Content Strategy
I started Shashi.co more than twenty years ago because friends and family kept asking the same tech questions: what laptop to buy, which antivirus to run. Writing the answer once beat repeating myself. The site has drifted toward enterprise AI and tech deals since, but last month I pulled two reports on the same set of posts. One was Blogger's on-site pageviews. The other was Google Search Console's Generative AI Features data, which shows how often a post gets pulled into an AI Overview. I expected overlap. I got two separate lists that barely touched.
The posts readers clicked chased breakups and rivalries. OpenAI cutting off Cursor's application programming interface (API) access. SK Telecom and Kakao giving away free artificial intelligence (AI) tools. A startup called Etched hitting a $21 billion valuation. Anthropic pulling ahead of OpenAI in business-to-business (B2B) spend. Every one of those posts cleared 300 views inside three days, then went flat.
The posts Google's AI Overviews kept citing were nothing like that. A Nokia fixed wireless access sale from May. A Honeywell restructuring writeup from June. A breakdown of 3GPP 6G standards timelines. A post explaining where Google moved the near-field communication (NFC) chip on the Pixel 11. None of these cracked the top ten for human readers. All of them are still generating daily citations months after publication.
Readers Want News. AI Overviews Want Answers.
A reader opens a marketing newsletter to find out what just happened and who is winning. That's an impulse click, tied to a news cycle that moves on in a day or two. An AI Overview exists to answer a specific question a person typed into a search bar: why did this acquisition happen, where is this component located, what does this technical standard mean for buyers. Nobody is scrolling LinkedIn for a 3GPP standards explainer. Plenty of enterprise IT buyers are searching for exactly that, and Google's model needs a page to pull the answer from.
The pattern showed up clearest in headline structure. My four best-performing AI Overview posts were all built around causal framing: "Why Intellect Just Bought Zaptic," "Honeywell Technologies Is Not a Rebrand," "Boomi Companion Shows Any Vendor How To," "Databricks Buys Electric to Chase Turso." Every one answers a why or a how. None of them are a press release rewrite.
Standards bodies write specifications, not summaries. Vendors write announcements, not explanations. That gap is where an analytical post earns a citation, and it stays open for as long as the underlying question stays unanswered anywhere else.
That gap is where an analytical post earns a citation, and it stays open for as long as the underlying question stays unanswered anywhere else.
A Few Posts Work Both Angles
Three posts landed on both lists. Salesforce adding Claude to Agentforce pulled 222 on-site views and 46 AI Overview impressions. Cisco's Teleport identity integration pulled 234 views and 36 impressions. Zoom Contact Center crossing $100 million in revenue pulled 47 impressions, though its on-site number never cracked the reader top ten.
Look at what those three have in common. Each one is current enough to trigger a reader's attention, and each one introduces a piece of architecture or a specific mechanism that a searcher would later type into a query. Vendor-supplied revenue figures like the Zoom number came from company reporting, not an audited filing, and I flag that in the post itself.
Two Calendars, Not One
I stopped trying to make every post do both jobs. The rivalry and valuation pieces earn their keep through reach: they build LinkedIn discussion and pull in newsletter signups. The M&A rationale and infrastructure teardowns earn their keep through compounding: a post from August 2016 is still generating citations because nobody else wrote a clean explanation of that particular technical question.
One calendar chases attention. The other chases the gap between what a vendor announced and what a buyer needs to understand. Both belong in the plan. Neither gets graded against the other's metric.
The Full List, Ranked
Two rankings, pulled straight from the reports.
On-Site Reader Traffic
| 01 | OpenAI cuts Cursor's API access | 356 views |
| 02 | South Korea assigns SK Telecom, Kakao, and KT free AI service for every resident | 344 views |
| 03 | Etched triples its valuation to $21 billion | 336 views |
| 04 | Anthropic leads OpenAI in business AI spend after Copilot hits 30 million seats | 315 views |
| 05 | YMTC files for a $4.9 billion IPO | 309 views |
| 06 | Cisco's Teleport identity integration | 234 views |
| 07 | Salesforce adds Claude to Agentforce | 222 views |
| 08 | XPeng's humanoid robot margins | 194 views |
| 09 | AI agents Instinct and Grok Bot disagree on who hits send | 178 views |
| 10 | Workday's Q2 AI spend and adoption metrics | 166 views |
AI Overview Citations
| 01 | Nokia's fixed wireless access sale and pivot | 119 impr. |
| 02 | Honeywell's restructuring | 63 impr. |
| 03 | Google Analytics AI traffic tracking | 59 impr. |
| 04 | Google moves the NFC chip on the Pixel 11 | 51 impr. |
| 05 | 3GPP 6G standards timelines | 50 impr. |
| 06 | Boomi Companion | 48 impr. |
| 07 | Zoom Contact Center crosses $100 million in revenue | 47 impr. |
| 08 | Why Intellect bought Zaptic | 46 impr. |
| 09 | Salesforce adds Claude to Agentforce | 46 impr. |
| 10 | ServiceNow Knowledge 2026 recap | 37 impr. |
What to Do Monday
| 01 | Pull your own Search Console Generative AI Features report next to your pageview report. Look for posts that appear on one list and not the other. |
| 02 | Rewrite your next M&A or product-change post around a why or how question, not a summary of the announcement. |
| 03 | Stop scoring analytical posts on day-one traffic. Check them again in three months instead. |
| 04 | Cite any number you pull from another source in MLA format, right where you use it. |
I ran the same comparison on LinkedIn, ranking a month of posts by engagement and by impressions. The two lists disagree there too.