The Bet: Owning the Entire Cold Email Stack

I learned the dependency lesson the hard way at 16.

I was running Ken Group - a marketing agency for course creators. We were working on a third-party platform that hosted courses. Amazing results. Clients were happy. I was at $16k MRR (huge at 16).

Then the platform raised their commission by almost 3x. Overnight. No warning.

We went from profitable to underwater in a single email notification. There was nothing we could do. Our entire business ran on someone else's rails, and they changed the price of the ticket.

That's when I made a rule: never depend on a single point of failure you don't control.

Fast forward to Ken AI. We're sending 1M+ emails per month. And for the first year, we did what every cold email agency does - stitched together Instantly + Clay + Smartlead, and made it work.

Until it didn't.

Most cold email agencies run on vendor tools. It's fast to set up, low risk, and you can launch an agency in a weekend. I don't blame anyone for going that route. We did it too.

But here's what happens at scale:

Vendor raises prices? You eat the margin. Vendor's shared IP pool gets flagged because some other customer sent garbage? Your clients' deliverability tanks, and they blame you. Vendor goes down at 11 PM on a Tuesday? You sit there staring at a dashboard you can't control, waiting for someone else to fix your problem.

And the costs are brutal. If we ran our entire workflow on existing tools like Instantly and Clay, it would cost us more than $40K per month in API and tool costs alone. At our volume, the vendor tax is a growth ceiling.

So we're building everything ourselves. The email servers, the sending infrastructure, the verification pipeline, the AI personalization engine, the deliverability monitoring, the campaign management. Every piece of the cold email process.

We built Ken Send - proprietary email infrastructure on dedicated IPs. Our own verification pipeline that triple-checks every email (filters out 78% of risky contacts before they touch our servers). AI personalization trained on 30,000+ words of real copywriting per client. Deliverability monitoring that replicates Gmail's spam algorithm and rewrites emails until they pass.

The results so far: Ken Send gets 25% better reply rates than the tools we replaced. 80%+ inbox placement. 4x reply rate vs industry average (3% vs 0.8%). Click rates at 16% - 8x industry average.

Those numbers aren't because we're smarter than Instantly or Smartlead. It's because we control every variable. When something underperforms, I don't file a support ticket - I go fix it.

Right now, the operational cost of running this model is high. Really high. And honestly - it's not yet worth it at our scale.

We're 8 people. Distributed between San Francisco and Moldova. Trying to build what most companies staff entire engineering teams for.

More things break. When you own everything, every failure is yours. No vendor to blame. Last month, a DNS misconfiguration took down sending for 3 hours. That was on us.

Engineering bandwidth is finite. Every hour on infrastructure is an hour not on product features. We're building a SaaS product alongside all this. The tension is constant.

Hiring is harder. I need people who understand email protocols, deliverability algorithms, IP warming strategies. Not exactly a deep talent pool.

It's exhausting. Some weeks I wonder if we should've stayed on vendor tools and just focused on growth. I'd be lying if I said the thought doesn't cross my mind.

At our current scale, a vendor-based approach would genuinely be easier and probably cheaper. I know that. My team knows that.

Because I've seen the ceiling.

The biggest lead generation agencies I know top out at $5-7M ARR. That's their ceiling. And it's not because demand runs out - it's because their margins compress, their vendor costs scale linearly, and they can't differentiate beyond the tools everyone else uses.

My goals are way higher than $5-7M.

Agencies using vendor tools will always be capped by those tools. Their deliverability ceiling is whatever Instantly can achieve. Their cost floor is whatever those vendors charge. Their innovation speed is whatever those product teams prioritize.

We don't have those caps.

When we find a deliverability technique, we ship it the same week. When we identify a cost optimization, we capture 100% of the savings. When a client needs something custom, we build it - not "request it as a feature" and wait 6 months.

The bet is simple: short-term pain for a long-term moat that no vendor-dependent agency can replicate. If we get through these next 3 months and every piece of the machine works together, the unit economics become absurd. Every additional client drops almost straight to the bottom line.

It's not worth it today. But it will be. And by the time competitors realize they need to build their own stack, we'll be years ahead.

I'm not saying every agency should build their own email servers. That would be terrible advice for most people.

But if you're building a company where the core product IS the infrastructure - where control and performance are the entire value proposition - then at some point, "buy" has a ceiling. And that ceiling is lower than you think.

We're betting that building everything ourselves is the harder path that leads somewhere most agencies can't follow. Ask me in 6 months if it was worth it. I genuinely don't know yet.

But I'd rather fail building something no one else has than succeed running the same vendor stack as everyone else.

If you're making a similar build-vs-buy decision in your business, hit reply. I'd love to hear how you're thinking about it.

Cristian