your best customers

Hello Predictable Revenue community,

“How the hell do they know who my best customers are?!”

That's the question a client asked me this week after receiving a cold email that felt suspiciously accurate. I'm sure you've received them too—emails claiming they've analyzed your case studies and built a special lookalike audience featuring companies that seem just right.

The simple answer is Clay. It’s a template, probably. It might be an AI solution but I doubt it. The real answer is more nuanced—and requires a deeper dive into the current state of AI-powered Sales Development Representative (SDR) tools.

tl;dr:

• AI SDR tools offer convenience but typically build mediocre lists.

• Bespoke, human-guided data beats generic, off-the-shelf data every time.

• Exceptional outreach combines human insight with specialized AI tools.

• Free Clay template below.

Here's the problem: AI SDR companies position themselves as “all-in-one” solutions. You get the data, the messaging, and the sending. All automated, no thought required. Convenient? Sure. But here's the catch—they're usually pretty mediocre at the most critical part: creating campaigns.

The advantage of an all-in-one tool is that it's easy; everything just works together without you having to think about it. The downside? Nothing is best-in-class. Everything's just…average. And that's not exactly their fault; excelling in multiple specialized areas is genuinely tough—even Apple struggles here. Think about it: Apple nails the integration of hardware, firmware, and operating systems. But software—especially AI-driven apps? They're hardly best-in-class. Doing one thing exceptionally well is hard enough; multiple things? Near impossible.

Let's unpack the core components any AI SDR platform must master:

1. Company Database (firmographics, technographics, signals, etc…)

2. Contact Database (complete with profiles, emails, and phone numbers)

3. Campaign Creation (selecting a small segment + finding the relevance)

4. Email Sending & Deliverability Infrastructure (yes, they’re different)

5. Messaging and Personalization

Each of these categories represents a fiercely competitive market on its own. Can one company realistically dominate five distinct, complex niches simultaneously? Probably not. Who's going to deliver better results: a specialized company excelling at one thing or an "all-in-one" player that's average at everything?

You might think "good enough" is acceptable here—but you'd be wrong.

Here's why: the most critical piece—campaign selection and design—still requires deep, human insight. Sure, AI can create "lookalike" audiences based on your existing customers. But think about this: your current customers probably aren't a perfect representation of your future customers. Likely, your list includes a mix of early adopters, perfect-fit recent clients, some less-than-ideal fits, and possibly emerging segments you're just now exploring.

Maybe you're about to attend a conference and want to target last year's attendees specifically, or perhaps a new vertical has suddenly become relevant. These nuanced decisions—picking precisely who to target and how to resonate with them right now—require human judgment. AI can automate parts of this process, but making the connections and crafting relevant messaging is something only humans can truly excel at.

It’s not for a lack of trying, we’ve been experimented with training models to output these exact types of ‘designs’ internally, tried out a bunch of vendors that make these claims, and nothing has been able to beat a human in the loop.

Here's the bottom line:

1. You still need a human in the loop to establish relevance and timing.

2. You can’t outperform your competitors if you're using the same "all-in-one" platforms that they are.

AI-driven SDR tools will undoubtedly become incredibly effective someday—but that day isn't quite here yet. (from the article, “We were losing 70-80% of customers that came through the door,” one employee said.) Saved you a click.

Back to my client's question. I created something in Clay faster than it took me to write this newsletter. Here’s the Clay template (and my affiliate link) I put together as an example. It's not perfect—I built it quickly as a demo—but it still does the thing. Am I going to use this for my outbound? No - it’s an overused attention hack. But did I do a thing really quickly? Yep. Am I a little to proud of myself about it? Probably.

Here’s the quick tl;dr how it works: the input is a company’s LinkedIn profile link, it enriches their profile, finds their case study page, creates an account classifier prompt, runs a lookalike account search, and uses the classifier to output the top 3 results.

If you plan to use this approach, here are two things to consider:

• Leverage Variables at Scale: Clay gives us the ability to insert variables into prompts at scale (run the same prompt but with merge tags). This lets us throw in data like company name, description, a scrape of their website, or the results of a Google search for their case study page into a prompt at scale. That’s the cool part.

• Use Classifiers for Automation: If you’re going to use this, make sure to write yourself a prompt that will look at each company and assess whether they’re a good fit or not, then do the same for your contacts. I wrote a little about this last week but it’s still on my mind. Last week I was focused on Account Classifiers and this week I’ve been working on Contact Classifiers. I’m finding that having both enables me to automate a good chunk of my list building process.

Once I come up with an idea for a campaign, I can just throw a big company list in and good fit companies/contacts get filtered by the classifiers and added to my campaign automatically. Once you’ve seen it in action, it feels like magic.

Why is this my go-to strategy? Simple: control and quality. Once set up, it's straightforward to execute, combines human insight with automated efficiency, and enables you to consistently use best-in-class tools without compromises.

So, how do cold email agencies know your best customers? They don't—not precisely. They're just leveraging smart tools and good-enough guesses. But if you want to go from good enough to exceptional, blend human strategy with specialized tools. That’s how you win.

Happy cold emailing and thanks for reading. You’re awesome.

Collin

PS - do you like easter eggs? The kind that developers hide in games… Well, here’s one just for you, loyal reader that made it to the PS. Here’s a simple Contact Classifier GPT I put together.

How to use it: Record a call where you talk about the types of companies that are a great fit for you, paste the transcript in, and it’ll output a prompt that you can use in Clay to filter any target list you feed it into Good Fits / Bad Fits.

PPS - how do you feel about my devious plan to hide cool stuff down here to get you to read the whole thing? Hit me back and let me know.

PPPS - receive any good cold emails that you’d like me to recreate a Clay template for? send ‘em my way and I might try this again next week.