private conversations with demmannd

The Thesis: Return on Intelligence (RO/I)

The metric redefining enterprise growth in the AI era.

There’s an absurd amount of money being spent to generate demand signal. Global ad spend crossed $1 trillion this year. U.S. digital ad revenue alone hit $294.6 billion, up nearly 14% (IAB/PwC). Add the $148 billion companies pour into martech - headed toward $215 billion by 2027 (Forrester) - and you start to see the scale of the problem.

Now ask what all that money is actually buying.

A few years ago, I ran a 90-day test at a multinational to measure the value of the signal we generated. Three campaigns drove half a million visitors and produced 3,800 lead forms. When I asked what we learned from the other 496,200 signals, the room went silent. 99.24% of the signal we paid for told us nothing.

This isn’t an outlier. It’s the norm.

Paid search is the biggest line item in most marketing budgets, yet accounts for under half a percent of total site visits. The other ~92% comes from channels nobody is directly paying for: direct, organic search, referral, organic social (Semrush, 2025). And increasingly, most of the buying journey never touches you at all - B2B buyers spend as little as 17% of their buying time with vendors. The rest happens somewhere you can’t see.

So volume was never the constraint. The constraint is what you do with what shows up.

Here’s the part that should stop you: none of that invisible signal ever touches your ROI number. Traditional ROI isn’t wrong, it’s just blind. It only counts dollars spent against outcomes you can attribute. Anything that doesn’t resolve into a tracked conversion doesn’t get scored poorly. It doesn’t get scored at all. It simply disappears.

"Volume was never the constraint. The constraint is what you do with what shows up."

The 92% of organic traffic you didn’t pay for. The 496,200 signals nobody could name. The 80%+ of the buying journey that happens off site. None of it shows up as cost, return, or loss. It’s just gone.

The reframe that simplifies the entire problem: If I only had one dollar to generate signal, what’s the smallest set of things I’d actually need to know?

With AI, that question matters more than ever. AI is exploding signal volume for everyone at once. When everyone has access to the same flood of information, the differentiator stops being who collects the most. It becomes who extracts the most value from the same data everyone else is drowning in.

Today, most of the signal your site generates gets thrown away. Because it’s invisible, it never shows up in ROI. ROI ends up measuring only a fraction of what actually happened.

In the AI era, the real metric is Return on Intelligence (RO/I): How much of what you already generate turns into something you can act on. Not how much signal you collect, but how much you can actually use.

That’s the problem DEMMANND is built to solve. Not a better report on the signal you already capture, but a system where your site and the intelligence reasoning over it are the same object. Nothing waits to be seen. The system reasons continuously over signal whether or not it resolves into a form fill. Every visit compounds instead of starting over.

DEMMANND measures what legacy systems can’t: intelligence itself

We’re not entering a category. We’re defining it.