Generative AI use in marketing workflows hit 87% this year, up from 51% just two years ago. Almost every team has adopted it. Almost none of them have agreed on what “good” looks like while adoption raced ahead, and that gap is exactly what’s starting to separate content that works from content that’s quietly working against you.
This isn’t a story about AI failing. It’s a story about platforms getting smarter about telling the difference between AI used with judgment and AI used without it, and starting to treat those two things very differently.

The Adoption Curve Nobody Predicted

That adoption curve is backed by Salesforce’s 2026 State of Marketing report. Enterprise teams are essentially fully in, at 94% adoption, and even micro-agencies under ten people have crossed 73%.

None of that is surprising on its own. What’s surprising is what’s started happening on the other side of that curve, at the point where the content actually meets an audience.

What “Obviously AI” Actually Means to a Platform

Google’s own Search Central team confirmed in a May 2026 documentation update that its long-standing spam policies now explicitly apply to generative AI content and AI Overviews inside Search, not just traditional pages. Google is on record saying it doesn’t penalize content simply because AI produced it. What it penalizes is scaled content abuse: publishing large volumes of thin, unedited pages purely to manipulate rankings, regardless of who or what typed the words.

One SEO industry study comparing publishing patterns found sites that published 50 to 100 quality AI-assisted articles with real human editing saw traffic increases of 30 to 80%. Sites that published 1,000 or more AI articles with little to no human review saw traffic drops of 40 to 90% — on the exact same underlying technology.

Same tool. Opposite outcomes. The difference was never the AI. It was whether a real person with real judgment stood between the draft and the publish button.

Why This Isn’t Just a Marketing Problem

The same pattern — adoption racing ahead of refinement — shows up everywhere, just wearing a different costume. In Corporate + B2B, AI lead-scoring adoption nearly tripled in two years, yet the performance gap between top-quartile and median teams widened right alongside it, because the tool alone was never the differentiator — how a team used it was. In Restaurant + Catering, corporate demand is growing fast enough that most operators are still running it through the same manual, referral-only process that worked when growth was slower.

Different industries, same root cause: the technology arrived. The judgment layered on top of it didn’t always arrive with it.

What Actual Craft Looks Like Now

None of this means stepping back from AI. It means treating the AI draft as exactly that — a draft, not a finished asset. In practice, that’s a real person checking every AI-assisted piece against three questions before it ships: does this sound like something a specific person on this team would actually say, does it add a real point of view a generic prompt couldn’t have produced, and would this survive a platform actively looking for signs nobody touched it.

If the honest answer to any of those is no, that’s the fix — not a full rewrite from scratch, just the human layer AI was never meant to replace in the first place.

Not sure whether your current content is reading as refined or as obviously automated? Book a quick call.