The precision playbook

A case study on stalking your ICP across the internet

I was recently meeting with the founder of a company approaching $1 billion in revenue. We were talking about their demand generation program, and at some point he paused and said, “Hey, I keep seeing ads from this company Oso everywhere I go online. I noticed you’re advising them. How are they doing that? And why can’t my marketing team pull that off?”

I laughed, because Oso’s marketing budget at the time was a rounding error on an enterprise marketing budget. And yet here was a founder of a nearly-billion-dollar company getting stalked across the internet by their ads. (And yes, this company was in Oso’s ICP!)

The combination of AI-powered tools and data has enabled small teams to execute sophisticated campaigns on a shoestring. This is a case study on how Ege Ayan, Oso’s digital marketer, ran that very campaign.

“When you’re small, you need to be precise,” Ege explained to me, “A broad campaign can burn through a limited budget.”

Instead, when you’re small, you have to be surgical, by using data to find exactly the right people and follow them around like you have a much bigger budget than you actually do.

Good creative doesn’t need to cost a fortune

Before you can target anyone, you need something worth targeting them with. Ege wanted a polished video ad that would appeal to developers while also signaling the maturity of the company and product. Without a dedicated creative team or budget for an agency, Ege had to be scrappy.

Ege sat with Oso’s DevRel team to brainstorm on several concepts. After many iterations, he came up with the concept of showing a developer adding a role — the kind of thing that’s genuinely painful without a good authorization solution — and landed on a punchy call to action: Just call Oso.

Once he had nailed the messaging, he created a detailed creative brief with frame-by-frame instructions. By the time he found an animator on Upwork, that animator could focus on execution rather than strategy.

Total cost: under $1,000.

In this case, Ege outsourced the execution in a cost-effective way. But with AI video tools like HeyGen, it’s also possible for marketers to produce professional video content without outsourcing at all.

Get precise about targeting

The video was the bait. But bait only works if you put it in the right water.

Now that Ege had an ad that addressed a specific pain point, he set out to find the audience to target. Here is how he got the ad to follow that founder I was talking to all over the internet.

Step 1: Find the companies with the pain using Sumble

Ege started with Sumble. Sumble analyzes company websites, job listings, and people profiles to figure out their tech stack and key technology initiatives. Using Sumble, you can use natural language queries like “find me companies with over 1000 employees who are using Apache Kafka in the east coast of the US” to build lists of accounts to target.

Ege used Sumble to identify companies actively recruiting for roles that had terms like roles, role based access control, authorization, or something similar. These would be prime targets for this campaign.

Step 2: Enrich the company list with Clay

Ege then fed those Sumble accounts into Clay and ran them through Claygent (Clay’s AI agent) with custom prompts to further enrich and qualify each account.

“They were basically separate angles to determine whether they have complex authorization,” Ege explained. “It would check for evidence of multiple roles in their app, custom roles, or fine-grained access control.”

Step 3: Find the right people on all social media platforms, including Meta

While Ege could build audiences on LinkedIn targeting specific companies, he wanted to see if he could get better ROI by advertising on Facebook and Instagram. He hypothesized that he would get a better response on those personal platforms than on LinkedIn, which is flooded with corporate ads.

The challenge? Most data enrichment provides corporate email addresses. To match people on Meta, he would need personal data.

(Side note: I have to laugh at the irony of looking for personal emails for B2B marketing, as I had spent years at MongoDB trying to figure out the opposite problem: what the hell to do with the thousands of gmail, yahoo, and other personal emails we collected through Atlas cloud sign ups.)

Ege started by using Clay to identify personal emails. This quickly got expensive and he was burning through Oso’s limited pool of Clay credits. So he pivoted to using an enrichment provider called Primer, which specializes in helping marketers find B2B audiences on platforms like Facebook and Instagram.

Step 4: Experiment with different channels, even controversial ones

Ege wanted to take his “target a B2B audience on B2C channels” approach a step further, and he decided to experiment with Google Performance Max.

If you ask ChatGPT what marketers think of Google PMax, you’ll probably get answer along the lines of: Marketers have a very mixed (often polarized) view of Google Performance Max (PMax). It’s one of those tools where people either love the results—or deeply distrust how it gets them.

But Ege’s view was that the usual objections assumed you were going in blind. He wasn’t. With a clean, Primer-built audience already in hand, he had something most PMax campaigns lack: a precise list of who he actually wanted to reach. That de-risked the channel’s biggest weakness. He set aside some experimental budget to see if he could get results from this channel.

The results

Across the board, the campaigns delivered lower CPCs and higher click-through rates than typical broad targeting, a direct reflection of how much more relevant the ads were to the people seeing them. The Sumble-sourced accounts, in particular, outperformed standard retargeting, which makes sense. Someone actively hiring to solve an authorization problem is a warmer prospect than someone who once clicked on your website.

PMax was messier, as expected. Conversion rates were lower and yes, there was some junk traffic. But Ege did the math: the volume was high enough, and enough of those conversions were real opportunities, that the channel earned its place in the mix.

And then there is the benefit that you can’t measure. Let’s go back to the story at the start of this post. That founder, running a company approaching $1 billion in revenue, with a full marketing team, thought Oso was everywhere. He felt outspent. He wasn’t. He was just out-targeted.

That’s the whole game. You don’t need to be everywhere. You need to be everywhere the right people are looking. With the right data and the right tools, a one-person marketing team can pull that off.

Appendix: The Tools

For anyone who wants to replicate this playbook, here’s what Ege used and what each tool does:

• Upwork & HeyGen: Ege found the animator for the video on Upwork, which is a good reminder that great creative doesn’t require an agency. You could also explore tools like HeyGen to make AI-generated video content. (I’ve written about using HeyGen in a previous post.)

• Sumble: Surfaces company hiring data as an intent signal. If a company is recruiting for roles that indicate a specific pain point, that’s a warm signal worth targeting.

• Clay: Data enrichment and list building. Pull in a list of target accounts and enrich them with firmographic and contact data at scale. Feed Claygent a prompt and a list of accounts, and it will research and qualify them automatically.

• Primer: Takes your account list and finds the right contacts within those accounts to match against ad platforms. Particularly useful for building LinkedIn and Facebook custom audiences without relying on personal email matching.