The Cold Email Tech Stack We Use for Australian Agencies (2026)

This is the full list of cold email tools we use to run campaigns for Australian marketing agencies, job by job, as the stack stood in October 2026. It also covers the tools we stopped using and the ones we haven't used hands-on, because most tool lists are written by vendors and skip both.

We've run cold email for 100+ agencies across Australia and New Zealand for 7+ years, and we've changed tools often along the way. If you're deciding what to buy, this shows how many pieces the job takes. If you'd rather not assemble them, it shows what you'd be handing to us.

It's one part of our wider playbook, cold email for marketing agencies in Australia.

What cold email tools do you actually need?

Five jobs, with AI across all of them: finding companies, finding emails, setting up domains and inboxes, sending, and handling replies and client review. Here's what we use for each, and what we used before.

We've changed tools often and will keep doing so. The tool matters less than the process: verified data, protected domains, honest copy, and checks before anything sends.

Sending: which platform do we use?

Instantly.ai, since March 2023. It handles all our sending, inbox warmup, A/B variants and sequences, and its API runs our retarget automation. Our site says: "For email outreach, we partner with Instantly.ai."

Our settings in one line: plain text, open and link tracking off, stop on reply, 30 new-contact emails per inbox per day, and weekday business hours. The full list, the pricing and a comparison with Smartlead are in Instantly vs Smartlead for agencies.

Domains and inboxes: Google Workspace, Cheap Inboxes and Cloudflare

Our standard setup for each client is 5 alias domains and 10 Google inboxes a month, 2 inboxes per domain. Each inbox sends a maximum of 30 new-contact emails a day, which gives 300 a day.

• Alias domains, never the agency's main domain. They are lookalike domains that forward to the agency's website. One burnt domain must never touch the agency's real email. When an engagement ends, we transfer the domains to the agency, because they're its IP.
• Cheap Inboxes buys and manages the Google Workspace mailboxes. Every sender we run is a Google Workspace mailbox.
• Cloudflare is the registrar and handles DNS.

Finding companies: Discolike, Google local results and tech-stack lists

Discolike is our main company-discovery tool, in use since September 2025. We use it two ways: lookalike search from a few verified seed companies, and query search by industry, phrase, geography and headcount. We run it in stages of about 100 rows and grade every row before we spend anything on emails.

The rule that goes with it: when unsure, BORDERLINE, never YES.

Our honest view of Discolike:

• Its built-in fit score behaves like a topic-adjacency detector, not an ICP filter. It once scored a cleaning company 0.99 as a property-management fit. We only trust low scores, as a reason to drop a row.
• The 11 to 50 employees band is unreliable. One batch filtered to it gave 5 usable companies from 50.
• Adding extra phrase fragments to a query widens the pool instead of narrowing it.

Lookalike discovery finds companies databases miss, but its fit scores aren't qualification.

Google local results are scraped city by city for trades and local services. That's about 20 businesses per city per keyword, so six cities gives roughly 117 rows. BuiltWith builds lists by technology. It catches common platforms like Shopify and WordPress well, and it misses more niche software.

Finding emails: AnyMailFinder, decision maker first

We run AnyMailFinder in a fixed order:

1. A decision-maker search on every company, keeping only addresses marked valid.
2. A company-email search on only the companies where no decision maker was found.
3. A hard check that the two lists never overlap, so no company gets both.
4. One contact per company.

AnyMailFinder verifies addresses in real time and charges only when it finds a verified email.

Across six 2026 runs, AnyMailFinder found a verified decision-maker email for 51.7% of companies; adding company inboxes took coverage to about 85%. The result depends heavily on who you're emailing. On lists of businesses with 50+ staff, decision-maker find rates ran 92 to 98%. On Australian trade and small-manufacturer lists, they ran 31 to 43%.

Here is how one 1,000-company Australian list played out:

Small local businesses often only publish a company address, so the company-inbox version of a campaign, which opens with Hi, and no name, is how you still reach them.

One good finder with real-time verification, run decision maker first and company inbox second, replaced our old five-tool waterfall.

AI: Claude Code skills, with a human in the loop

We use AI (Claude) heavily for research, list grading, copy drafting and reporting, with human strategy and review. In practice, custom Claude Code skills do the repetitive, checkable work:

• Plan a list, then grade every Discolike row YES, BORDERLINE or NO with evidence.
• Run the email-finding sequence and format the results into campaign-ready tabs.
• Clean company names so {{companyName}} reads like a person wrote it.
• Draft copy by adapting real past emails from our library of 593 campaigns and 2,824 emails, then check the draft in code. The library holds no reply or meeting data: frequency is not effectiveness.
• Rewrite copy for retarget campaigns, and draft the review email to the client.

One measured, from-scratch 500-lead campaign reached the client's review Sheet in 22 minutes 48 seconds. About 5 minutes of that was AnyMailFinder. Copy is the slowest step, at 3.6 to 5.3 minutes per writer run.

AI does the repetitive, checkable work. Every number, client name and claim is checked in code or by a person before anything is sent. Matching against do-not-contact lists is code, never AI judgement. And every client-facing email is a draft that a person sends.

Curious how this would work on your own list? Book a call.

LinkedIn, replies and client review

HeyReach is our LinkedIn partner. We've used it hands-on since July 2026, so we make no performance claims for it yet.

An AI reply tool classifies replies as interested, information request or meeting request, and for some clients drafts a response. A person approves every forward or send.

Client review happens in a shared Google Sheet (the leads, one tab per segment) and a shared Google Doc (the copy), followed by one review email. As our site puts it, "nothing launches without your green light." Reporting comes from Instantly data.

Which tools did we stop using, and which have we never used?

Some tools we haven't used hands-on, so we won't review them. We haven't used Firmable or Smartlead for our own campaigns (we've only come across Smartlead on a client's own account), and we haven't run campaigns on Findymail or Hunter. We had an Apollo account in 2024 but haven't run client campaigns from it. For Smartlead, we compare it on public facts in Instantly vs Smartlead for agencies, and for Firmable and Apollo in Firmable vs Apollo for Australian B2B data.

How do you choose your own cold email stack?

Five principles from what we've learned:

1. Protect your main domain. Send cold email from lookalike domains you own, never from the domain you use for clients.
2. Verify every address. Expect to throw away 20 to 30% of a raw list.
3. Pick one email finder. Run it decision maker first and company inbox second.
4. Grade every company before you pay for emails. A wrong YES costs credits and trust.
5. Automate the checkable work and keep strategy human. Code for matching and checks, AI for drafting and grading, people for decisions.

Or have us run it. Our Agency Outreach System is the done-for-you version of everything above, and how we run campaigns walks through each step.