Why I Stopped Betting Our Pipeline on Organic Search

For most of the past three years, I kept a running line in our quarterly planning docs: “content compounds.” SEO traffic built on itself. A post that ranked in 2022 still pulled leads in 2025. The logic held until the floor shifted, and I kept writing “content compounds” in the plan long after the data was saying something different.

The Number That Changed How I Think About Inbound

The number that got my attention was not a ranking drop or a traffic alert. It was a single figure from B2B SERP tracking data: when an AI Overview appears in the search results, organic click-through rates fall by roughly 61 percent. AI Overviews now appear on more than 80 percent of B2B commercial queries. The implication is that ranking at position one no longer means what it used to mean for most of the queries that matter.

I ran that figure past a few founders in my network earlier this year. Most of them nodded, said something like “yes, SEO is changing,” and moved on. The reaction I was watching for, the one where someone actually reorganized their pipeline assumptions around it, was rare. Including from me, for longer than I should admit.

The 61 percent CTR drop is not a forecast. It is SERP tracking data from queries already running at scale. The question is not whether AI is going to intercept your search traffic. It already has. The question is what pipeline assumptions you are still operating on that were built before the interception rate was this high.

I kept telling myself the traffic dip was temporary. A Google update, a blip in rankings, a content gap we could fill. That explanation was comfortable because it pointed toward a fix I already knew how to execute. Publish more. Improve the quality. Fix the internal linking. These are real interventions that used to work. The issue is not that they became wrong. The issue is that they are solving for a channel that is structurally different from the one they were designed for.

At some point, the honest accounting is: the distribution assumption I had built our inbound plan around was no longer the one operating in the market. That is the admission I kept delaying.

What AI Search Actually Did to B2B Commercial Traffic

AI Overviews answer the buyer’s question before they click. Research by Ahrefs on 146 million Google SERPs found AI Overviews appearing on 21 percent of all keywords and 57.9 percent of all question-style searches. For B2B commercial queries, the coverage is higher. A buyer researching their problem reads the AI-generated summary and forms a view before they ever see an organic result. Your ranking is downstream of that summary, not above it.

The structure of the buying journey makes the problem compound. Forrester and HBR research consistently shows that roughly 83 percent of the B2B buying journey completes before a buyer contacts a vendor. That was already a challenge when SEO traffic drove first contact. You were reaching buyers who had done most of their evaluation. Now, AI intercepts even that late-stage search traffic and provides the summary directly.

The distribution work I have written about before focused on building repeatable acquisition loops: product-led referrals, content-to-trial, signal-based outbound. The AI search disruption sits on top of every inbound content motion. Even a well-built content engine loses reach when the traffic channel compresses.

The Ehrenberg-Bass Institute has quantified the baseline challenge for B2B pipeline programs: at any given moment, 95 percent of your addressable market is not actively buying. A content strategy that only captures in-market buyers through search is already optimizing for 5 percent of the available revenue opportunity. With AI now intercepting a meaningful share of that 5 percent before they click through, the channel math is worse than most founders are assuming in their planning documents.

What surprises me is not the data. It is how slowly the implications diffuse into actual budget decisions. Founders I respect are still measuring content success against keyword rankings as the primary metric. Not because they believe the channel works the same way. Because the old metric is legible and the new one is not yet settled.

Why Cold Email Is Not the Substitute

When an inbound channel compresses, the instinct is to shift budget to outbound. That instinct is understandable but the math is broken there too. Cold email positive reply rates fell to 3 to 5 percent in 2025, following Google and Yahoo sender authentication requirements in 2024. Volume-only outbound models are structurally broken in the same way that volume-only SEO is: the signal-to-noise ratio in the inbox collapsed at exactly the same time AI made volume production cheap.

I watched this in detail reviewing outbound programs from a cohort of B2B teams over the past year. The teams that doubled their sequence volume in response to lower reply rates did not recover pipeline. They drove deliverability into the floor and ended up with fewer replies per month than they started with, at higher cost per meeting. Adding sequences to a broken model compounds the failure.

Signal-based outbound still works. Sequences triggered by first-party intent data: someone who visited the pricing page twice, engaged with a specific piece of content, or attended an event connected to your category. These hit conversion rates meaningfully higher than cold sequences to a purchased list. The overhead work that most AI sales tools still ignore: prospect identification, qualification, and signal interpretation. That is exactly where the effort needs to go, not into higher volume at the same targeting quality.

But signal-based outbound is a tactical fix on a structural problem. If your inbound SEO is compressed and your cold email is broken at scale, you need different distribution, not a higher-volume version of two broken channels.

The Two Distribution Motions That Survive

The two distribution motions that survive AI-mediated search are answer engine optimization and an owned audience. AEO puts your content inside the AI’s answer rather than below it. Owned audience bypasses algorithmic mediation entirely. Everything else: keyword-ranking SEO, high-volume cold email, generic content marketing. All are operating on a compressed channel.

Distribution motion	Algorithm dependence	Compounding over time	AI disruption resilience
Traditional SEO (keyword ranking)	High	Yes, until algorithm shifts	Low: AIO intercepts traffic above organic
Cold email (volume-only)	Medium (inbox placement)	No	Low: reply rates at 3 to 5 percent
Signal-based outbound	Low (intent data quality)	Moderate	Moderate
AEO (Answer Engine Optimization)	Moderate (AI citation signals)	Yes	High: your content is the answer
Owned newsletter or community	None after opt-in	Yes	High: no algorithm between you and reader

AEO is not a rename of SEO. The structural requirements are different. SEO optimizes for ranking signals: domain authority, backlinks, keyword density. AEO optimizes for citation signals: answer-first structure, FAQ schema, tables, and authoritative external citations. A post written to rank at position one and a post written to be cited in a ChatGPT or Perplexity response look different. Most content teams are producing the first kind and wondering why their AI visibility is low.

Owned audience is simpler to describe and harder to build. A newsletter subscriber opted in. There is no algorithm between your next issue and their inbox. A 40 percent open rate on a list of five thousand people is immune to the AI Overview CTR collapse. The subscriber who has read you for eighteen months responds differently than a first-visit organic click. The relationship accumulated before the search interception happened, and it persists regardless of what Google or any AI decides to do with commercial query traffic next quarter.

What AEO Actually Requires You to Change

AEO requires four specific structural changes to how content is written. Most B2B content teams are producing none of them consistently. The changes are not difficult, but they require accepting that the goal of a piece of content has shifted from “rank high on this keyword” to “be the answer an AI extracts when a buyer asks this question.”

The first change is answer-first paragraphs. Research on AI Overviews shows that AI systems extract the first one to two sentences after a heading as the authoritative response. If you open a section with context-setting, background, or framing before your actual answer, the AI skips your answer and uses a different source. The pattern is: heading, then direct answer in 40 to 60 words, then developed argument. Every H2 section in this post follows that structure deliberately.

The second change is FAQ schema. FAQPage structured data signals to search and AI systems that your content explicitly answers discrete questions. Posts with properly implemented FAQ schema produce roughly 2.7 times the citation rate in AI-generated answers. The questions need to match what buyers actually search: specific, question-phrased, answered in complete sentences with no hedging.

The third change is tables. Articles with HTML or Markdown tables receive approximately 2.5 times more AI citations than prose-only articles. AI parsing favors clearly delineated comparative information. If your topic has a taxonomy, a comparison, or a decision matrix, a table is a citation optimization, not a design preference.

The fourth change is authoritative external citations. Anthropic’s published research on how AI models evaluate content credibility, and academic work on AI training signal preferences, both suggest that AI models favor content citing primary research. A post that references a named study and links out to it signals verifiability. A post asserting claims without citation signals opinion. The authority signals matter to AI citation in ways they never fully mattered to traditional ranking algorithms.

This is where I should be honest about what I changed on this site. The FAQ section, the tables, the answer-first paragraph openings after each heading: those are the specific outputs of working through the AEO framework and applying it to long-form content. If you find this essay in a cited AI answer, that structure is the reason.

What an Owned Audience Moat Actually Looks Like

An owned audience moat is a subscriber list where your content reaches readers without algorithmic mediation. Once someone opts in, there is no search engine between you and their inbox. The data on newsletter performance has become concrete enough to treat this as a measurable distribution asset rather than a brand-building side project.

Research on top-performing newsletters found that leading publishers achieve 38 to 55 percent open rates, versus the industry average of roughly 20 percent. Publishers on one major newsletter platform collectively made over $25 million in 2025 and reached 350 million people by mid-year. Many of these are solo operators or small B2B teams, not media companies with editorial teams.

The model that worked for the top performers has a consistent pattern. They validated content on social media first, watching which short-form takes generated enough engagement to indicate genuine reader appetite. The highest-performing social content became the source material for longer newsletter issues. The list they built over eighteen months became a distribution asset no algorithm could compress.

For B2B founders thinking about how distribution actually works: a subscriber list is not just a content channel. It is a pre-qualified group of buyers and referrers who have repeatedly chosen to hear from you. A person who has been on your list for a year has seen how you think, how you handle being wrong, and how you work through complexity. That relationship converts differently from a first-time visitor who arrived through a search click and spent forty seconds on the page.

Video compounds the owned audience effect. The argument for video as the default B2B signal is that a real face on screen is the proof of human attention that AI cannot mass-produce. That argument extends naturally into an owned distribution context. The newsletter gets opened because subscribers know who you are. A video makes that relationship tangible in a way that text alone does not. I have heard from people who run their own video-first outreach at Sendspark that the warmest inbound responses come from prospects who have been newsletter subscribers for months. They already knew the point of view. The video confirmed the person behind it.

The Decision: Which Moat Are You Building?

The practical decision for a B2B founder in mid-2026 is binary. Either invest in being cited by AI, or invest in an owned audience that reaches buyers without algorithmic mediation. Doubling down on SEO content volume as a growth engine is neither option. It is optimizing for a channel that is structurally compressed.

The right specifics depend on your addressable market. If your ICP is narrow (fewer than 20,000 qualified companies in total), account-based marketing remains the primary motion. AI search disruption matters less to an ABM program because you are not relying on buyers discovering you through search. You are mapping accounts, identifying intent signals, and reaching them through targeted outreach. The handoff problem in ABM, making sure the context earned in early-stage outreach survives into a meeting, is still where most pipeline leaks, independent of how leads arrived. In those narrow-market ABM motions, video as the engagement touchpoint is the differentiator that survives AI commoditization of text outreach.

If your addressable market is broader, the combination of AEO content and a newsletter is the moat worth building. The AEO work makes you citeable when someone asks a question you should own. The newsletter accumulates readers who will be buyers or referrers when they eventually enter the buying window. Both compound. Neither depends on where any algorithm decides to send its attention next quarter.

The wrong move is to treat this as a gradual transition planned for later. “We will keep doing what works and slowly add newsletter.” If the SEO channel is already compressed, the budget going into pure keyword-ranking content is producing less pipeline per dollar now than it was twelve months ago. The decision to rebalance is not a future consideration.

One thing I would not abandon: content that earns genuine authority. The difference between content that builds authority and content existing purely for keyword ranking is whether it answers a question better than anything else available. If it does, it is good for AEO, good for traditional SEO, and worth the investment. If it does not, it is expensive noise. The check I now run on every planned piece of content at dearmer.com.au: would an AI cite this as the best available answer to the question it implies? If no, the question is whether to write it better or not write it at all.

The Closing Thought

Two years ago I would have told you our biggest distribution risk was execution: publish consistently, produce quality content, keep up with the category. I was measuring the right things for the wrong channel.

The floor shifted in 2024 and kept shifting through 2025. I kept writing “content compounds” in quarterly planning docs while the compounding thesis was becoming contingent on structural changes I had not made yet. That is the founder mistake I want to be honest about here. Not that I missed AI search when it was still hypothetical. I missed it when it was already in the data.

The distribution that survives is the kind that either owns the AI’s answer or does not depend on it at all. Both require a different content investment than keyword-density SEO. Both are worth making now.

When is the week you stop checking your organic keyword ranking as the primary success metric and start checking how often your content gets cited in an AI answer, or what your newsletter open rate is doing over a twelve-month subscriber cohort?