Traffic was a stand-in for value.

Picture a man on a street corner with a clipboard, counting cars every fifteen minutes. His best count of the month comes at twenty to six on a Thursday. Nobody on that block has moved in eleven minutes. A cyclist slides by in the bike lane, gets home on time, and never makes the clipboard.

A lot of dashboards work like that clipboard. They count how many cars are on the road. Most days that's a fine number. On the worst day it's a lying one, because a jam is when the road is fullest and moves the fewest people.

Traffic was a proximate metric for value. Visits were easy to count, and for a long time more visits really did mean more people getting something from you. So we counted visits. Then the link between the count and the value started to slip, and a lot of slides didn't notice.

Advertising already went through this

I've watched this happen once already. When I was interim CMO at PenFed, I got to see the advertising side up close. Ads started out sold on how many people saw them: impressions. Over time the industry moved toward people taking action. As a marketer I wanted attribution based on what people actually did, clicks and applications and accounts, not on how many eyeballs went past.

Impressions stayed. They became one number among several, worth something, but not worth what a conversion is worth.

Visits are going to go the same way. A visit will become like an impression: valuable, but not the same value as someone doing the thing you exist to help them do. We'll get better at tracking both.

At PenFed I was VP, Digital & Innovation, leading about 120 people across digital, marketing, design, analytics and supporting engineering. The numbers I'd put on a slide from those years are application completion above 60 percent and product acquisition up 12 percent year over year. Neither one is a visit. A started application is a car on the on-ramp. A finished one is a trip.

Bikes didn't disappear

Think about how people get around. We walked. Then we got wheels, carts, animals to pull them, then bicycles, then trains and cars. Every time, there's a sense that the new way will make the old ways pointless. It never does. I still have to teach my kids how to ride a bike.

The web is in one of those moments. A big share of the visitors to any site now aren't people. According to Imperva's 2025 Bad Bot Report, automated traffic made up 51% of all web traffic in 2024, the first time in a decade that bots outnumbered humans, and bad bots alone were 37%. And more and more of the automated kind are agents reading on a person's behalf: someone wants a specific piece of information, and instead of clicking through your site, they send an AI to go get it.

Human browsing gets squeezed into fewer visits, but it stays. A new mode of travel changes the mix: who takes which mode, for which trip, and what each trip is worth.

The web is becoming one big library

Here's the picture I find most useful: a librarian who doesn't hand you a stack of books. She reads the right pages across a hundred of them and hands you one page with the answer. AI agents are starting to treat the whole web that way. They go out to a lot of sites at once, pull what they need, and put it together for whoever asked.

This isn't brand new. At Wiser, we went out to millions of pages a day to understand what certain products were selling for, and used all of it together as the raw material for what we delivered to customers. Years before that, when I was one of five people on Gannett's Design Innovation Group, we worked on a project that built news websites on demand, assembled from content available across hundreds of Gannett's news sites. Agents are a much bigger, much faster version of both.

So the question for a marketer changes. It used to be "how do I get them to my site?" Now it's also "when my content gets pulled into someone else's answer, does it still sound like me, and do I get credit?" How you write and structure your content decides whether you're recognized when it's used somewhere you'll never see.

Fewer visits, more intent

Five years ago the conversation was all about blocking bots. Keep them off the site. That's shifting to something more useful: you want some, not all. An agent that finds your content and puts it in front of a person who needs it is discovery. Block it and you've closed a door your customers might be walking through.

And the people who still show up are worth more. If an agent did the browsing, the person who comes to your site anyway came on purpose. They want to learn about you, interact with you, buy from you. For marketers who are aware instead of scared, that's a real opportunity. Less traffic that was never going to do anything, and a sharper read on the traffic that will.

Run your marketing in 2030 the way you ran it in 2020 and you'll find out the hard way which of your numbers were real and which were just convenient.

Write down what the number stands in for

One sentence, plain words: we count X because we believe it means Y. Put it at the top of the page where the number lives. Most teams never have. When they try, they often find different people believed in different Ys all along.

Then find a measure closer to the Y. Trips completed instead of cars on the road. For a publisher, returning readers or paid subscriptions. For a bank, funded accounts. These numbers are smaller and they show up later. That's the price of being closer to the truth.

Don't throw the old counter away. Take it off the goals page and keep it where the team looks when something seems off. Traffic still tells you whether anyone is showing up and where they're getting stuck. A sudden drop means go look. So does a sudden spike, because sometimes a spike is a jam.