhaust that channel, hit everybody up again on the next channel, hit everybody up again on the next channel, and then use direct mail last because it's the most expensive. But if you have a small TAM, that's a good way to use outbound marketing.
Recently joined the company. Our most successful trigger of 2023. This is, of course, a classic, and it was a great one in 2023 to be able to email people as soon as they join a company and say, "Hey, congrats on the new role. As you're taking over this role, I'm sure you're going to try to shake up the company. You know, we can help you do that. Would you want to talk?" In 2023 and even into 2024, that's been a really reliable signal for us. I'll never forget one of my customers in 2023 received an email after they shortly joined the company, and they said, "Oh my gosh, they definitely manually wrote this one. This was amazing. It's so timely. You know, can we do this as well too?" I said, "Oh, of course we can do this. This is super easy. Clay makes this extremely easy." I think it's a signal everyone should try. In my agency, this is one of the triggers that we actually have all of our customers try. So, even though I'm saying it's the best signal of 2023, it's a little bit used. I don't know what more of a shelf life it has, but it still gets great responses in 2024, and you should be using it for your trigger campaigns.
Social signals. The greatest trigger of all of '24. In all the campaigns that we ran in the past year, social signals were the most successful for us. This included engaging with people that recently posted on LinkedIn, people who engaged with other people's LinkedIn content, people that posted about certain topics. It was all around just an amazing signal, and it was one of the best that we put for all of our customers. I'll tell you a story to really highlight this. We were working with a company that was doing offshore staffing, and we ran six different triggers. Some of those triggers included: the company was founded in the last 2 years, and maybe they're going to need more people and they're growing quickly. We also included that the person who joined the company that we're reaching out to was a new hire, our 2023 best signal. We also included: does this company already hire international talent? And we also included a signal for: does this company not include the hiring international talent? And that way, we had messaging for all of them. But one of our signals was also: if we detected that they posted on LinkedIn. All we would do, we would say, "Hey, I saw your LinkedIn post about X," and we would summarize what their LinkedIn post was. And if you're going to copy this, play, make sure you also filter out anybody who's posting political things. Just make sure you don't put that in your email. And we got a better response rate when we reached out with the LinkedIn post, and we were literally saying, "Hey, I saw your LinkedIn post about blah blah blah. You should recruit one of our virtual assistants." That's all we did. And I know the gurus are going to hate this, but I have the numbers, and this worked better than all the relevant signals that you think would work. We also were using campaigns where we were targeting people who were engaging with content and posting about certain keywords, and those also were extremely high performers. So, as we go into 2025, we're extremely bullish to continue with social signals of LinkedIn posts and content engagement in all of our outbound campaigns.
How to structure your email. I've gotten far too many emails where somebody sends me seven emails in a row, and all of them are follow-ups via just a reply to. And you know what that does? That just reminds me of the six other times that I completely ignored your message. The way that we sequence our messages at Growth Engine X is we always use about a 3 to 5-day delay in between emails. Standardly, I don't think this matters at all. You shouldn't follow up with people two days in between, and I don't really have data that five is better, but one day is definitely bad. We also send an initial email, and then we send a follow-up email, and then we send a third email with a completely different subject line. We know that our best response rates come from our first email in a sequence, so we don't want to keep going and keep reminding them, "Hey, we're the people you didn't respond to." We want them to completely forget about that other sequence. And if we talked about saving money in the first sequence, we want to talk about how we can help them make more money in another sequence and completely shake up our offers.
Successful email one frameworks. If you've been following any cold email guru on LinkedIn, I would say the most tried-and-true method that everyone almost agrees on is to use email 1 to initially introduce your offer. First line: Why you, why now? Explain why you're reaching out to them right now. Second line: Clearly explain your offer in as short as you possibly can. And your third line: Show social proof that the person you're reaching out to isn't the first person you've ever sold to in your life. And then on line four: Call to action. Some people are advocating for soft call to action, "Could I send you more information about this?" "Is this interesting?" We go for, "Would you want to chat next Thursday?" Just because we're trying to generate leads here and we actually want to get that feedback. This is how we structure a lot of our email ones when we first start campaigns.
Add more context in email 2. Mark Twain famously said, "I would have written you a shorter letter, but I didn't have the time." I think about this all the time when we're doing cold email copywriting. The way we usually use email 2 is we think about all the things that we had to delete from email 1 that were great context, and we move that to email 2. The reason we do this is because I know email 2 is going to be threaded to email 1, so we can use this to dive deeper, to keep our email still short, but add that more context that if they're interested, they can scroll down and they can still see email one, and the context is all there. When we send email 3, we're going to have to reset, but in email 2, use that to add the things you had to delete from email 1 that you really wanted to include.
Make email three about lowering the friction for somebody to respond. All right, let's think about the psychology here. You sent somebody email one, they didn't care. You added more context about how great your offer is, and you added case studies, and still they didn't care. At this point, they don't care, they're not responding. Already, what we want to do in email 3 is we want to lower the friction for them to respond. There's implicit objections when you send somebody an email that you need to get over, and we need to lower that friction. If you say, "Hey, do you want to book a call about this?" they might not respond because they want to protect their time, but they might be slightly interested. And so, we lower the friction by maybe sending some kind of lead magnet, sending something else for free, custom Loom audit, whatever it might be. Lower that friction. You've already hit them twice, they're not responding. Come up with a different call to action to just get a hand-raiser instead of a booked meeting.
Your breakup emails are ruining your email sequences. When we think of a breakup email, the amount of times I received a breakup email, "Oh my gosh, you haven't responded to me. You must be getting chased by an alligator. My stuff is so important. There's only a reason why that you wouldn't respond to it." I think a great frame of mind is Orin C. If talks about how you need to have status alignment when you're selling something. You are the expert, you can help solve a problem. You don't need to be begging people for a response. So, when we do a breakup email, we are usually using it as a way to see if we're talking to the right person in the company. We're not saying, "Oh, you haven't responded. I'm never going to talk to you ever again. Bye." We're not begging them; we're just saying, "Hey, you might have been the wrong person. Is there somebody else in the company that we should be reaching out to?"
Bonus: Clay is the only tool in the world that you can use to name other people in the department automatically, and that's currently a pretty underutilized strategy. So you can actually name other people in the department in a way that they're not used to seeing in an automated cold email, and we get tons of responses on those kinds of things.
Change the value propositions in your cold email sequence. Sometimes when I'm reviewing people's cold email sequences, they keep talking about the same value propositions. They don't switch to any other different pain points. What you want to do is if they're not responding to the first value propositions, switch it up. They might not care about making more money; they might only care about saving time because that's where they're focused on in their business. So, if you're first talking about saving money, in your next email, talk about how you can help them make more money or save time. Switch things up because what you think they might care about isn't exactly the same thing, and you should meet them where they're at. So, what that's a question I often think to myself when a campaign isn't working. We've thought about it. We're helping people save time. We're helping them save money. Nothing's working. We've had campaigns where we literally say, "We can help you save 3 hours a month on your sales tax. Are you interested?" And that works for some offers. That's as deep as you have to go. If it's not working initially, we want to take it a step further. Why do you care about losing weight? Well, I care about losing weight so I can spend more time with my kids and I could live a healthier, longer life. Why do you care about your CRM having full data inside of it? Well, I'm paying these reps to sell, not to just enter into spreadsheets all day. And so, you could say, "Hey, Clay helps people keep all of their data up to date. Tons of platforms are saying that Clay helps you keep all of your data up to date so your SDRs can spend more time actually selling." Far better value proposition. So, keep those value propositions in mind. If that's not initially working, break it down and talk about the "so what."
Don't use all your best content in email one. This is a mistake I used to make, especially when I first started my cold email career. We would have all of this data and we would pack email one with all these automations. "I got your website traffic, I got, you know, the new employee you just hired, I got the keyword from your open job post." Stuff that all into email one. Email two and three, nothing else was left. And so then we had a great email one, and then just generic follow-ups. Sometimes we'll hold some of that data back so that one, it looks like we re-researched the company, and also we can change the messaging to be relevant in a different way.
How to use AI effectively. I know AI is this big buzzword. Everybody thinks it's this new thing and it's going to revolutionize everything, which it will. But the more things change, the more things stay the same. Often times, I see people using AI just to use AI. The way we use it, think about what would you do if you were to manually reach out to this person, and how can you work backward and use AI to say the same thing? When using AI, show your work. What I mean by this is sometimes when we send cold emails, we're scraping data from multiple sources. The more sources you're pulling into an email, the more you're actually putting yourself out there to be wrong and for some of the data to be incorrect. What we used to do is we used to say, "Hey, from the numbers I see, I'm assuming you get 50,000 website visitors per month." We don't send that anymore because of the amount of people who came back to me. One person even said, he said, "Is the point of this email to be so wrong about our web traffic that I respond?" And I just never forgot about that. And so, what we do is we show our work now. And so, we will say, "Hey, I was doing a research on Similarweb, and they said you have 50,000 site visitors." So now it's not us; we were doing our homework, we were looking into it, and Similarweb is the person who's wrong, and we could also say, "Hey, Crunchbase told me that this was true about your company," or "G2 said this." When you're using AI and you're scraping data from sources, when you can show your work, so that the more that you put out there, the prospect sees that you did the research. But if it's wrong, they don't blame you; they blame the data company.
I was wrong about this AI strategy in 2024. I had a great idea. I thought we were going to add in PS lines where we would create an analogy between our company and what we did, and then the company that we were reaching out to, to give clarity on our offer and add some relevance. What I found is that when we get personalization right, people respond and they say, "Oh my gosh, thank you. You're the first person to send me an actually personalized email. Really appreciate it. Yes, let's book a call." That's usually the kind of response we get. I don't think I've ever gotten somebody to positively respond and say something about the analogy. I even send emails to local business owners, and we'll call out a local restaurant and we'll say, "Hey, we think this local restaurant's great. Have you ever tried it before?" We'll have people completely ignore the message that we sent, and then they just talk about how much they love that restaurant. So, when we send these personalizations, I know that they're working based on the way that people are reacting. I've never gotten a reaction to the analogy. I was wrong, and I'm sorry.
Mention customer case studies in your outbound emails. Cent has really changed the game of the kind of data that you can collect for your outbound campaigns. One of our favorite things to do is collect the case study off of the customer website. If there's a way that we can directly tie it to the offer, we'll do that, or we'll make a nice PS line because nobody really knows that you can automate this kind of research, and it really makes your outbound email stand out. So, let's say we're trying to sell advertising services, and we say, "Hey, notice that you worked with Intercom in the past and you helped them with XYZ." You really want to include that so it looks really relevant. What you then can say is, "How are you getting more customers like this? What's your plan to close another enterprise deal like Intercom?" Something to connect the dots about how your advertising services could help them do that. And that's just one thought. There's tons of ways you can tie in case studies. And then, if there's no way for you to tie in the case studies, sometimes I even like to just send a PS line, "Hey, if we never connect, I saw the case study of how you guys helped Intercom do XYZ. Really impressive. Kudos." Even if we never meet.
Don't use AI to write your entire email. One of the things that I see people still trying to do is they're using OpenAI, Claude, whatever it might be, to write their entire email. When that's not really necessary, one, you're not totally sure what it's going to output on the other side, and two, you don't need it to output everything. You want to be in control of your messaging so you can split-test things and understand what's really working. Often times, in this previous part of this video, we said we want to mention a case study. I don't need AI to say, "I saw the case study on your website about..." I don't need AI to generate that. We can keep it as static text. We don't need it to all create that. We could focus all of our horsepower, all of our energy into prompting, into just getting that AI line as good as we possibly can, and we don't have to focus about all of the other things.
This was Eric Nski from Growth Engine X. If you like these 25 topics, we can do a part two for 25 more.