Why I'm spending less time on Google SEO and more on AI citations

Two weeks ago I shared the title and meta audit. My pages were ranking but hardly anybody was clicking. 1,570 monthly impressions on Google, 15 clicks, 1% CTR.

This week I pulled the dashboards to check on progress and something strange showed up. The Google story stayed roughly the same. The Bing Webmasters story didn't.

The Two Dashboards

Google Search Console: 2,580 impressions over the last month. 18 clicks. CTR of 0.7%. The impressions are climbing fast, up 64% in 7 days, but clicks aren't keeping up. That's the conversion problem the title and meta sprint is fixing.

Bing Webmaster Tools: 32 impressions. 3 clicks. CTR of 9.38%.

Bing's number looks tiny next to Google. Two orders of magnitude smaller. But the conversion rate is more than 10x better. Same content, same site, two completely different audiences responding very differently.

That gap is the most interesting thing in the data right now, and it needs an explanation.

What's Actually Going On

My best read of this is, Google search is still mostly a discovery engine. Someone types a keyword, gets 10 results, picks one based on a 2-second scan of titles and metas. This changes slightly with the intro of their AI overview but broadly speaking, Google still operates as a discovery engine. This is the world where title optimization matters most.

Bing search in 2026 is increasingly something different. A meaningful share of Bing's traffic comes via Microsoft Copilot, Edge's integrated AI, and other AI-grounded answer surfaces. Is a completely different paradigm as the user isn't choosing between 10 results anymore; the AI has already pre-filtered to 1–3 cited sources, and the user is checking the work. When they click through, they're already 80% convinced your page is the right answer.

Look at where Bing is sending the citations: 18 AI citations across 6 distinct pages on my site, all in the role-specific topic cluster. Data engineers (6 citations), full-stack developers (5), career independence for software engineers (3), mobile developers (2), how-to-ship-a-side-project (1), frontend engineers (1). Microsoft Copilot has decided these pages are authoritative for senior-engineer questions, and it's grounding answers in them. When someone clicks through from a Copilot answer, the click converts almost an order of magnitude better than a click from a Google SERP.

This is, in my opinion, a structural shift in how search-driven traffic behaves.

The Implications

The optimisation approach is getting re-shaped as a click from an AI-grounded answer surface is worth roughly 10 clicks from a traditional SERP.

Optimizing for AI citation has been treated as a side bet for the last 18 months. Nice if you can get it, but Google is where the real traffic is. The numbers in my own dashboard say something different. The real traffic from Google is converting at 0.7%. The smaller traffic from Bing-via-Copilot is converting at 9.38%. If you weight each impression by its conversion likelihood, the Bing-sourced traffic is doing far more work per pageview than the Google-sourced traffic. Worth flagging is that Bing has far fewer impressions than Google, so the CTR percentage carries higher variance. The pattern is still striking, but treat the 10x number as directional for this case study.

Google volume keeps growing, and I'm not abandoning it. What I am saying is that the time I was about to spend on the next round of title rewrites is probably better spent on the AI citation surface. Schema enrichment, llms.txt completeness, the .md routes that AI crawlers prefer over rendered HTML, the structured data choices that make a page citable.

There's a deeper version of this argument too. AI-grounded answer engines reward a different kind of content than Google does. Google rewards keyword match and authority signals; Copilot rewards specificity, structured data, and a particular kind of self-contained expertise. The pages getting cited on my site by AI (seen through Bing Webmasters) aren't my highest-impression pages on Google. They're the ones with the cleanest semantic structure and the most direct answers to specific senior-engineer questions.

That's a different content brief than "rank well on Google." It's worth pulling apart.

What Actually Drives Citations

Reading my own data alongside what I'm seeing on other senior-engineer sites, four things seem to drive whether a page gets cited:

1. The page answers a specific, scoped question rather than covering a broad topic. "Side project ideas for data engineers" beats "side projects" because the AI assistant grounds on the more specific match.

2. The semantic structure is clean and self-contained. Each section has a clear claim, each paragraph delivers on the claim, and the answer isn't buried under three paragraphs of throat-clearing. AI assistants extract the citable bit; if your citable bit is paragraph 4 inside a flow argument, you lose to a competitor whose citable bit is the H2.

3. Structured data is in place and accurate. Article, ItemList, FAQPage, HowTo schemas. Not all of them on every page, but the right one for each page type. The pages on my site getting cited are the ones with ItemList or HowTo schemas correctly deployed.

4. The page is reachable by AI crawlers. This is the failure mode most operators don't realize they have. Cloudflare's AI Crawl Control is on by default and blocks AI crawlers at the edge regardless of what your robots.txt says. Vercel, Netlify, and several other hosts have similar features now. Audit your edge layer before you audit your content.

These four things don't require new content. They're a structural retrofit on what you already have. That's where I'm spending the next two weeks.

The Re-Allocation

For me, the practical implication is a re-allocation of attention. Less time fine-tuning Google titles for the next 0.5% of CTR lift. More time auditing structured data, expanding the schema coverage, deploying .md routes for the highest-value pages, and making sure my site is fully reachable by every major AI crawler.

The bet is that growing my AI citations by 2-5x over the next quarter is worth more than growing Google clicks by 2x. I'll have data on whether the bet works by midsummer.

What I'm Building This Week

Continuing the title and meta audit deploy across the remaining /topics/* pages. Drafting the structured data retrofit plan for the role-specific topic cluster. Watching whether AI citation count keeps climbing at the same 2.5x-in-14-days rate it's been climbing at; if it does, that's a real pattern worth pressing on.

Reply to this email with one of:

• Your current Bing impression count and AI citation count (most operators have never looked; this is your reminder)

• A guess at why YOUR Bing CTR is what it is, if you've checked (it'll be higher than your Google CTR, almost universally)

• A page on your site you suspect could be AI-cited but isn't (happy to take a look at the structure)

P.S. The toolkit: Visibility Audit → Business Model Canvas → Building Loop → SEO Strategy Playbook → Sovereign Idea Workflow → Title & Meta Audit Workflow. All free, all designed for builders with limited time.

P.P.S. Know a senior engineer who's still optimizing only for Google? This is the issue for them. Forward this email. They can subscribe at thesovereigntechnologist.com.

That’s all for this week.

See you next Thursday.