Key Moments
I sent 10,000,000 cold emails and learned this
Key Moments
Cold emailing 10M emails revealed that low inbox volume per inbox (30 emails) and a 3-email sequence are key to avoiding spam filters, but personalization is more effective than AI analogies.
Key Insights
Maintain an inbox volume of approximately 30 emails per inbox or per domain to stay under spam filters, scaling horizontally with multiple domains and inboxes.
Target 40-60% open rates, but understand that Apple iOS updates automatically trigger open flags, making reply rates the more reliable metric for engagement.
Email sequences should not exceed three emails; email number one is the most important, with diminishing returns on subsequent emails, and reusing target lists quarterly is recommended.
Leverage granular filtering tools like Clay to build a highly specific Ideal Customer Profile (ICP) by 'waterfalling' data points, such as company age, funding status, and founder experience.
Social signals, particularly engagement with or posting about LinkedIn content, were the most successful trigger in 2024, outperforming other common triggers.
When using AI for personalization, 'show your work' by citing data sources (e.g., 'Similarweb says you get X visitors') rather than making direct, potentially inaccurate claims.
Maintaining deliverability: Keep inbox volume low and scale horizontally
To avoid spam filters, it's crucial to maintain a low email volume per inbox and per domain. The current practice is to send around 30 emails per inbox. This is achieved through horizontal scaling, utilizing multiple domains and inboxes rather than relying on single platforms that might ban bulk sending. Tools like Instantly.ai or platforms with done-for-you setups can aid in managing this infrastructure smartly. By keeping the perceived volume low for email providers, you increase the chances of landing in the primary inbox rather than the spam folder. Warming up new domains and inboxes for approximately three weeks before launching a campaign is also recommended.
Open rates are a signal, not a definitive metric
While targeting 40-60% open rates on cold email campaigns is a good general goal, it's important to acknowledge the impact of Apple's iOS update. This update automatically triggers open flags for users with email accounts tied to the Apple Mail app, skewing open rate data. Consequently, open rates should not be solely trusted but can serve as a useful indicator of whether emails are landing in the inbox. Emails in the spam folder, for instance, will not trigger open trackers. A reply rate below 1% (excluding out-of-office replies) is a stronger indicator of deliverability issues than just low open rates. For troubleshooting, focus on reply rates and, if using custom domain tracking, consider setting up more domains and inboxes to start fresh.
Optimize email sequences: Three emails, then refresh
Across numerous A/B tests and customer data, the most effective email campaigns consistently consist of a maximum of three emails. Email number one typically yields the highest engagement, followed by diminishing returns for email two and three. Sending beyond three emails often leads to annoyance and an increased likelihood of being marked as spam. Instead of extending sequences with the same list, it's more effective to reuse your Target Addressable Market (TAM) list every quarter. This allows for a strategic pause, giving sufficient time (e.g., three months) for changes to occur within a prospect's business or role, making a subsequent outreach more relevant to their potentially shifted priorities. Forgetting previous cold emails is common, so the key is the timing of the business context, not repeated nagging.
Granular targeting and ICP refinement with 'waterfalling'
Effective cold emailing hinges on precise targeting. The 'spray and prey' tactic of broad filtering (e.g., industry plus employee count) is ineffective because a marketing director at a 20-person company is vastly different from one at a 500-person company. Instead, leverage tools like Clay to build a 'golden ICP' by 'waterfalling' data points. This involves layering criteria to create highly specific audience segments. For example, targeting accounting companies could first filter for 'new companies' (less than 2 years old), then 'recently funded' companies, and finally, companies where the CEO has no prior CEO experience. This tiered approach allows for distinct messaging for each refined segment, ensuring relevance and maximizing the impact of outreach by understanding what specific characteristics make a prospect most receptive.
Social signals emerge as the top trigger
In 2024, social signals proved to be the most successful trigger for cold outreach campaigns, surpassing other common signals like company age or new hires. This category encompasses various activities, such as individuals recently posting on LinkedIn, engaging with content, or discussing specific topics. For instance, when working with an offshore staffing company, referencing a prospect's recent LinkedIn post about a relevant topic led to significantly better response rates than relying solely on traditional triggers. Such personalized outreach, directly mentioning the post and drawing a connection to the service offered (e.g., 'I saw your post about X, you should recruit one of our virtual assistants'), demonstrated a higher level of engagement. It's crucial to filter out sensitive topics like political posts when implementing this strategy.
AI's role: Personalization over generic analogies
While AI offers powerful personalization capabilities, its application needs to be strategic. The most effective use of AI in cold email is to enhance genuine personalization, not to automate generic analogies. For example, referencing a local restaurant or a specific case study like a company's work with Intercom can be highly effective if done well. However, attempts to create analogies between your company and the prospect's situation, even with an AI's help, often fall flat. Prospects are more likely to respond positively to emails that feel deeply researched and tailored to their specific context. When using AI to gather data, 'showing your work' by citing sources (e.g., 'Similarweb indicates X website visitors') is essential. This approach attributes any inaccuracies to the data source rather than appearing to have made a poorly researched, direct claim, thus maintaining credibility.
Structure your email sequences strategically
The structure of your email sequence significantly impacts its effectiveness. The first email should be concise, typically following a 'why you, why now, your offer, social proof, call to action' format. Email two can then add more context that was cut from email one, serving as a deeper dive for interested prospects without overwhelming them initially. Email three should focus on lowering friction for a response, acknowledging that explicit objections exist. Instead of a direct meeting request, consider offering a lead magnet or a custom audit to encourage engagement from prospects who might not be ready for a full commitment. Breakup emails should not be about begging for a response but about finding the right contact person within the company, perhaps asking for an internal referral.
Vary value propositions and avoid using all your 'best content' upfront
Continuously switching up the value propositions within your cold email sequence is essential, as prospects may prioritize different benefits at various times. If an offer focused on saving money isn't resonating, pivot to highlighting how you can help them make more money or save time. Understanding the 'so what' behind the prospect's needs is key – why do they care about saving time or money? Digging deeper into these motivations can uncover more compelling messaging. It's also a common mistake to pack all your best data and personalization into the very first email, leaving subsequent follow-ups generic. Holding back some personalized details for later emails can make the outreach seem continuously researched and relevant, allowing for different angles and further engagement.
Mentioned in This Episode
●Software & Apps
●Companies
Cold Emailing Best Practices
Practical takeaways from this episode
Do This
Avoid This
Common Questions
It's recommended to keep inbox volume low, around 30 emails per inbox and per domain. Scaling horizontally with multiple domains and inboxes is the key, rather than increasing volume on a single domain.
Topics
Mentioned in this video
Referenced as a platform that can provide company information, suggested to be used and cited when showing your work to prospects about data accuracy in cold emails.
Cited as a data source for website traffic, but the speaker cautions that data from such sources can be incorrect, and it's better to show your work by attributing the source.
A tool recommended for list building, filtering, and creating segmented messaging based on ICP criteria. It's highlighted for its ability to manage complex data points and personalize outreach.
Mentioned as a data source that can provide information about companies. The speaker advises showing your work by attributing data from sources like Crunchbase to avoid being blamed for inaccuracies.
Mentioned as a platform that will ban users for sending too many emails, necessitating the use of multiple domains and inboxes for cold outreach.
Used as an example of a customer for advertising services, highlighting how a case study involving them can be used to draw connections and frame outreach.
The agency represented by the speaker, providing expertise and data-driven insights on cold email campaigns and outreach strategies.
Mentioned in relation to a specific email strategy involving a caveat to the typical three-email sequence rule.
Mentioned as a tool that can be used for email outreach, but cautioned against using it to write entire emails. It's suggested to use AI for generating specific lines or aiding in research.
Found this useful? Build your knowledge library
Get AI-powered summaries of any YouTube video, podcast, or article in seconds. Save them to your personal pods and access them anytime.
Try Summify free