Skip to content
Walk WestNovember 20, 20263 min read

Stop Reporting Traffic: One Year After Our CEO’s Keynote

Stop reporting traffic to your boss. A year ago, our CEO, Greg Boone, stood in front of a room of communications professionals and argued that AI alone won’t save a company. The year since has made a narrower point for us: the click is no longer the scoreboard.

We said then that AI would answer the question before anyone saw your site. We said the work that holds up is the work other people vouch for. Here is how that looks a year later, and what we would put on the scoreboard instead.

Clicks kept disappearing

About two in three US Google searches now end without a click. SparkToro’s study, as reported by Search Engine Land, covered January through April 2026 and put the figure at 68%, up from about 60% in 2024. SparkToro cautions that its panels differ between years, so treat the gap as a direction, not an exact measure.

When an AI Overview sat on the results page, SparkToro estimated clicks fell by nearly 60%. That is the shift we described a year ago, now measured. Someone asks a question, reads the answer, and leaves. Your page can supply the answer and still get zero visits.

HubSpot changed its scoreboard

HubSpot taught a generation of marketers to get found on Google. Its CMO now says the company’s blog lost about 80% of its organic traffic as AI answers spread, and he credits channels built outside search for what held up: a newsletter, a podcast network, a YouTube network. By his account, about 90% of HubSpot’s content leads come from channels that barely existed in its mix five years ago. The blog supplies about 10%. He also says revenue grew nearly 20% in 2025. Search Engine Journal ran the full interview.

Those are HubSpot’s own figures, and outside analysts dispute the exact size and cause of the traffic drop. We hold the numbers loosely. The direction is the point. HubSpot stopped treating traffic as the scoreboard. It now tracks revenue per visit, how often AI answers name it, and pipeline that content influenced.

Pick a scoreboard you can defend

Traffic is the number you can no longer defend in the CEO’s office. Everyone’s graph bends the same way, so a falling line says nothing about whether your work is working.

We count attention instead. We call these attention-qualified leads, or AQLs: signs that a person paid attention, not just landed on a page.

  • Completions: a video watched to the end, a podcast episode finished.
  • Saves and shares: a post your customer screenshots and sends to a colleague.
  • Citations: an AI answer names you or links to you. Count the names and the links separately.

Then add the number that closes the loop: booked meetings. A real person agreed to talk to you.

None of this is clean attribution. HubSpot’s CMO calls its AI-citation signal correlative and still treats it as a core investment. A rough signal you trust beats a precise one that stopped meaning anything.

Try this in the next 90 days

Record a three-episode podcast arc around one real tension your customers face. Pick a question they argue about, not a topic you want to promote. Invite a different guest to each episode and ask each one to share it with their own audience.

That one project touches most of the scoreboard. People can finish an episode. Guests can share it. And other people talking about you gives AI answers something to cite besides your own site. Run it for 90 days, then check completions, shares, citations, and booked meetings against your traffic report.