Why you’re getting this: You’ve interacted with Haven or AdTribe before, and this is my weekly-ish update on what I’m learning while trying to grow both of them. Zero pressure to stick around, just click unsubscribe if you don’t want to get it.
Alright, let’s get into it.
LinkedIn is the new Facebook
On Tuesday I posted about a box of Cheerios on LinkedIn.
The box says 12 grams of protein on the front. Then you read the small print and it’s 12 grams only if you add milk. The cereal itself has 8. I buy anything with the word protein on it, so I took it personally.
As I’m writing this, that post has 14,523 impressions and 84 comments.

A few days earlier I posted why media buyers quit their jobs, based on 355 interviews our recruiters did. 349 impressions. Our October numbers post got 338.
That’s LinkedIn right now. Post something useful about your business and a few hundred people see it. Post about cereal and you get 37 times more.
But let’s be clear. Posts like that are good for the ego and they will never bring in business. I’d rather have 300 impressions that hit the exact people I want to work with than 10,000 from people who came to talk about cereal.
(The likes still feel good though lol)
The good news is that useful posts can blow up too. On Monday Illya, our Head of Performance, posted a breakdown of how Grüns runs its Meta ads. They have 1,300+ ads live, and he mapped all of it by funnel stage and gave the full breakdown away as a free download.
It went bananas. Over 100k impressions and 1,500+ likes in two days.
And here’s the part I care about:
320 people downloaded the breakdown
135 of them are ecommerce brands
75 said they spend over $50k a month on ads
6 calls booked for our creative strategy offer
Not all 6 are a fit, but a few are exactly the brands we want to talk to.
My cereal post got me a list of protein cereals to try.
Here’s Ilya’s post, go give him some love.
I rebuilt my AI SDR for Haven
Haven is the easy one to do outreach for, because everything is based on intent. If a founder posts that they’re hiring a media buyer, they need what we sell right now. I want to be in their LinkedIn and in their inbox the same week.
A few months ago I built the first version of this with Claude Code. It found hiring posts and sent the first message for me. It sent 271 of them and booked zero calls.
So last week I rebuilt it. The new one does all the boring work and leaves the talking to me. It runs on a small server three times a day, and here’s how it works:
Finds the posts. It searches LinkedIn for 17 phrases like “hiring a media buyer” or “looking for a creative strategist” and only takes posts from the last 48 hours.
Reads every post. Claude checks six things. Is it a real hiring post? Is the author the founder? Is it a role we place? Is it an agency, a DTC brand, an app or a SaaS company? Is the company in the US, Canada, the UK, Australia or Western Europe? And does the post say “no recruiters”? If it’s not sure, it says no.
Checks the profile. For the ones that pass, it opens the author’s LinkedIn profile and checks again with their real location and job title.
Adds them to my CRM. Name, company, the role they’re hiring for, a link to the post, a two-sentence summary and their email. If we already talked or they’re already a client, it skips them.
Sends a connection request. 10 a day max, so my LinkedIn account stays safe.
Keeps the pipeline updated. When someone accepts, it moves them to “To message”. When I write to them, “Messaged”. When they answer, “Replied”. I never update the CRM by hand.
Sends me a list every morning. At 6:30 I get a Telegram message: who to message today, who answered me, and who hasn’t replied in 5 days. For each person it suggests a first line, one short question about their post with no pitch.
The one thing it doesn’t do is send messages. I send every message myself.
It’s also pretty strict. In the first 5 days Claude read about 1,000 hiring posts and kept 31 founders. The most common reason for a no: the post wasn’t from the founder. Second: it was a recruiter or someone hiring for a client.
Is Haven a real business?
Haven started as a side hustle. One of our white label partners asked if we could find a full-time media buyer in Europe for his agency. I asked our HR to do it. Then more and more clients started asking for the same thing, so we built a separate website and went with it.
Financially though, Haven was just one of AdTribe’s offers. Everything sat in one P&L, so it was hard to tell if it was a valuable business or not.
When Haven became a team of 4, it was time to find out. So we split it into its own P&L.
It is a real business. Here’s what I see for the first nine months of the year:
Haven brings in about 13% of our total revenue and 14% of our profit
Gross margin: 59% at Haven, 53% at AdTribe
Profit margin: 25% at Haven, 21% at AdTribe
AdTribe’s best month was 29% bigger than its worst. Haven’s best month was almost 9 times bigger than its worst
Haven lost money in 3 of those 9 months in the early month when we didn’t have enough distribution
(To be fair, Haven doesn’t carry much overhead yet. Admin and accounting still sit on the AdTribe side)
The two businesses are hard in completely different ways.
Haven is easy to sell. The hard part is fulfillment, because finding great talent is very hard. And every engagement is a one-time project, which is why the months jump around so much.
AdTribe is the opposite. There are millions of media buying agencies out there, so even booking a call with a prospect is hard, and expensive if you run ads. The good part is that once we sign a client, they usually stay. Some have been with us for more than 5 years.
That’s it for this week. Stay hydrated, get some rest, and try not to lose your whole weekend to GTA 6 haha
Talk soon,
Dmitry
