At a recent exhibition in Chicago, artist Charwei Tsai displayed a thousand pieces of glass with a single diamond hidden among them. This prompts the question: How can you find a diamond in a sea of glass, and would you know it if you did?
With AI, it is now possible to understand exactly where one company sits among an entire industry. We can find new prospects that exactly match our best clients, and do so at scale. We can even test if our words are differentiated as we write them.
Drawing on techniques developed through collaboration between marketers and data scientists, two breakthrough use cases highlight how these tools are reshaping both positioning and prospecting.
Data-Driven Positioning
Every company needs to understand how to set itself apart, and to position effectively, it helps to identify what already exists in the market.
AI makes it possible to compare hundreds of firms quickly based on how they present themselves online. For instance, analyzing website copy can reveal common messaging patterns and potential whitespace.
Transformer-based language models, the backbone of many generative AI tools, are particularly adept at finding meaning in text regardless of exact wording. These models embed sample copy in a multi-dimensional space, allowing us to visualize how similar or distinct each company is from its peers.
In an analysis of the websites of 700 early-stage venture capital funds, for example, the first finding was that most VCs aren’t very differentiated at all. Their sites follow familiar structures—explaining how they invest, what sectors they target, and how they support portfolio companies. The words often seem repetitive and empty.
But a transformer model helped identify less common themes, such as how a venture firm might focus on being transparent in giving feedback or on the importance of conviction-based investing. This model can uncover the unique strategies used by different types of funds within the dataset, such as diverse-led funds.
By connecting these patterns with outcome data—like startup exits or total capital raised—it’s possible to see how the most successful VC’s position themselves.
This method can also guide copywriting, for example when writing a new website for a VC. Feeding potential phrases or taglines into the model during the writing process provides a quick check to see if they stand out against the backdrop of the broader competitive set. The iterative process leads to editing with clarity and precision.
While the example here focuses on venture capital firms, the same approach applies across industries, and offers a scalable alternative to traditional competitive analysis.
The Twin-Client Project
Predicting what messages will resonate is only one of the challenges that we face as marketers. We are also tasked with identifying new prospects that are likely to convert.
The same transformer embedding models can “match” potential clients with existing ones. By analyzing how top clients describe themselves, it’s possible to build a profile that serves as a blueprint for finding similar organizations. This concept is sometimes called finding “client twins,” and it allows marketers to focus their searches rather than casting a wide net and hoping for the best.
In other words, they allow us to judge the quality of prospects before we ever reach out to them.
For example, a digital asset trading firm that targets institutional investors faced the challenge of navigating a large, opaque market. With over 20,000 potential prospects globally—and the relatively limited transparency of the crypto world itself—manual outreach would be ineffective without some way to prioritize these companies.
Using transformer models and web-scraped data, one analysis surfaced less obvious prospects such as international shipping firms, payments startups, and family offices. These organizations shared traits with the firm’s best clients, giving clues to their needs, what they might buy, and even how best to sell to them.
AI-powered models offer new ways to answer longstanding marketing questions. The ability to understand where one company stands within its universe of competitors and prospects is a powerful tool. When done right, it can reveal the diamond in a sea of glass.
Brian Dema is the Founder and CEO of the Department of Growth, which manages marketing and sales for startups and venture capital firms, and builds complete teams using fractional experts.
Han de Jong, PhD is a freelance data scientist and a Staff Scientist in the Neuroscience Department at the University of California, Berkeley, where he maps the brain to study addiction.
Illustration by Khytul Abyad
