GEO is getting plenty of attention. Proof that it works is harder to find.

For marketing leaders, that raises some reasonable questions. How do you measure AI visibility? How do you know optimization is actually driving the results? And with ChatGPT, Gemini, Copilot, Perplexity, and Google all working differently, what does success even look like?

SOLIDWORKS wanted answers.

As a leader in 3D design, SOLIDWORKS was already seeing buyers turn to AI platforms for information. The question was whether deliberate optimization could increase the likelihood that SOLIDWORKS content would appear in those answers.

BOL developed a program designed to find out. We optimized a defined group of pages, compared their performance against the rest of the site, and measured the results across multiple AI platforms and independent data sources.

The results make a strong case for GEO.

SOLIDWORKS was Losing Ground in LLM Answers

SOLIDWORKS was losing visibility in AI answers. Across the site, Microsoft Copilot and Bing citations had fallen 19%. As more buyers turned to AI for answers, SOLIDWORKS needed to reverse that decline and find a way to measure whether GEO was actually working.

Attribution is murky. When sitewide numbers move, it could be an algorithm update, seasonality, or the brand itself. Crediting optimization is usually guesswork.

One engine isn't the market. Google, Microsoft, ChatGPT, Gemini, and Perplexity each choose sources differently. A win on one proves little about the rest.

Brand strength isn't a strategy. Established brands assume AI will cite them by default. The data said otherwise: across the web, AI assistants were citing solidworks.com less than they had a year earlier.

BOL needed to do two things: reverse the decline and prove that the optimization was driving the results.

Design the GEO Program to Survive Scrutiny

BOL's approach wasn't about gaming AI citations. It was about systematically making SOLIDWORKS a better answer to the questions its market was asking.

1. Optimize around what buyers and AI engines are looking for

BOL developed a GEO strategy for each page, identifying the search terms, questions, and related topics buyers are exploring across traditional search and AI platforms. From there, we identified gaps in the existing content and built an optimization plan designed to make each page more useful, authoritative, and easier for AI engines to cite.

• Search and LLM query analysis. Identify the keywords, prompts, related questions, and comparisons the page needs to address.
• Content gap analysis. Determine what information is missing or underdeveloped based on search behavior, AI prompts, and competing content.
• Content architecture. Add or restructure sections so important questions are answered clearly and comprehensively.
• Direct answers and FAQs. Give search engines and AI platforms concise, useful answers to the questions buyers are asking.
• Authority signals. Strengthen pages with internal links, supporting proof, relevant media, and connections to deeper SOLIDWORKS content.
• Technical optimization. Make sure pages are structured and accessible so search and AI systems can find, understand, and cite the content.

2. Compare optimized and non-optimized pages

Optimizing 36 pages also gave BOL a way to compare their performance against pages that had not received GEO optimization. Because both groups were on the same domain and measured during the same period, the comparison provided a clearer view of the potential impact of the work.

3. Measure across multiple AI platforms

Results were tracked across Google AI Overviews and AI Mode, Microsoft Copilot and Bing, ChatGPT, Gemini, and Perplexity. Data came from Google Search Console, seoClarity, Conductor, and Bing Webmaster Tools.

4. Measure performance across the entire optimized set

Rather than relying on a handful of high-performing pages, BOL tracked the percentage of optimized pages that appeared in AI results. This showed whether gains were occurring broadly across the program.

5. Track progress monthly

AI visibility became part of SOLIDWORKS' ongoing SEO and GEO reporting, allowing the team to track performance over time and continually prioritize new opportunities.

Skepticism-Busting Results

Across every AI source measured, BOL-optimized pages outperformed the rest of the site.

• Nearly 2x more likely to be used by AI. Averaged across all four sources, 70% of BOL-optimized pages were used by AI assistants, versus 40% of the rest of the site.
• 7.4x more exposure in Google AI answers. Each optimized page appeared inside an AI Overview or AI Mode an average of 33,147 times, versus 4,461 for non-optimized pages (18 May – 4 Sep 2026).
• 100% cited by Microsoft Copilot and Bing. Every one of the 36 optimized pages was cited at least once.

Growth where the rest of the site shrank

On Microsoft Copilot and Bing, BOL-optimized pages grew their average citations per page by 13%. Non-optimized pages dropped 30% over the same period, a 43-point gap on the same domain. Sitewide, Microsoft citations fell 19%.

Share of pages cited as a source in an AI answer

Sources: Google Search Console, seoClarity, Conductor, Bing Webmaster Tools. Google and Microsoft figures use different measurement windows, labeled above.

Why This Program Worked

It answered the attribution question. A same-site control group meant the brand, the domain, and the timing were held constant. The difference between the two groups was the optimization.

It didn't rely on one engine. Optimized pages won on Google, on Microsoft, and on ChatGPT, Gemini, and Perplexity. A result that repeats across independent systems is hard to dismiss as a fluke.

It measured breadth, not a lucky page. Share-of-pages metrics showed the lift was spread across the optimized set, not carried by one standout.

It exposed the cost of doing nothing. Non-optimized pages lost 30% of their Microsoft citations. That turned GEO from a nice-to-have into a risk leaders could see.

It focused where the business needed it. Prioritizing AI and AI-CAD topics put the effort behind the conversations SOLIDWORKS most needed to win.

The Bottom Line

GEO doesn't need more believers. It needs evidence. SOLIDWORKS started with the same questions every skeptical CMO asks, and BOL answered them with data from the brand's own site.

With this program in place, SOLIDWORKS now has:

• 36 optimized pages outperforming the rest of the site on every AI source measured
• A same-site control group that isolates the impact of optimization
• A cross-engine measurement framework spanning Google, Microsoft, ChatGPT, Gemini, and Perplexity
• A monthly prioritization cadence ready to extend to the remaining 90 core pages
• Proof that stands up in a boardroom

If your leadership team is skeptical about GEO, good. That's the right place to start. Let's show them the data. Let's build it together.