Most of what people have written over the last year about how AI disrupts SEO focuses on the wrong thing.
People want to know whether Google's AI Overviews are stealing their clicks, whether ChatGPT is replacing traditional search, and how their content will survive the zero-click environment. These are all real concerns, but they're symptoms of a more fundamental shift.
At Schwartz Marketing Lab (S.LAB), we tracked 118 B2B SaaS companies over an 18-month window to understand who has been gaining traffic and who has been losing it since AI search entered the fold. The results were almost evenly split: 53% of companies declined and 47% grew.
The split wasn't explained by execution quality, publishing frequency, or domain authority. Those variables didn't consistently separate the winners from the losers. What did explain it was something most content programs don't measure.
The property that separated them
The companies that lost traffic had something in common: a large portion of their content was built around queries that AI systems can now resolve without sending anyone to a website. This included templates, job description lists, basic how-to guides for well-documented processes, step-by-step tutorials for mature tools. Despite being evergreen and carefully produced content assets, AI could satisfy the answer without a click. The user didn't need their website anymore.
The companies that grew had something different. As we worked through the cohort data, we began calling it Durable Demand.
Durable Demand prevails when two properties exist together:
The audience renews when new people keep arriving at the same problem for the first time. Every year, a new cohort of engineers learns DevOps fundamentals. Every year, new teams go through their first SOC 2 audit. Every year, new marketers learn attribution modeling from scratch. The concept doesn't change; the people who need to learn it do. That's an audience that replenishes itself.
Engagement is required when the content can't be reduced to a paragraph summary. A framework for thinking through a problem, a methodology that requires reading and applying rather than skimming and extracting, a conceptual explanation that only makes sense once you've worked through the reasoning — these can be summarized by AI, but not substituted by it.
The clearest example: Harness holds the #1 ranking for "what is a devops pipeline" even when Google displays an AI Overview directly on that SERP. The AI tried to answer the question in the results. Users still clicked through, because understanding a DevOps pipeline requires reading an actual explanation. That's Durable Demand working exactly as the data predicted.
A simple test before you invest
Ask yourself three questions before you invest in a piece of content or continue to invest in an existing one:
Does the audience renew itself? Will new people keep encountering this problem year after year, or do most people solve it once and move on?
Does it require engagement? Can someone extract the full value from a summary, or do they need to read, think, and apply?
Can AI satisfy it without you? Search the primary keyword in an incognito window. If an AI Overview appears, ask honestly: could the reader accomplish their goal from that overview alone?
Content that passes all three questions has structural resilience that doesn't depend on where Google's algorithm goes next, or how capable AI systems become at answering general questions. Content that fails one or more is increasingly fragile regardless of how well it's executed.
What this changes
The old content strategy question was: is this topic relevant?
The new question is: does this topic have Durable Demand?
Relevance tells you whether people will keep searching for something. Durable Demand tells you whether they'll still need your content after AI generates an answer. Those are different questions, and the gap between them is where most traffic losses are actually happening.
If you want to go deeper — the full mechanism breakdown, portfolio health model, and how the 118-company research shapes into a practical system — that's exactly what the Durable Demand Framework was built to explain.
Top comments (0)