How To Unlock Smarter Target Account Lists Using AI

If you work in account-based marketing, your default process for building an ICP account list probably looks a little something like this: open your ABM platform of choice, punch in your firmographic filters, and out comes your segment. 

But there are only so many common filter criteria you can use to build a list. What happens if you run out?

The data problem for niche ICPs

Platforms like 6sense and Demandbase aren’t just powerful, they’re invaluable for modern ABM. When you’re running a robust, full-funnel, omnichannel ABM program and need a mega-brain to power the whole thing, they’ve got you covered and then some. 

Using these platforms, you can easily leverage traditional firmographics to build your TALs, such as a company’s headquarters’ country, revenue range, employee count, industry, NAICS code, etc. This is table stakes.

But what if you want to segment by something more specific, something that doesn’t come with a pre-existing filter?

Say you’re going after small, mom-and-pop CPA and tax preparation firms. By nature, these are tiny accounts, and even with the best data vendors, the available firmographic data is sparse and riddled with inaccuracies. Even if you filter by a NAICS code specifically designated to tax preparation, you’ll find that applying it simultaneously includes loads of non-tax prep firms as well as cuts out plenty of valid tax prep firms.

On a small scale, the solution to this particular problem could be to visit each company’s website (if it has one) and check its services pages to see whether it actually offers tax prep. If your list is 30,000 accounts, though, that’s when it’s time to ask AI to do that job for you across the entire list.

And it will. Just like that.

Or imagine that you sell a SaaS product built for e-commerce sites, specifically companies with massive SKU counts, like auto parts stores. There’s no filter for that—and really, why would there be? But it’s easy to have an LLM go through each of those accounts, estimate the number of SKUs on their sites, sort them, set a threshold, and filter them for you.

Both of the above cases are real-world examples where AI helped us unlock infinitely better targeting against accounts that actually matter.

Getting creative with signals

Which brings me to my main point: AI is an opportunity for marketers to get creative about signals. Whether it’s the approximate number of SKUs on a site, the mention of a specific service, or recent earnings reports mentioning something you care about—say, you’re trying to go after companies that are heavily investing in AI. You could easily ask an LLM to scan earnings reports at scale and establish an “AI-ness score” for each of your accounts.

In other words, AI lets you segment by “weird” stuff that nobody would ever have as pre-packaged data, but which can drive incredible performance for you. If you can imagine it, you can probably capture it using AI-assisted tools.

Clay is one of the tools available that’s really good for this sort of work. But it’s just an orchestration layer, kind of like a big spreadsheet where you can run prompts against every row using different models. Other options are available, and even 6sense shared at a recent event that they acknowledge this as a current product gap and are working to build similar AI-assisted data scraping and enrichment functionality within their platform.

In an effort to solve this in a scrappy way, one weekend before I had the right Clay subscription tier, I even vibe-coded my own homebrew solution using a Postgres database, Python workers, and the OpenAI API.

The ultimate takeaway is that this is not a sales pitch for a particular tool, but rather a plea to get creative with segmentation using any possible tool in the toolbox to dig deeper and get more granular with account list targeting. We’re talking about a mindset rather than a tool dependency.

How to build the case internally

If you’re trying to convince someone internally that this approach is worth investing in (and it is), I’d start by thinking about the junk that consistently shows up in your MQLs or lead flow. Most teams can easily pinpoint a common bad-fit attribute that keeps reappearing. I’d argue with a high degree of confidence that you can likely use these tools to mitigate it.

While you’re at it, start running sanity checks on your lists with basic prompts like: Is this site operational? Is this company still in business? In a time when marketers are constantly internalizing a “do more with less” philosophy, this is a concrete way to trim the fat before you start spending.

Right now, teams that are willing to layer AI-powered qualification on top of their existing stack have a real competitive edge. It’s a massive opportunity for differentiation, and it may not be open forever, because the big ABM platforms are already catching up.