How I built a cold outbound sales email machine

Profiles on building Jan 23, 2020

By Charles Cushing, Co-founder and COO @ Galleon

In 2016, I became responsible for sales at Galleon, the B2B SaaS company I co-founded. At the time, I’d never sold anything in my life.

Early on, we decided to try cold email marketing. Over the course of the year, we turned a barely-functional sales jalopy into a Lamborghini of a cold email distribution.

Cold email is a highly saturated channel. Getting it right takes time, if it’s possible at all. It may also only become effective in conjunction with other outreach methods. That said, if you can get it to work, the payoff is huge. It’s one of the cheapest and quickest outbound methods to scale.

Building Galleon’s sales machine took hundreds of hours of research, process mapping, documentation-writing, hiring, and coding. It was also not without dead ends or mistakes. Here, I’ll share a few of the lessons we learned along the way.

Step 1: Define your customer-message matrix

The goal of cold email is to find “customer-message fit”. You can’t scale until you have a message that reliably produces customers to open and respond.

Before you send a single email, you should create hypotheses around these two questions:

Who is your target customer? What industries and companies they work for? What positions and titles do they hold? Are they managers or individual contributors? Is your user the same as your buyer? (For more on this, read up on how to craft an Ideal Customer Profile, or ICP).

What message will get them to open your emails? What problem is your buyer facing? What value prop are you offering? How do you make your pitch? A master class on email copy is beyond the scope of this post, but again, there’s lots of info online.

Time required

A few days


  • For target customer: A set of bullets describing each customer persona you wish to test.
  • For messages: A set of subject lines and email bodies — 1 to 3 for each customer persona.

Tools used

  • LinkedIn for exploring personas
  • Google Sheets for keeping track of customer-message combinations

Step 2: Build your MVP system to run tests

Now comes the hard part: testing. You should test each customer-message combination to see what converts. If you have 3 personas and 3 messages each, that’s 9 tests. You’ll need to run each test enough times to get meaningful data. This adds up fast: 3 personas x 3 messages x 100 emails/test = 900 emails.

At this stage, you do not want to over-invest in infrastructure. Your system may mangle data and require lots of manual work — and that’s ok. It’s not time for perfection yet. Right now, you only need three things: email addresses, a way to reach them, and a way to collect conversion rate data.

Where do I get customer emails?

This is a trick question — before you start thinking about emails, you should build a list of the companies you want to target. Account-Based Marketing is a powerful concept, and it’s helpful to think of campaigns as targeting companies, not just individuals.

How do you build your first company list? At Galleon, our target customer was ecommerce companies. I spent a few hours Googling “Buy X online” and copy-pasting the top 5-10 hits. This exact method won’t work for everyone, but it’s ok to be creative and do things that don’t scale.

Once you have a list of companies, you’re ready to collect individual email addresses from LinkedIn. There are many great tools for magically acquiring email addresses from LinkedIn profiles. My favorite is LeadIQ.

Your lead-scraping process should go something like this:

  1. Go to a company’s LinkedIn page
  2. Click “View employees”
  3. Pop open the profiles of anyone who seems to fit in your customer-message matrix
  4. Read each profile
  5. If the person seems like a good fit, capture their email with LeadIQ
  6. Repeat for the next company

I scraped our first couple hundred leads myself. As tedious as it was, I learned a ton about our ICP in the process, and would recommend it to anyone getting started.

How should I send emails?

A common first instinct is to look for a CRM. Unfortunately, most CRMs don’t actually help you send emails. For that, you’ll need sales acceleration software.

After a brief, ill-fated run with Hubspot, Galleon switched to It was what we needed — a lightweight way to send email campaigns with auto follow-ups, and track stats as they came back. (Enterprise-grade name brands like Outreach, Salesloft, etc. are probably too heavy for this phase.)

You might want a CRM to track deals, but it’s usually not necessary for emails. We used Copper as a stop-gap and it was fine. These days, lots of people use Airtable.

Now that you’ve got emails and a way to send them, you’re ready to run some tests. Drop your contacts into your sales acceleration platform of choice. Tag your personas. Upload your messages, and fire away.

As emails go out, you should get stats back from your sales acceleration tool. Watch the conversion rates and the feedback customers are sharing. Use this info to kill the ineffective experiments and push more volume through the promising ones. Rinse and repeat.

There are three main reasons why it’s vital to set up an MVP system:

  1. You can verify whether cold email works at all, without making a huge upfront investment
  2. You can more easily re-factor your data models, something that becomes 100x harder to do once you’re using enterprise-grade software
  3. You can figure out which features and processes are mission-critical for you.

For Galleon, we learned that, yes, we could get acceptable conversion rate with email. We re-factored our data model literally dozens of times as we experimented with different sales pipeline configurations. And we made tons of process discoveries. For example:

  • We came up with 8 persona tags that became embedded in our system, used to programmatically determine who should receive high-touch vs. low-touch treatment.
  • We realized there were several non-standard features that we needed from our system, like the ability to support custom Salesforce picklists, and the ability to interleave other tasks (like cold calling and InMail) with email.
  • We were able to map out all of the status changes that needed to be automated. E.g. if a Company was switched to “Disqualified”, we needed an automation to auto-disqualify all of the attached contacts, prevent the adding of new contacts, and offer a default status to contacts if the company was later un-Disqualified.

This is just scratching the surface. Systems are inherently complex, and the MVP allowed us to iterate rapidly with minimal cost.

Time required

3-12 weeks, depending on your product.


  • An initial company list
  • An initial contact list with emails
  • At the very least, software that lets you capture emails from LinkedIn, send email drip campaigns, and capture response rates and sentiment.

Tools used

Note: This was NOT a good long-term stack and I don’t recommend it. Still, it meaningfully boosted our email throughput with minimal setup.

Step 3: Build your Ironclad system to push volume

You’re ready for Step 3 when you can say with conviction that you’re ready to ping 5,000 new contacts. Now, prepare to invest in your Ironclad system. Building a truly robust email machine system may take months, but if done well, it’s worth it.

The first thing you’ll want to do is to translate your learnings from your MVP into a system requirements doc. Hopefully, you can get very specific, including the data models you want, the sources and sinks for each piece of data, and the automations and features you require.

Then, you’ll want to go hunting for the key SaaS tools to support your system. The sales SaaS market is very crowded and it can be tough to tell what a given product does without getting a demo or trial version. As a reference, we did demo calls with probably two dozen different vendors for Galleon, and got trial versions of perhaps half a dozen.

Unfortunately, just picking the right tools usually isn’t enough. You’ll also have to make sure the tools can talk to each other. After narrowing it down to our few favorite systems, we tried configuring our data models in each of them, then established whether the data could be transmitted via direct integration or API. Surprisingly, this process knocked out several name-brand competitors for us, like

Finally, the right tools may simply not exist for your use case — either to deliver functionality you need, or to connect your systems. In this case, you may need to build custom internal tools. At Galleon, we home-built an internal tool called Broadside to manage all of the data transmission between our lead enrichment tool (LeadIQ), our sales acceleration tool (Apollo), and our CRM (Salesforce). How we chose these systems is a subject for another long post.

As a final step, we trained workers in India and the Phillipines to perform the remaining low-complexity human tasks, such as scraping leads from LinkedIn based on title, and doing certain types of data cleanup.

The result was our Ironclad system: a robust machine that could keep data clean, make sure no leads got dropped, and could support us through our next 10x of growth.

Time required

4-12 weeks to build, years in use.


  • An software + human machine that takes as input a list of target companies, and delivers as output sales meetings booked on sales reps’ calendars.

Tools used

  • LeadIQ for LinkedIn lead scraping
  • Apollo for sending email campaigns
  • Salesforce as our CRM
  • Broadside, a custom tool we built to mediate the other tools

Final thoughts

Using our MVP system at Galleon, it took me 4 days a week just to reach a dozen or two accounts. Using Broadside, it took me 2 hours per week to reach hundreds. Building a sales machine can be grueling, but don’t give up too early — the results are worth it.

A few final pieces of advice upon entering the madcap world of cold email:

Test, then automate. Early on, it’s often faster to do things by hand, rather than trying to find the perfect software tool.  Many companies seek automation from Day 1, but this is a mistake. Going 100 miles per hour is only useful if you’re driving in the right direction.

Always expect change. Even once you build your Ironclad system, you can’t fire the blacksmith. It’s inevitable that you’ll want to make tweaks to your process. At Galleon, we later added in email customization, InMail, and cold calling.

Learn from your competitors. You need to stand out from the crowd. It’s helpful to be subscribed to your competitors’ newsletters so you can ensure your messaging is truly different.