Behind the scenes of a content machine without AI slop, built on my own July telemetry

"Organic traffic is falling off a cliff, paid ads are burning budget and the board has imposed a hiring freeze. I have no way to generate B2B meetings."

A 34% year-on-year drop in B2B organic traffic is a structural change affecting an entire category of queries. Our July telemetry showed that fewer visits can deliver more pipeline - as long as you separate visibility from demand and conversion.

The data rewriting the rules of search

Seer Interactive shows that for queries triggering Google AI Overviews, organic CTR falls from 1.76% to 0.61% and paid CTR from 19.7% to 6.34%. Ahrefs, on a sample of 300,000 keywords, confirms a 34.5-58% CTR drop in position one. Search Engine Journal adds that the presence of AI Overviews cuts clicks by 38% while the zero-click share rises from 54% to 72%.

Gartner estimates a 25% fall in traditional search volume by the end of 2026. Companies heavily dependent on informational content saw sessions drop by as much as 64%. Pew Research shows that with an AI Overview present, users click a traditional result only 8% of the time, and 83% of such searches end with no click at all.

AI Overviews eat mainly informational, non-branded queries, while branded phrases stay relatively protected. The conclusion for a CMO: the decline is about query categories, and the team was working under rules that just stopped applying.

Three assumptions that stopped working

Positions 1-3 delivered most of the traffic. Informational traffic converted through retargeting. More content meant more pipeline. Each of those levers now behaves differently: AI Overviews take the simple answers, retargeting loses contact with users who never landed on the site, and content volume without telemetry generates pure cost.

Rising CPC means rescuing pipeline with a bigger paid search budget burns money fast, because paid CTR also drops around AI Overviews.

Three-layer telemetry as the foundation

Visibility

Impressions, average positions and brand presence in citations inside generative answers. In this layer I also check entity consistency and schema. The diagnostic rule: stable impressions plus falling CTR equals a domain that is indexed but not clicked. Optimisation then goes towards citations, and clicks become a side effect.

Demand

Branded query volume, direct traffic and dark social signals. Growing branded search signals that the brand appears in ChatGPT, Perplexity or Copilot answers, and users return later directly, bypassing the SERP. Periodic tests in AI engines reveal which category questions get answered and which brands get cited.

Conversion

Leads, demos and opportunities from destination pages, deal value and cycle length. This layer settles everything: AI-search traffic converts 18-23 times better than traditional organic, and user value is three times higher.

The July numbers that changed my content machine

14,138 impressions (+57%), 218 reactions (+145%), ER 1.54%. The personal series averaged 925 impressions and 25.3 reactions per post; the pure education series 443 impressions and 4.1 reactions. Follow conversion fell from 2.1‰ to 1.2‰ despite +57% reach. Sunday turned out to be the best slot, averaging 1,014 impressions.

In July I dropped the series averaging 443 impressions per post - the best decision of the month. The rule that came out of it: technique without a personal stake sinks. Education always inside a story or a number from operations.

July's paradox is that the month won on volume and quality but lost follow conversion. The cause sits in the mix: too much lecture, too little human, too weak a "follow, because…" close.

What I discarded and why

• Definition articles, because AI Overviews answer them without a click.
• The pure education series, because 6 posts produced 443 impressions and 4.1 reactions.
• Volume without telemetry, because it generates cost without pipeline.

I kept destination pages, operator stories, TCO breakdowns and 20-query tests in AI engines. Those connect visibility with demand and conversion. The working template: price, a real-life analogy, a hard number with a date, an honest counterpoint, and a question about the reader's own number.

How I measure effectiveness today

• Share of citations in AI engines.
• Branded queries.
• Direct traffic.
• Leads from destination pages.
• Deal value and cycle length.

The language of the board report

• Quantity down, quality up.
• AI-search conversion 18-23 times better.
• AI-search user value 3 times higher.
• Higher deal value, shorter cycle, higher win rate.

Fewer visits, more pipeline - if you have three layers of telemetry. The board conversation then stops being a budget defence and becomes a presentation of a new growth architecture.