§01the finding
We took 200 B2B SaaS companies across 20 categories and split them into two groups: those with at least one public comparison page (“X vs Y” or “alternatives to X”), and those without. Then we ran the top competitive queries for each company across all four models and measured citation rates.
The gap was 72%. Companies with comparison pages were cited in 58% of relevant competitive queries. Companies without them were cited in 34%.
62% of the 200 companies we audited had no comparison page at all. It’s the highest-ROI GEO asset you can build, and most companies haven’t built it.
§02the data
The effect holds across all four models, but the magnitude varies:
Perplexity shows the strongest effect (91% lift) because it actively crawls and indexes comparison pages. ChatGPT and Claude show strong effects because comparison pages contain exactly the structured recommendation content models look for. Gemini’s lower lift reflects its higher variance overall.
§03why it works
When someone asks an LLM “should I use Stripe or PayPal?” — the model is looking for a source that has already done the comparison. A comparison page is that source.
The mechanism is simple: LLMs prefer to cite content that directly answers the question being asked. A comparison page titled “Stripe vs PayPal — which is right for your business?” is a direct match for the query “Stripe vs PayPal.” Your homepage is not.
The secondary effect: comparison pages tend to rank well on Google for “X vs Y” keywords, which creates a feedback loop. Ranking well → getting crawled more → getting indexed more deeply by Perplexity → getting cited more.
§04anatomy of a high-citation comparison page
The comparison pages that get cited most share a specific structure. Here’s what it looks like:
- Opening verdict. A one-sentence answer to the question at the top. “If you’re a developer-first startup, choose Stripe. If you need consumer payments from day one, consider PayPal.” Models extract this and quote it directly.
- Honest trade-off table. A 4–6 row comparison table with specific, factual cells. Not “better”/“worse” — specific data points. Pricing. Supported countries. Integration complexity. Setup time.
- When to choose each. Two explicit sections: “choose us if you...” and “choose them if you...” This is the most-cited section. Models extract the decision criteria verbatim.
- Third-party validation. One or two external reviews or ratings that support your positioning. Embedded in the text, not just linked.
- Clear URL. /stripe-vs-paypal or /alternatives-to-paypal. The URL is part of the ranking signal.
§05build yours this week
You don’t need to build 12 comparison pages. Start with one: your most common competitive evaluation. The brand that comes up most in your lost deals, the one that prospects mention in demos.
Write it honestly. Acknowledge where they win. If you can’t acknowledge where they win, the page will read like marketing and won’t get cited.
Publish it at a clean URL. Submit it to Google Search Console. Share it in a relevant Reddit thread. Then run a check in 30 days and see if it moved your citation rate on competitive queries. In our test group, 87% of companies that published their first comparison page saw measurable lift within 45 days.