Old is new...again

One thing I’ve noticed while advising AI companies—and as we work on the launch of Drift—is that some of the age-old marketing concepts still resonate. Surprisingly well, in fact.

In the middle of a technological arms race, it’s easy to assume all the old rules are obsolete. After all, sub-20 employee AI companies are popping up like mushrooms after a rainstorm, and getting gobbled up just as fast by enterprise acquirers. Things are moving fast, adoption curves are steep, and everyone’s betting on breakout speed. But in the rush to define the future, it’s worth remembering that some of the frameworks from the past can still help us make sense of what’s happening now.

Meet BDI and CDI: Still Standing

Case in point: BDI and CDI. Brand Development Index and Category Development Index (yes, the terms sound dusty, but stay with me). These are old-school CPG marketing terms, the kind of thing most marketers encountered in a slide deck once and haven’t thought about since. But in the AI space, they’re proving particularly relevant.

Here’s the idea: BDI tells you how well your brand is performing relative to the total market. CDI tells you how well the category is performing—how familiar, accepted, and in-demand it is. The takeaway is simple: if your CDI is low, your brand efforts will struggle unless you get really lucky or—as VCs hope—you find lightning in a bottle. Typically, you can’t grow share in a category that no one understands.

AI’s Awkward Adolescence

That’s exactly the spot many generative AI products are in right now. The technology has gone mainstream insanely fast—ChatGPT found that lightning and went from 400 million users in February to 800 million by the end of April, But, most AI categories are still undefined, or worse, misunderstood.

That means if you’re building something novel, chances are you’re not just marketing a product. You’re marketing a new kind of solution. A new way of working. Maybe even a completely new behavior… and maybe that behavior is threatening or intimidating to the user. Regardless, that puts you squarely in the territory of category creation.

Who Built the Category? Who Benefited?

Companies like Salesforce and Gong figured this out early. Salesforce didn’t just sell CRM software, it sold a whole new model: SaaS. And who benefited? Well, HubSpot’s job was a whole lot easier because Salesforce built up the CDI. Gong didn’t just sell sales analytics, it created “Revenue Intelligence” as a category, and educated the market into seeing the need for it. These companies understood that before customers could want their brand, they needed to understand the value of the category itself.

The New AI Playbook Is the Old One (Mostly)

In AI, we’re seeing the same pattern emerge. Startups like Cohere opted to differentiate by owning a category—in their case, “enterprise-grade LLMs”—rather than compete head-to-head in the generic chatbot arena. It’s a smart move: carve out a space, define it clearly, and shape the criteria for success within it. Of course, building a new category usually means investing more time, capital, and narrative muscle.

New Categories Require New Mental Models

Here’s another challenge: creating a category means doing more than renaming your product features. It means helping people build a new mental model. And that requires education, onboarding, storytelling, and patience.

Many AI tools ask users to interact in ways they’ve never done before, through prompts, chat interfaces, or natural language instructions. That’s exciting, but also unfamiliar. Research shows that first-time users of AI tools often don’t know what to ask, how to start, or what the tool can actually do. They lack a framework for success. If you don’t guide them, most will simply bounce.

A Tale of Two Onboardings

This is why onboarding in AI products matters so much, and why Snap’s rollout of “My AI” in Snapchat was such a great example of how to do it incorrectly. The feature dropped into everyone’s chat feed with no explanation or opt-in. Users found it confusing at best and invasive at worst. The backlash was immediate: app ratings tanked, and “AI” became the most common (non-profane) word in 1-star reviews.

Compare that to Microsoft’s approach with Copilot. They didn’t just add AI features quietly into Word or Excel. They gave the suite a name (Copilot)and launched a full narrative around how it could change how people work. They made the unfamiliar feel intuitive by telling stories, showing examples, and giving the category room to grow. Is it flawless? Not even close. But Microsoft recognized it was moving into uncharted waters that required more handholding and explanation than you’d get with a typical feature upgrade.

The Real Work Is Framing the Future

The lesson: new categories need scaffolding. They need clarity. And they need a story that helps users see the future, and see themselves in it.

So while the tools we’re building today are wildly advanced, some of the best thinking for how to bring them to market comes from decades-old playbooks. BDI and CDI may sound like relics, but in an AI-first world, they might be more important than ever.

Because if the category hasn’t been sold, odds are the product won’t be either.

At Drift, this lens is helping us make smarter choices about how we introduce AI into workflows users already know, and how we shape the category we’re helping to define.