Tactical Names

I’d spent years developing my own rules for evaluating domain names. Shorter is generally better. Real words beat invented ones. And a good name should be easy to say and spell.

But when GoDaddy’s daily list arrived with thousands of expiring domains, applying those rules consistently was another matter.

After scanning enough names, I’d start relying on instinct. And a promising name at $500 could be a terrible investment at $15,000.

I wanted something that could help me make better buying decisions, not just find interesting names.

I didn’t need a language model to evaluate every domain. Most of the initial filtering could be done faster and more cheaply with conventional scoring.

So I divided the work. A scoring engine processes the daily file in about fifteen seconds, and only the top three hundred candidates go through AI for deeper analysis.

Each receives a recommendation to ignore, flip, or build, along with an explanation and pricing guidance when the data supports it.

The experience is designed to help me spot interesting names quickly, then understand why they’re worth considering.

After two months of iteration, Tactical Names worked, but it had become increasingly complicated.

I experimented with two conversation-first versions. Neither felt right. They looked generic, and I was adding features without necessarily improving the experience.

Rather than keep refining something I wasn’t happy with, I started over.

I preserved the original code, opened a fresh database, and rebuilt the product around a simpler brief.

Two nights later, I had a much cleaner experience. Those two months of learning helped me recognize what belonged in the product and what didn’t.