Cold Email & Outreach Strategy

One of the founders on a recent mastermind call runs a product that costs $13 a month. He has spent more than a decade building an organic engine that sends millions of free clicks to his site every month, and he came to the call with a specific question: which AI platform should he build an automated email lead gen system on top of? The plan was to have AI produce a valuable custom asset for each prospect, a tailored report or a visual, and use that as the opening move. His theory was that if AI can generate that asset for a nickel, the economics of cold outreach finally work at a price point where they never did before.

I gave him the honest answer, which is that I have never seen cold email work when the annual contract value is under about $2,000, or for a product priced under roughly $400 or $500 a month. That is not a law of physics, and I told him to test it anyway, because founders who failed at something have a habit of telling everyone else it is impossible. But you should understand what you are swimming against before you spend six months finding out.

Start with the response rate math, not the tool

Outbound needs a high contract value because response rates are far worse than the dashboards suggest. On a decent cold campaign, maybe one person in a thousand is genuinely interested. On a great one, two or three. When you do win, that win has to be worth $5,000 or $10,000 or $50,000 to make up for the nine hundred and ninety-nine people who did nothing.

The open rates in your sending tool are close to fiction. Those 50% open rates are mostly cybersecurity bots scanning inbound messages, and clicks have the same problem: assume roughly 80% of the clicks you record are bots. If you want to know how many real people took in your message, assume one in ten when you are doing everything right, and be ready for it to be closer to one in twenty.

Now run the AI asset idea through those numbers. Say the asset costs you five cents to produce. If only one in twenty of your emails is genuinely seen by a person, you are really paying about a dollar for each human who lays eyes on it. If one in a hundred of those people turns into a qualified lead, you are at roughly $100 per lead. That is fine if your customer is worth $15,000 a year. It is a catastrophe if your customer pays you $13 a month, and no amount of clever prompting closes a gap that size.

Before you pick a platform, get these three numbers on paper:

• Your true annual contract value. Not the aspirational number, the actual average across the accounts you close today.
• Your real seen rate. Strip the bot opens and bot clicks out of your reporting before you calculate anything.
• Your all-in cost per qualified lead. Include the sending infrastructure, the data, the AI cost, and the hours somebody on your team spends running it.

One founder who runs outbound for a living put a useful benchmark on this. In-house, with no agency in the middle, he gets his cost per lead to around $18 on contracts worth $1,000 to $2,000 a year. Add agency markup on top of the same motion and it stops working, because the markup eats the margin that made the campaign viable. At a modest contract value, you almost certainly have to run outbound yourself.

What AI actually changed about cold email

For about a year, AI personalization in cold email was genuinely novel. You could send a message that was specific enough to the recipient that it felt hand-written, and open rates, replies, and meetings all reflected that. A lot of founders built real pipeline in that window. Then the tools got popular, the whole market adopted them, and the advantage collapsed into the baseline. Everyone is inundated now, and the same message that stood out eighteen months ago blends into thirty others.

This is the part of the story the AI-will-fix-it theory misses. AI did not make outreach cheaper for you. It made outreach cheaper for everybody, and the second-order effects are what hurt:

• The novelty is gone. A personalized cold email was remarkable eighteen months ago. It is now the baseline, so it no longer buys you attention on its own.
• Volume is up everywhere. Cheaper production for you means cheaper production for every competitor sitting in the same inbox.
• Deliverability got tighter. Inbox providers responded to the flood by getting stricter, so simply getting delivered takes more work than it did.

What has not stopped working is real personalization in the first email, and the effect size is big enough to be worth the effort. Founders running this at volume report that sending something genuinely personalized in the first message roughly triples reply rate, and it triples the positive reply rate too, not just the total. The lift comes from doing research the person on the other end can tell you actually did. That is the piece AI helps with, and it is a very different claim from the idea that AI lets you brute-force a low-ACV product into working.

The list is where outbound is won or lost

Most founders spend their energy on copy and sequence design when the biggest gains are sitting in the list. The best approach I have seen is a hybrid: use a model like Claude to research and assemble a list of companies that match your criteria, then load that account list into LinkedIn Sales Navigator, which requires the top-tier subscription, and use its filters to find the right people inside those accounts.

There is a filter most people ignore and shouldn't. Narrow to people who have posted on LinkedIn in the last 30 days. They are more active online in general, and they reply at higher rates even when you reach them by email. That list will be smaller than you want, so work it first, then go back for everyone else. The sequencing matters because your sending reputation is built on your early results.

On data sources, one member made a point worth repeating. Scraping tools pull from a lot of places on the internet, much of it stale, while people tend to keep their LinkedIn profiles current, at least at the executive level. If you are targeting senior buyers, LinkedIn data is usually fresher. If you are targeting frontline employees who don't maintain their profiles, they are probably not maintaining a presence anywhere else either, so no tool is going to save you.

• Build the account list first. Use AI research to decide which companies belong on it before you go looking for people.
• Filter for active humans. Prioritize prospects who have posted in the last 30 days, then work the rest as a second wave.
• Enrich, don't just scrape. Apollo for the underlying list and Clay for detailed enrichment is a combination several members use well.
• Check your assumptions on seniority. Profile data is only reliable for the people who have a reason to keep it current.

Only pay for the people who raise their hand

If you are going to spend real money producing a custom asset for each prospect, do not attach it to the first email. Offer it and wait for the yes. The opening message says you have built something specific to their business and asks whether they want it, and only the people who reply get it. A lot of senders do this, so it is hardly novel, but it cuts production cost sharply and it usually helps deliverability too, since the first email carries no link and no attachment.

• Offer first, deliver on reply. Produce the expensive asset only for the people who have already raised a hand.
• Generate on click. If you do send a link, build the asset at the moment of the click rather than in advance.
• Screen the bots out. Put a Cloudflare check in front of the generation step so security scanners cannot trigger it.

That last one matters more than it sounds. If roughly 80% of the clicks on a cold campaign are automated scanners, a generate-on-click model without a bot check means most of your AI spend goes to security software rather than buyers. One founder in the community added a Cloudflare check in front of a similar asset on his own site and saw bot traffic drop enormously. That single change can take most of the waste out of the model.

When the answer is that outbound is the wrong channel

Sometimes the useful advice is that you are forcing a channel that will never fit your price point. If your product sells for $13 a month, cold email is unlikely to become your growth engine no matter how cheap AI gets. SEO and affiliates are what tend to work at that price, and if you already have a large organic base, the better question is what else you can build on top of that traffic.

The other path is to change what you are selling rather than how you are selling it. Build an enterprise version. Go after companies that could deploy it across 100 or 300 seats and pay $1,000 or $2,000 a month for a company-wide rollout. Now the contract value is high enough that the math works, and the same list-building and personalization effort that was hopeless against SMB buyers becomes a reasonable investment. That move usually beats optimizing a channel that was never going to fit.

• Under $2,000 ACV. Lean on SEO, affiliates, and product-led motions rather than cold outbound.
• $2,000 to $15,000 ACV. Outbound can work, but only if you run it in-house and keep cost per lead tight.
• $15,000 and up. Outbound works well, and it justifies deep personalization and account-based effort.

Treat email as a touchpoint, not a closing tool

The biggest shift in how we think about cold email is what we expect it to do. The days of sending a cold email and booking a demo directly off it are mostly behind us. It still happens, but it is much harder to pull off than it was two years ago, and building a growth plan on that assumption sets you up to be disappointed.

So we use email almost entirely to get a cold contact onto our website. Once they are there, retargeting can pick them up, and they enter the top of a funnel with several other channels feeding it. If we deliver a few valuable pieces of content along the way and get them to the site, we count that as a win, even if nobody books anything that week. The quality of the content is still the single biggest factor in whether any of this works.

• Lead with content, not an ask. The email's job is to deliver something worth reading, not to book a meeting on the first message.
• Optimize for the site visit. Getting a cold contact onto your website is the win, because that is where the rest of your system can pick them up.
• Hand off to retargeting. Once they land, always-on retargeting keeps you in front of them at a fraction of the cost of another cold send.

That reframing takes the pressure off the email itself and puts it on the system around it. Outbound stops being the thing that has to close and becomes the thing that gets you into the consideration set. Founders who make that shift stop obsessing over reply rates and start asking whether the whole set of channels is compounding, which is the harder question and the one worth answering.