One cold email used to need four paid tools. A scraper to find the business, an enrichment pack to find the contact, a verifier to check the address, a sending platform to send it. Four subscriptions, each priced per seat, each holding a piece of our data.
None of them talked to each other. The list got exported and re-imported at every step, and replies were rare. Nobody could say why.
- Reply rate
- 0.39%→4.75%
From 3,360 of my own sends.
One engine instead of four bills
I built OutreachOS to do all four jobs in one app, on a server I already pay for. It starts with a search term and ends with a tracked campaign, and nothing leaves the system in between.
- Leads come out of Google Maps and Google Search on proxy-backed workers.
- Each lead's site is scraped, scored for fit and enriched.
- Every address gets an MX and RCPT probe, catch-alls included, before a campaign can touch it.
- Five inbox pools handle warmup, sending, verification, backup and quarantine.
- Sequences run with A/B/C variants and pick a winner on their own.
- Sending stops the moment someone replies, clicks or unsubscribes.
The finding that mattered most
Once every send, open and reply lived in one place, the pattern was hard to miss. Role inboxes like info@ and sales@ were most of the old volume and almost none of the replies.
A person reads their own inbox. Nobody owns info@.
So the tool now keeps person emails only. Role inboxes are filtered out before they reach a sequence.
What changed
The reply rate went from 0.39% to 4.75% across 3,360 sends. That is roughly twelve times better, and most of it came from writing to fewer, realer people rather than from clever copy.
Four subscriptions became one server. And because the data never leaves, every campaign teaches the next one something.