Scale Your Agency or Get Better Reply Rates

What I Watched This Week

Two videos came across my feed this week that are both worth breaking down. They cover completely different angles of the cold email game, but when you look at them together, a pattern shows up. Both creators are trying to solve the same underlying problem: how do you generate consistent, scalable results when the old playbook is dying?

One is a Hormozi consultation clip with a cold email agency doing $2.2M per year trying to figure out how to get to $10M. The other is a demo of using AI to build micro SaaS tools as cold email lead magnets. Very different surface topics. Same root question: what actually moves the needle now?

Let me break both down.

Video 1: The $2.2M Cold Email Agency and the Performance Model Trap

What the Video Is About

This is a coaching session clip where a cold email agency owner walks Hormozi through his business. The agency works with B2B SaaS companies with TAMs over 10 million users and B2B service businesses with LTVs over $50K. They got to $2.2M on $10K/month flat retainers and the owner wants to switch to a performance model to scale faster.

The core pitch: lower the entry price to $5K so it's easier to close, then stack performance fees on the backend. The owner mentions sending 10 million emails per month for some clients and has one client paying around $75K to $100K per month on the performance model. He frames that one client as proof the model works.

Hormozi pushes back immediately. He points out that $10K per month for 10 million emails is absurdly underpriced. Then he gets to the real issue: performance models only work when you control the money flow or have absolute payment transparency. His Shopify agency example is the clearest version of this. If you run the Shopify store, you see all the revenue, you take your cut, and you remit the rest. Clean. But if a brick and mortar guy closes a $50K deal and you're supposed to get a percentage of that, how do you know? You set up a system that rewards them for lying to you, and some of them will.

Where Hormozi Is Right

He's completely correct about the enterprise customer qualifier. He frames it like this: if you're dealing with businesses that have real assets, obligations, and something to lose, they tend to honor contracts. If you're dealing with SMBs on a performance deal with no transparent tracking, you're operating on goodwill. US contracts don't give you much practical recourse, so the payment risk is real.

I've seen this exact issue play out. The performance model sounds incredible in the pitch meeting. The math makes perfect sense. The client loves it because the downside risk feels low to them. But the moment results come in and the invoices start climbing, some clients get creative with attribution. Suddenly that deal "would have closed anyway." You're in a conversation about whether your outreach actually caused the revenue, and there's no clean way to win that argument without owning the tracking layer entirely.

Hormozi's suggestion is solid: charge a significant one-time upfront fee to find message-market fit, then introduce a second tier once you've proven the messaging works. He even says you can double or triple your close rate by switching from a $1,500/month recurring offer to a $5,000 one-time fee. Same logic applies here. One-time transactions have less psychological resistance because there's no open-ended commitment.

What I'd Push Back On

The agency owner is underselling himself and Hormozi kind of lets him off the hook on that. If you're sending 10 million emails per month and you're charging $10K, you're not even covering infrastructure at any meaningful scale. That's a unit economics problem before it's a pricing model problem.

When I look at the numbers he gives, Hormozi does the napkin math and comes out to roughly 20 clients at $10K each. That's a mid-market agency volume, not enterprise. The one $75K-$100K/month client is the outlier, not the proof point. One whale does not validate a model change.

His 30% annual churn number is the most important number in the conversation, and it gets glossed over. He says it's hard to measure because he keeps changing the offer. That's a red flag. If you can't measure churn on your current offer, you don't actually know if your model is working. Fix your tracking before you change your pricing structure.

The real problem he has is what I'd call a positioning problem dressed up as a pricing problem. He's trying to serve both SMBs and clients with $50K+ LTVs on the same team with the same infrastructure. Those are two completely different client profiles with different risk tolerances, different contract behaviors, and different economics. You can't build a performance model that works for both at the same time.

My recommendation: pick one. If you want the performance model to work, go upmarket exclusively. Qualify hard on company size, revenue, and whether they have CRM infrastructure you can plug into. If you want to keep the volume, charge based on email volume like a utility. Price per million emails, not per month. Transparent, scalable, and doesn't require you to argue about attribution.

If you want to see how we think about pricing for outreach infrastructure, the cold email tech stack breakdown covers how we structure costs at volume. And if you're building an agency with enterprise clients in mind, the enterprise outreach system is a better starting point than trying to retrofit an SMB model.

The One Thing Worth Implementing

Split your offer into two stages. Stage one is a fixed-fee engagement to find message-market fit. Stage two is the scaled send once you know what converts. This removes the recurring-vs-one-time friction from the sales conversation and lets you qualify clients before you're locked into a performance deal with someone who has no CRM and a bad memory about which deals closed.

Video 2: Micro SaaS Lead Magnets and the Reverse Lead Magnet Concept

What the Video Is About

This one is a demo and framework video built around what the creator calls the "reverse lead magnet." The evolution goes like this: generic PDFs stopped working. Then AI-generated custom copy sequences started working, got overused, and stopped working. Now the play is building tiny web applications using Claude Code that solve a real problem for your prospect and delivering them as the cold email hook.

The claim is 4% reply rates across verticals that were previously considered impossible to crack via cold email, including AI automation agencies and cybersecurity firms. The examples are concrete: an SEO agency lead magnet that scans a prospect's backlink profile and generates an AI search optimization report, all triggered from a single URL click. An AI automation agency tool that detects what software stack a prospect is using and generates pre-built workflow automations they can import in one click.

The pitch is that you build the tool once per vertical, and it works across every lead in that space. An SEO agency tool works for dentists, chiropractors, e-commerce brands, anyone who has a website. Build it once, send it to thousands.

The mechanical process: write a prompt for Claude Code describing what you want the app to do, let it build the frontend and backend logic, host it, and use the app URL as your cold email CTA instead of "can we jump on a call?"

Where This Is Genuinely Smart

The concept of reciprocity as a conversion lever is not new, but this execution is. The previous version of this idea was "reply and I'll send you a custom email sequence." That worked for a while because it implied personal effort. This version delivers something that functions immediately, requires no implementation work from the prospect, and actually demonstrates competence rather than just claiming it.

The retargeting angle is something almost no one talks about. When you send someone to a URL, that URL is a landing page. You can fire a Google Ads pixel, a Meta pixel, or any other tracking event when they load it. So even if the prospect doesn't book a call, you now have them in a retargeting audience across multiple platforms. Your cold email just became the top of a multichannel funnel. That's a meaningful upgrade from a reply that goes nowhere.

The email capture element is also worth noting. If the app requires them to enter an email to access their results, you've turned a cold prospect into a warm opt-in. They gave you permission. Now you can follow up by email without it being a cold outreach anymore. That changes the legal and psychological dynamic entirely.

The scalability math also holds up better than it sounds. If it takes one hour to build a tool per vertical and that tool works for every lead in that vertical indefinitely, the cost per campaign drops toward zero after the first build. Compare that to personalized first lines, custom video loom messages, or individually crafted case studies. This wins on unit economics at volume.

Where I'd Add Nuance

The 4% reply rate claim is the headline, and I believe it's real for specific niches. But I'd want to know what volume those numbers are based on and whether the reply quality holds up. A reply that says "what is this?" counts as a reply. What matters is meetings booked and pipeline created. Reply rate is a vanity metric if the replies aren't converting.

The other thing worth thinking through: this strategy has a shelf life. Right now it works because it's novel. The perceived effort is high because building a custom web app still feels expensive and rare. As more agencies adopt this, the novelty wears off. Prospects will start recognizing the pattern. "Oh, another AI-generated tool in my inbox." The strategy itself is sound, but the execution window where it has maximum leverage is now, not two years from now.

There's also a deliverability consideration that doesn't get mentioned. If your cold email contains a URL and that URL is going to a freshly created domain or an app with no domain authority, some inbox providers will treat it with suspicion. You want to send the URL in a follow-up or gate it behind the initial reply, not put it in the first touch. Use tools like Instantly or Smartlead to test deliverability before you scale any campaign where the CTA is a link click rather than a reply.

The technical barrier also needs honest acknowledgment. Claude Code is genuinely impressive and the demo makes it look simple. But "describe what you want and it builds it" skips over the prompt iteration, the debugging when the app breaks for a specific browser, the hosting setup, the domain configuration, and what happens when 500 prospects hit the app at the same time and it goes down. These are solvable problems, but they're not zero-effort problems. Build the first version with enough time to break it before you send it to real prospects.

For finding the right prospects to send these to, I'd recommend using a verified B2B list so you're not wasting app builds on bad data. The ScraperCity B2B database is where I'd start for building targeted lists by industry vertical before deploying this kind of campaign.

The One Thing Worth Implementing

Pick one vertical you already serve or want to serve. Build one micro tool that solves a specific problem for that vertical. Something that takes a URL as input and gives back a useful output. Test it on 100 prospects before you scale. Measure meetings booked, not just reply rate. If the meeting rate is above 1% on cold, you have something worth scaling. If it's not, the tool isn't solving a real enough problem yet.

The Pattern Across Both Videos

Both of these videos are grappling with the same core problem from different directions. The market for generic cold email has gotten saturated. Flat retainers at commodity prices are hard to defend. AI-written sequences that everyone recognizes as AI-written are losing their edge.

The answer both creators are pointing toward is the same: raise the perceived value of the initial interaction. One is doing it at the pricing and delivery model level, moving to performance to signal confidence. The other is doing it at the first-touch level, delivering something real before asking for anything.

The mechanism is different but the psychology is identical. You're trying to shift the prospect's perception from "this person wants something from me" to "this person already gave me something." That shift is what changes reply rates, close rates, and churn rates simultaneously.

I've written about this in the context of offer construction before. The best cold email doesn't ask, it delivers. And if you can deliver proof of competence before the first sales conversation, you've removed the biggest objection most prospects have, which is "I don't know if these people can actually help me." For practical scripts that apply this framework, the killer cold email templates page has versions built around lead-with-value structures.

One thing I'll add from my own experience: the clients who pay the most and stay the longest are almost always the ones who received something useful before they ever got on a call. Not a PDF. Not a generic case study. Something specific to their situation. Whether that's a custom audit, a personalized tool, or a targeted insight about their business, the pattern holds. When you put real effort into the pre-sale experience, you attract clients who respect what you do. The clients who are shopping on price alone never see the value because they were never looking for it.

What to Do This Week

If you run a cold email agency and you're thinking about switching to performance pricing, do one thing first: audit your current client base and separate the ones with clean CRM tracking from the ones without it. Only the first group is viable for performance deals. The second group needs to stay on a fixed model until they have the infrastructure to track attribution cleanly. Do not change your pricing model company-wide until you've validated it with the subset of clients where payment risk is near zero.

If you want to test the micro SaaS lead magnet concept, pick your highest-converting niche and build one tool this week. Keep it simple. A site scanner that outputs a one-page report is enough to test the concept. Send it to 50 prospects, track replies, and track meetings separately. Give it two weeks before you make a judgment call. The data will tell you if it's worth scaling.

Both of these ideas have real merit. Neither is a magic button. The operators who win are the ones who test systematically instead of overhauling everything at once based on one video.