If your AI strategy still treats the model as a co-pilot, you're missing the bigger shift reshaping visibility, intent data, and pipeline math.

This week's edition covers:

  • Kevin's Take: why treating AI as an independent reviewer, not a co-pilot, is the shift most CMOs are sleeping on
  • The Signal: Signal: three AI visibility metrics that actually mean something, plus what to do when two-thirds of searches end without a click
  • Tools & Tactics: G2's intent data just got a lot bigger, and how to rebuild keyword research for answer engines
  • Quick Links: 90% of your AI citations live off your site, vertical AI breaks the PLG playbook, publishers lock crawlers out of your earned media

Kevin's Take

AI as an independent reviewer, not a co-pilot, is a game changer

I read this Ethan Mollick piece on co-existence and the end of co-intelligence and I really enjoyed his experiment where he stopped using AI as a collaborator on his book concept and instead pitched the idea to AI as an independent reviewer, like a skeptical editor or a peer who didn't owe him anything. That's a tiny reframe but it's actually a huge one, because most everyone using AI in marketing right now is using it as a co-pilot, a helper, an assistant that's been steered into agreeing with them. We prompt, it produces, we tweak, it agrees, we ship. There's no real disagreement in that loop, and I think that's why so much AI-generated marketing work feels hollow.

What I've started doing is setting up AI as an adversary or an outside critic rather than a teammate. When I'm working inside Claude Code, I'll spin up a separate project with its own system prompt that doesn't know my context, doesn't know I'm the one asking, and is told to behave like a skeptical CMO at a competing company, or a customer who's already burned out on the category. I paste in the work and ask for the holes. The output is wildly different from what you get when you ask your friendly co-pilot to review something. The friendly co-pilot finds nits. The independent reviewer finds the actual reason the idea won't land. Mollick's framing of co-existence makes me think this is where the real shift is happening, not in efficiency gains but in giving AI enough autonomy and distance from your own thinking that it can actually push back on you.

If you figure this out, you're going to make better decisions faster, because you'll have something close to a second opinion on tap for every brief, every deck, every launch plan. If you're still stuck in prompt-and-praise mode, you're going to keep creating work that your own AI told you was great, and then wonder why nothing's working in market. The real game changer for me has been to stop asking AI to help me and start asking it to disagree with me.

That's it for this week. Talk soon.

— Kevin Kerner, CEO, Mighty & True

The Signal

Three AI visibility metrics that beat share of voice (1 min read)

AI share-of-voice metrics from visibility platforms extrapolate from tiny prompt samples and mislead marketers. The piece proposes three alternative metrics built for an infinite-query LLM environment.

Why it matters: If you're paying for an AI visibility tool, you need to know whether the dashboard you're showing your CEO is actually measuring anything real.

Two-thirds of Google searches now end without a click (1 min read)

SparkToro/Datos data shows 68% of Google searches ended without a click in early 2026, up sharply as AI Overviews keep users on-platform. AI Mode is expected to compress organic CTRs further. (via Danny Goodwin)

Why it matters: If your demand gen still leans on organic search traffic, your pipeline math is breaking — you need to rethink how your brand shows up inside the answer, not after it.

Tools & Tactics

G2 unifies buyer intent across four review sites (1 min read)

G2 unified buyer intent data across four software discovery platforms (G2, TrustRadius, Software Advice, GetApp) into a single feed. The expanded signal set covers more of the in-market research journey for B2B software buyers.

Why it matters: If you rely on G2 intent to prioritize ABM and pipeline, your signal volume and account coverage just expanded — check how this changes your scoring model.

Rebuilding keyword research for answer engines (1 min read)

HubSpot guide on adapting keyword research methodology for answer engines, focusing on nuanced, personalized AI search queries. Covers how to identify and target queries that LLMs surface in 2026.

Why it matters: If your SEO playbook still starts with a keyword volume spreadsheet, you're optimizing for a search behavior your buyers have already moved past.

Strategy

  • How to shape what AI says about your brand — Jason Barnard breaks down how LLMs form brand perceptions from your digital footprint and what to do to influence recognition, credibility, and recommendation (via Jason Barnard)

Read Inference — long-form essays

Inference is our long-form essay series on the future of B2B marketing. Where Drag and Drop is the weekly signal, Inference is the deeper analysis. Subscribe to get the new pieces in your inbox.

▶ Read Inference

Know a great guest?

Know a marketing leader with a strong opinion and real results to back it up? We are booking guests for the Tech Marketing Rewired podcast. If that is you (or someone you know), hit reply. Bonus points if you have a hot take that will make people uncomfortable.

▶ Pitch yourself / a guest →