933 positive replies from 685,403 cold emails.
This is every campaign we've sent for our clients, pulled from Instantly on Oct 13. The ones that worked and the ones that didn't.
Pulled from Instantly on Oct 13, 2026. A positive reply is a prospect saying yes to the offer in writing. It isn't a booked meeting. How we count
Six clients we run outbound for.
Who they sell to, what we built for them, and what came back.
411 positive replies, written in 8 languages.
We pulled 2.88M venues off Google Maps and wrote to each one in its own language.
We changed only the list. Positive replies went up 3.4x.
The database knew 12% of the hotels. We found the rest on Google Maps.
16 enterprise buyers said yes in 15 days. 12 were C‑level.
Each email pointed at a gap between what the company says and how its staff rate it.
The offer we added did 4.6x better than the one they came with.
Their favorite industries got zero yeses, so we followed the ones that replied.
Stopped on day 4. First deal closed on day 12.
Each email named 2 competitors ChatGPT recommends instead of them.
We stopped asking for a call and asked for a mailing address.
We scraped 1M auto shops for $23, then cut 58% of the list that turned out to be dealerships.
What 197 campaigns taught us.
We compare every campaign against the rest. These are the patterns that kept showing up.
The list beats the copy.
Point B ran the same 2-step email, the same free audit and the same mailboxes twice. The database list: 27,735 emails → 15 positive. The list we built from Google Maps: 20,121 emails → 70 positive. That's 3.4x per prospect contacted. So now we start with the list, and build it ourselves when the databases don't have it.
At small businesses, info@ wins.
Across 98,173 emails to leisure venues, enquiries inboxes ran at 0.556% positive per email and named contacts at 0.296%. At a venue with 1 to 5 sites, the owner reads info@ anyway. So we stopped paying to find their personal email.
Specific ideas beat a general pitch.
Same client, similar volume. A pitch about the whole category: 18,116 emails → 14 positive. Three ideas specific to each company: 20,175 emails → 42 positive. So every email we write now has something in it that's only about that prospect.
Write in their language.
Spanish 1.006%, German 0.976%, Polish 0.762% positive per email, against 0.353% for English to UK named contacts. We picked the language from each venue's own website, so a German-speaking venue in Switzerland got German.
Match the inbox provider.
Recipients on Google Workspace converted at 2.83%. Recipients on Microsoft 365 converted at 0.74% and said no 4x less often, which told us the emails weren't being seen. We added 30 Microsoft mailboxes, and Microsoft inboxes now get email from Microsoft senders.
Kill it on day 4 if it's not working.
An AI-search offer to DTC brands: 5,125 emails → 1 positive. We stopped it on day 4 and rebuilt it for B2B companies running Google Ads. The rebuild got about 6x the positive replies per email, and the client closed its first deal 12 days in.
How a campaign gets built.
Map the market
Who's in, who's out, exact titles, deal size, and who you don't want us emailing. If your deal size is too small for cold email, we'll tell you on the first call.
Build the list
We use contact databases when they have the market. When they don't, we build it from Google Maps, government registers and directories. One client's list covers 2,879,712 venues in 83 countries.
Check every company
Simple rules go first. On one list they cut 94% of 950,103 venues before we spent anything on AI. Then AI reads each website left and says yes or no, with a reason we can check.
Find the inbox that gets read
We crawl contact and team pages ourselves, look for the owner, and use info@ when that's what the owner reads. Every address is checked by a different tool from the one that found it. On one list, 12% were dead and never got an email.
Write for each prospect
What we personalize depends on the market. It might be their employee reviews, the competitors ChatGPT recommends instead of them, or something from their own site. We check every email before it goes out for blank fields, wrong language and phrases we don't use.
Send, then hand you the replies
Separate sending domains so yours stays safe, 2 inboxes each, about 30 emails per inbox a day, plain text, no tracking. 351 domains and a 1.75% bounce rate across 685,403 emails. Positive replies land in your inbox and CRM as they come in.
Cheaper than you'd think.
| Job | Volume | Cost |
|---|---|---|
| Google Maps scrape | 1,618,218 venues in 35 countries | $35 |
| AI qualification | 8,903 company websites read and judged | $3.87 |
| Traffic history | 20,000 domains × 18 months | $26.40 |
| Founding-year check | 8,907 companies against the national register | $8.91 |
| Native-language emails | 12,710 emails in 8 languages | $337 |
Some of the people who said yes.
We don't share their names, just who they are.
- The president of a listed paint maker's India business
- The chairman of a crop-protection manufacturer
- 3 CHROs at Indian manufacturers
- The founder and CEO of a developer-infrastructure company
- The capital-markets arm of a major Canadian bank
- A casino-resort operator
- A national cricket federation
- The CEO of a 49-location fitness group, 6 minutes after the first email
- A VR arena in a London shopping center
- A go-kart track
- A tiny-house hotel, and a lodge at the edge of a national park
- A city's chief administrative officer
What you get.
What we do
Get positive replies into your inbox. You sign off on the list and the emails before anything goes out. The domains and mailboxes are yours, and you pay what they cost.
What we don't promise
Booked meetings, revenue or a set reply rate. Turning a reply into a client is on your team, and it's where a lot of these deals were won or lost.
The full numbers
Every result on this page comes with how many emails it took. The campaigns that flopped are in the totals too.
Tell me how many opportunities you need. I'll tell you how many emails that takes.
15 minutes on your market and your deal size. If cold email isn't your channel, I'll say so.
How we count.
- Source
- Every client workspace we can read in Instantly, pulled on October 13, 2026. The earliest send is from May 2024.
- Sent
- Emails the platform delivered.
- Human replies
- One per prospect, auto-replies excluded.
- Positive reply
- Someone replied and said yes to what we offered. If the offer was a sheet, "yes, send it" counts. It doesn't mean a meeting got booked.
- Not included
- 2 workspaces with expired plans (one is our own outbound), campaigns on other platforms, and deleted campaigns. So the real totals are higher.
- No dollar figures
- Instantly gives every opportunity the same default dollar value. That's not a real deal size, so we leave it out.