One good number beats ten press releases

My first job out of Northwestern was at Forrester Research in Kendall Square. I lasted five months. It taught me more about getting covered than the five years after it did.

At Forrester we produced numbers. And the thing I noticed, before I understood why it mattered, was that nobody argued with the numbers. They argued about what the numbers meant. That’s a completely different, much better conversation to be at the center of.

Then I became a reporter and spent twenty years on the other side of it. The pitches that got on air were the ones with a figure I couldn’t get anywhere else. Not the best-written ones. The ones I needed.

The only peer-reviewed research in this field says the same thing. Researchers at Princeton and Georgia Tech built a benchmark and tested tactics against generative engines: adding statistics, citations and quotations lifted visibility 30 to 40%, and as high as 115% for content sitting in fifth position. Keyword stuffing, padding and persuasive language measured at zero or negative. Muck Rack found the signature in the wild too. Press releases cited by AI carried twice as many statistics and two and a half times as many bullet points as average.

This is the whole reason we are a collective and not an agency. Original research needs someone who can design the survey, someone who can find the story buried in the results, and someone who can produce it in a form an editor will actually run. Most shops have one of those three people and outsource the other two. We keep all of them under one roof, which is why we can say “let’s go count something nobody has counted” and then actually go do it.

Run the survey. Publish the benchmark. Then watch competitors cite you while arguing with you. You give away one number and you own the sentence around it.