Starting AI SEO for SaaS with a Single Q&A Page

A friend recommended AI SEO, so I tried it on a few SaaS projects. SaaS is software subscribed to annually; buyers rarely go straight from search to the vendor page. They bounce between Google, ChatGPT, Perplexity (AI search tools), and G2/Capterra (software review sites). The goal of AI SEO is to make it easier for AI to cite you when answering; being found by Google is just part of it.

Day one: Clarify who is searching. Open the product page, click the browser address bar, and copy the URL. Then open an AI chat tool like DeepSeek or Doubao, click New Chat, and paste this into the input box: "I work in enterprise software operations. My target customers are team leaders of 20-100 person teams. List 15 questions they would ask from discovering a problem to comparing vendors, phrased as searchable queries." Hit Enter, and expect to see lists like "How to choose," "Who to compare with," and "Is it needed for small teams." Then input: "Categorize into Awareness, Comparison, and Purchase Risk." You should see three groups of questions. Next, open Google, enter three of these questions, and see if you appear in the results. Then open OpenAI or newly accessible Claude, paste the same questions into the chat box, hit Enter, and see if you appear in the answers.

The pitfall is stuffing keywords right away. I used a 4-week rule matching method, comparing question terms against page content to check coverage. It's more useful than keyword stuffing. Question terms need to be specific, covering pricing, migration, and data permissions.

Day three: Change one page of content into something machines dare to copy. In the product backend, click Content - New Page (using a common backend as an example). In the title bar, enter "Is XX suitable for teams under 50 people?" Do not write "Why Choose Us." In the body area, enter three paragraphs: Direct Answer, Applicability Conditions, and Evidence, keeping each around 30-40 characters. Click Preview; you should see a short title and clear paragraphs.

Add FAQs, at least five, phrased naturally like human speech, e.g., "Can we try it first?" Insert a table with columns: Suitable For, Not Suitable For, Onboarding Cost, Data Permissions, Pricing Model. Leave blank if no public info exists; don't fabricate. Use WeChat's voice-to-text feature to dictate the page logic to your team; after two weeks, it's faster than typing. Create a rule-matching table: left column for question terms, right column for presence/absence. If uncovered, fill it in.

To do AI SEO, write clear answers first. AI likes to copy explicit sentences. If you write "industry leader," "seamless," or "advanced," it can't grab onto anything. A pitfall I encountered was letting the model fill in competitor parameters, resulting in non-existent data written very smoothly. Later, I only let it organize existing materials, with manual verification before output.

After a week, check if it entered the "answer pool." The answer pool is the batch of webpages AI might cite when answering. Ask the same batch of questions again after a few days, switching between different tools. Ask: "Recommend a few XX SaaS options suitable for teams under 50 people, and explain why." Check if you're in the recommendation list. If you don't appear, modify the page: change "We support..." to "Teams under 50 can start with the basic version; export and permissions are in settings." If you do appear, check if the reasons are skewed. If it says you're cheap but you actually focus on compliance, add evidence. Only do three Q&A pages per week. My testing shows smaller pages are more stable.

Next steps: Take three competitors and make a comparison page, then ask AI for recommendations to see if it treats the page as evidence. Make answers copyable first, then talk about exposure.

---

📌 This article is compiled from Hacker News, original link: https://news.ycombinator.com/item?id=49597248

Copyright belongs to the original authors. This is a compilation and independent analysis based on public reports.