44,212 emails, 7 paid jobs for a pest control firm
James Shields put 44,212 emails behind 7 paid jobs for an Ohio pest control firm, while Nick Abraham spent the week telling operators their two-variant copy tests measure nothing. Both are arguments about which number you should actually be judging a campaign on.
Landing in the inbox
Christian splits sends by MX record, Illingworth narrows reads to five numbers
Christian says he pulls the MX record for every address before sending, then runs Google and Microsoft prospects as two separate campaigns, because the Google half has come back several times higher on the same list, copy and offer. His point is that a blended send hides the gap in the average.
open rate is a fugazi.
Campaigns and numbers
Shields says 44,212 emails got a pest control firm 7 paid jobs
James Shields says four campaigns for an Ohio pest control family business returned 7 paid jobs in 60 days. The run contacted 23,047 businesses with 44,212 emails, got 690 replies, and 41 of those agreed to a free inspection. He says the business had never sent a cold email and all prior work came from referrals and Google.
one offer: a free termite inspection
Christian holds his line that dials beat email per meeting
Christian held his position that calling books cheaper than email, backing it with 700k+ cold calls across 100+ B2B companies. For Eden Studio, 41,150 emails and 32,577 dials against the same market gave calling 33 of the 53 meetings, roughly 987 dials per meeting against 2,058 emails. He also says calling a positive reply instead of sending a link takes show rates from around 40% to close to 90%.
Their MDRs book about 6 qualified meetings a day.
Workflows and tooling
Lieben runs a campaign in 15 minutes, Illingworth cuts a week to an afternoon
Michel Lieben says he ran a campaign from one prompt in about 15 minutes in Claude Code with ten tools connected, tested on the 50 CEOs of the Forbes AI 50. FullEnrich found 48 of the 50, and the run wrote 13 emails and loaded them as drafts while he was on a break.
now takes an afternoon, and nothing about the output got worse.
Where they disagree
Abraham says tiny split tests tell you nothing, and kept saying it
Nick Abraham listed five things he stopped caring about after 150M cold emails for 1,000+ clients: clever copy, tiny A/B tests, big case study numbers, branded domains and sender names. His example of a worthless test is variant A at 2 positive replies against variant B at 1, a technical 100% lift that says nothing about the next 100,000 sends.
there are genuinely people who believe small split test data is valuable data
Also on the timeline
Gmail suggested a reply asking which tool sent the cold email
Frank Sondors says Gmail auto-suggested a reply to a cold email that declined and then asked whether a sending tool was used. One sender, one instance.
Instantly says its AI agents return 9.9 opportunities per 1,000 emails
Instantly says its new outreach agents produce 9.9 opportunities per 1,000 emails against a platform average near 1.25. Vendor's own numbers, with no list or offer detail attached.
Frequently asked questions
Should I send the same cold email campaign to Gmail and Outlook users?
No. Christian found that splitting campaigns by MX record (the server that handles email for a domain) between Google and Microsoft prospects showed significantly different reply rates on the same list and offer, so blending them together hides the actual performance gap.
What's a realistic conversion rate from cold email for local services?
One pest control business got 690 replies from 44,212 emails, which converted to 7 paid jobs, so judge local services campaigns on closed work rather than reply rate alone.
Is cold calling or cold email more cost-effective per meeting?
In one head-to-head test on the same market, calling generated 33 of 53 meetings at roughly 987 dials per meeting, while email needed about 2,058 emails per meeting, making calling cheaper per meeting when you have phone access.
How much do tiny A/B tests in cold email actually tell you?
Tiny A/B tests like comparing 2 positive replies to 1 positive reply show misleading results and don't predict performance on larger sends, so they're not reliable for making campaign decisions.
How fast can you build and launch a cold email campaign?
One campaign was built in about 15 minutes using AI tools that found prospect data, wrote emails, and loaded drafts automatically, though deciding who to target still remains the step that controls your reply rate.