← Back to essays

Learnings from Sending 100,000 Cold Emails Per Month

This essay covers practical tips, history of learning, industry direction, skills acquired, tech stack, data tracking, and the role of automation from my experience with cold email at scale.

Results Summary

We closed 9 deals worth $102,000 from cold email alone. Average deal size (excluding Sykes at $38,900) = $8,900. This represents roughly $50k per month over 2.5 months of active work.

The Four Components of Cold Email

1. List-Building

This comes from the ICP. List-building is founded on a hypothesis around a certain trait of a Person/Company that means they are a good fit.

  • Person → are they the decision-maker over the offer?
  • Company → Employees, Industries, Technology Stack, Predicted Revenue, Lookalike company to client, past experience, marketing efforts, recent content

Tactically: Platforms like Ocean.io, clay.com, builtwith are good for building lists for agencies. Cheaper option: Apollo.io's $60/month plan. Expensive: Prospeo, Leadmagic & Findymail. Verification with Million Verifier.

2. Copywriting

Generally, there's 2 elements to copywriting in cold email:

  1. How you are pushing pain/promoting your offer
  2. How you are making that personalised to them
Typically an accepted format:
Line 1 = Why You?
Line 2 = Why Now/Problem Statement?
Line 3 = Social Proof?
Line 4 = Call To Action/Offer?
Line 5 = feels personalized

3. Personalisation

AI personalisation & data extraction for websites is mainly useful to filter a list based on a variable e.g. all wix sites. However, it also plays a role in making the prospect think that you have taken the time to research them.

Personalisation Hierarchies:

  1. Best when used as a trigger for reaching out - LinkedIn posts, web traffic, case study on their website, recent news, hiring, software they've used, tech stack, competitor mentioning, recent funding
  2. For the sake of it - restaurant lines, weather lines — these still work because they signal to the prospect that you have done research

4. Deliverability

  • Sending good offers that don't go to spam often — biggest cause of deliverability issues is people marking as spam
  • Warmup for 2-3 weeks
  • Diversify infrastructure — gmail/outlook & other third party providers
  • No links anywhere, no spam copy, 25-30 emails per inbox, warmup campaign copy
  • Use servers related to the country you are sending to
  • 50% of inboxes in reserve capacity
  • Always buy separate domains. Never use own company domain

History of Learning

1. Had some experience with Clay & Smartlead earlier in the year when I tried to start the agency

2. Watched Nick Abraham talk about Cold Email. Launched campaign that night to Restaurants in Sydney with existing lead list on Instantly.ai. Got one close — Bernadette Gore

3. Weekend — Watched all of Nick Abraham's channel on Youtube

4. Started launching Smartlead Campaigns & setting up domains

5. Fired out emails continuously in large lead lists

6. Went deep into Eric Nowoslawski & Clay.com rabbit hole

7. Today — Collectively made $120,000 from Cold Email in 2.5 months of work

How I Would Learn It Again

My thinking on learning goes:

  1. Ingest tonnes of content about the topic — like literally 100 hours over 1-1.5 weeks
  2. Then, act on this content & create something according to that work
  3. Go back & watch content, which is more nuanced the second time around

Top 3 Mistakes

1. I would have learnt 2x quicker if I spent more money straightaway

I was fear paralysed by spending the money I earned the hard way via cold calling. As soon as we got clay, the learning curve was much quicker — being able to integrate with other tools.

Takeaway: Spend more money initially when learning/investing into a new skill. It catapults the learning curve.

2. Consumption/Creation Ratio Issues

Initially, the ratio of consumption to learning was really good. When watching Nick Abraham, I took away bullet point notes. But when I started watching Eric — I watched too many videos without trying to action what was said.

Takeaway: The consumption/creation ratio is not linear. The more you watch, the less you need to watch — especially of mainstream stuff.

3. Been more intentional about the learning process

This includes more documentation to come back to, trying to setup a system. Even simple notion pages can be helpful because that allows you to track your thoughts & come back to them.

Takeaway: Document thoughts throughout. Do a better job with documentation & note-taking.

Where Cold Email is Heading & Impacts for GHC Studio

General Industry Norms

Future of Cold Email

The Skill Acquired & Why It's Useful

Cold Email has been the catalyst for the most growth I've had in terms of thinking about outbound sales & how to effectively utilise it as a channel.

Your Offer is the Most Important Thing

To a cold audience, your offer is the thing that matters the most. Your personalisation, lead magnet only gets you so far. Because of the objective data & sheer amount of data you receive from cold email, you get very quick feedback on:

Demand Capture offers necessitate extreme amounts of volume. You are trying to cast a net as wide as humanly possible to reach as many people as possible.

Key Insights

Tech-Stack & Technology's Role

The High-Level Process

  1. List-Building: Apollo.io & Enrich IQ — scrapes the data. Clay.com for filtering based on intricate variables
  2. Personalising: Clay.com — superior data enrichment platform. Combines with OpenAI's API, Apify, Zenrows
  3. Sending: Million-Verifier for email validation. Smartlead for sending platform

Investment: We invested roughly $18-20,000 into cold email and got back $100,000 from it.

More Nuanced Role of Technology in Sales

Technology & the competence of web-scraping now allows you to get proxies/signals for different companies that accurately determine whether a company is in the market.

For example — exposure to how you can aggregate data from SimilarWeb & other providers makes it much easier to build cold-call lists with companies that find their websites valuable & are likely to pay more in the sales process.

The Role of Automation

My naive prediction for cold email was that we could simply send out 5-6k emails per day with basic homepage design. Close at a 8-10% clip & make tonnes of money.

Problem: This is not very profitable because typically those interested or that would buy because of a homepage offer are acting on desire. And desire historically does not pay as well as pain.

Nevertheless — Cold Email is fun, interesting & alluring because it is automated. Once you understand fundamentals of list-building, data enrichment & sending out emails, the things that are optimised over time are:

AI SDR Prediction

AI SDRs cost $3-5k per month for between 5-10k emails. That means you get 20 qualified leads per month. Paying about $500 per meeting. Average agency/SaaS company closes at 20% clip. Meaning that you pay $2500 to acquire a customer. For a $10k project, that would leave us with around $1500 profit. Horrible.

Tracking Data

Cold Email forces you to think about data: how many emails did you send, how many were replied to, how many replies were positive, how many positive replies booked meetings, how many booked meetings become 2nd calls, how many 2nd calls become closes.

Initially when we started, I cared a tonne about reply rates. We even got 8-12% reply rates on some campaigns where we filtered only for gmail accounts.

But you stop caring about reply rates when you sit on 10 calls on a Friday & realise 90% of them are broke. And only 2 of them was ever going to be a close.

Then, you realise: It was never about replies. It was about qualified leads that actually care about their website & have a reason to be on call.

Three Main Data Points We Track

  1. Sales numbers throughout different channels - % of revenue from % of clients, amount of projects below $5k, AOV from different channels
  2. Every Cold Email Campaign - which ones have performed the best historically — reply rates, positive replies, bounces etc
  3. Cold Email performance over time - past week, month & 2 months — allows you to see if overall performance has been good/bad

Caveat: I probably could've done this 1 or maybe 2 months earlier but no earlier than that. It takes a lot of data to be able to make a decision about what are best next steps. I'll be cautious going forward that I don't pre-emptively make decisions based on limited data or time.