Siege Media looked at 116 B2B websites and asked a simple question: which type of page predicts AI search traffic best? The answer was versus pages. They came out about twice as strong as the next content type.
Ross Hudgens, who founded Siege Media 14 years ago, talked about it for an hour and a half on Edward Sturm's podcast. He covered how to build a versus page, where to put it, when to stop making them, and the one positioning mistake that keeps brands out of AI answers.
I run Distribb, and comparison content is one of the first things SaaS founders ask us about. So I watched the whole interview and wrote down everything you can apply to your own site this month.
We pulled 13 live B2B SaaS comparison pages and checked each one's format, table placement and wording. 11 of the 13, among them HubSpot vs Salesforce and Wrike vs Asana, sit on a dedicated landing page rather than a blog post. 11 built a side-by-side feature table, and 6 placed it within the first screen, including Pipedrive vs HubSpot and ClickUp vs Asana. None of the 13 pages called its own product best overall, all choosing a narrower, specific position instead.
What Siege Media found in 116 B2B sites
Ross started in search as a link builder, doing lead generation in-house for an insurance company. He had a boss who micromanaged him, so he started Siege. Fourteen years later he is still running it.
When Edward asked him what he does more of in 2026 than two years ago, he did not hesitate. Versus articles.
His team ran a correlation analysis across their clients. The sites with the most AI traffic, measured as total traffic and not traffic per page, were the sites with the most versus articles. Siege published the full study on their site. It covers 116 B2B Google Analytics properties and 1,112 transactional pages over the 90 days ending May 24, 2026. Versus pages had the strongest link to AI search traffic of every page type they tested, roughly double the next one.
The study also includes a caveat I liked. Content explains about 28% of the difference in AI traffic between sites. Brand authority and category fit explain the rest. So versus pages are the biggest lever inside your content, and your content is one lever among several.
→ If you only build one new page type for AI search this quarter, the data points to versus pages.
Why versus pages work in AI search
Ross's explanation is about where buyers are when they ask. Comparison questions happen late, when someone already has two or three options and wants help choosing. That is exactly the kind of question people now take to ChatGPT or Gemini.
Those questions also rarely show up in the prompts most companies track. Brands track "best CRM for startups" and similar category prompts. Buyers also type "HubSpot vs Pipedrive for a five-person team". When the prompt contains your name and a competitor's name, you are guaranteed to be part of the answer. The only question is what the AI says about you, and your own versus page is one of the sources it reads.
There is an old marketing name for this, comparative advertising. What changed is who reads it. It used to be only the buyer. Now it is the buyer and the model answering the buyer.
→ A versus page is the one page where you know in advance that you will be in the answer.
Put it on a landing page, not the blog
This was the most useful answer in the interview for me. Ross thinks the ideal versus page lives on a landing page, not on the blog.
His reasoning: the reader is at the very end of the funnel, so you can be more direct with your calls to action. Siege measured the two setups against each other for clients and saw a difference between the blog version and the landing page version.
In his words, it should feel like a traditional landing page with an "us versus them" headline, context and a call to action above the fold.
→ Treat the versus page as a sales page that happens to rank, and give it a sales page's layout.
The comparison table goes near the top
Ross wants a feature table on every versus page: what we have, what they have, side by side. Screenshots help.
He is not strict about the exact position, but it should come shortly after the headline. Edward pushed on this because he sees many sites bury the table far down the page. Searchers come for the table. Language models also handle tables well.
One warning from Ross: get the competitor's pricing right. If you get it wrong, you can annoy them enough to receive a takedown request, and the wrong number becomes context for the AI answer as well.
→ Check the competitor's live pricing page on the day you publish, and again every time you update the page.
Keep the page about two brands
A mistake Ross sees often: the page is titled "X vs Y" and then halfway down it turns into "and we are the best alternative to Y". He says to keep the versus page focused on the two brands.
Alternatives work better as a separate roundup on the blog. Someone searching "Y alternatives" wants several options, not only you. So you end up with two different pages doing two different jobs: the versus page for the head-to-head decision, the alternatives roundup for the person still shopping.
His other advice on tone is to position yourself humbly. Say where the competitor is good and who it suits. Then say who you suit. A fair page is also easier to get approved internally.
→ One versus page answers one decision. Anything broader belongs on a different URL.
Which competitors get a page, and when to stop
The target keyword is simple: your name versus the competitor's name. To choose which competitors, Ross looks first at sales calls. Which names come up when prospects say what else they are evaluating? Search volume is a good proxy on top of that.
That also tells you when to stop. If a comparison has no search volume and never comes up in deals, you have made enough pages. Competitor-versus-competitor pages, where you compare two other brands, are hard to rank unless you are very authoritative.
One idea he only started considering this year: three-way comparisons. In markets with three established players, people actually search for all three together. He used to think that was too broad to rank. Now he thinks it can work, and if you are the fourth brand, it can be a way to get into a conversation that normally excludes you.
He also pointed at HubSpot as a reference. They have a software comparisons library linked in their footer, a hub page that collects all their transactional pages. It gives every comparison page a place in the site structure and internal links.
→ Build the pages your sales team already needs, link them from one hub, and stop when the list of real competitors runs out.
Editorial pages, not a template
Edward asked the question every SEO is thinking: can I just generate a templated comparison page for every competitor and link them from a hub?
Ross's answer was no. Make them editorial. Products change fast, especially in B2B, and a contract can be worth hundreds of thousands of dollars. That means a comparison with small search volume can still be very valuable, and it deserves real work.
There is a second reason. A good versus page is also a sales asset. When a prospect says "how do you compare to Monday.com?", your sales rep can send the page. Nobody sends a spun template to a prospect.
→ If your sales team would not send the page to a prospect, it is not ready to publish.
When the client refuses to publish one
Ross admits the hard part is never writing these pages. It is getting the client to agree. Even "best tools for X" roundups on the client's own site are a difficult conversation, and a direct comparison is harder.
What changes minds is proof: when the client sees a competitor doing it and it working, the tone of the conversation changes quickly.
And when a brand still refuses, there is a workaround. Pay a YouTube creator in the niche to make the comparison. The same content then comes from them and not from you. Ross adds one detail: ask the creator to use your preferred framing in the chapter titles and the description, for example "best for marketing teams", so the text is there for models that cannot watch the video. In Edward's own tests, Gemini read YouTube transcripts correctly and the other models did not.
→ If you cannot publish the comparison yourself, make sure someone credible publishes it with your positioning in the text.
Publishing comparison pages without a content team
Ross said something that stuck with me: "these pages aren't hard to do". For an agency with a team of writers, that is true. For a SaaS founder with no content team, a set of editorial versus pages, an alternatives roundup, the pages around them and the links that make them rank is months of work that keeps getting pushed to next quarter.
That is the gap we built Distribb for SaaS to close. You connect your site once and it runs on its own:
- Buyer-intent keyword research on your market and your competitors, so the comparisons and alternatives your buyers search for end up in your content calendar.
- Comparison and alternatives articles written in your brand voice and published to your CMS on schedule, 30 a month on Pro and 90 on Accelerator.
- Original research tables: Distribb scrapes real data, builds a source-backed table and puts it in the article, the kind of proprietary context Ross keeps coming back to.
- Backlinks from real businesses through the backlink exchange, contextual links placed in relevant content, with no cold outreach.
- AI visibility tracking across five AI engines: which prompts mention you, which pages get cited, and how you compare with competitors.
Pro is $97 a month. Accelerator is $495 a month and adds more volume plus a human on our side who reviews every piece before it goes out. Both start with a 3-day free trial.
If you are still building the basics, our SaaS SEO checklist covers the rest of the foundation.
Stop calling yourself the best overall
When Edward asked about the most common mistakes new clients make, Ross named two. The first is weak competitive pages: spun up quickly, no screenshots, no design, no clear position.
The second is calling yourself "best overall". Ross pushes clients to decide who they are for instead. His view is that language models try to place each brand in a sub-segment: best for small teams, best for enterprise, best for agencies. If you say you are the best for everyone, you give the model nothing to place you with.
He gave a real example. A client appeared on many of the ranking pages for a keyword and still did not show up in the AI Overview. His read: the client described itself as best overall, while every other site used more specific framing, so the AI had no segment to put them in.
→ Pick the segment you win, say it on your site, and ask your partners and affiliates to use the same words.
Branded subreddits: who they are for
The second big topic was branded subreddits, a community on Reddit that the brand runs itself. Ross's answer to whether they are worth it: if you are small, not really. If you are large, yes.
His qualifiers:
- Your customers buy repeatedly.
- Your brand gets around 20,000 searches a month, or less if your contract value is very high.
- People have questions about your brand that are not easily answered on your website.
- Your brand sentiment is mostly positive. With negative sentiment, a community tends to amplify the bad posts.
The example he likes is a post on Chime's subreddit asking whether Chime is a good bank. The company could never publish that question on its own site. In its own community, a happy customer can answer it.
His recommended first six months: answer branded questions, the "is [brand] legit" type. There is little competition for those queries, and the people asking are current or near customers, which helps the community grow. Siege estimates the search value of a branded subreddit at around 2 to 4% of a site's current traffic value, before any AI effect.
→ A branded community only helps the brands people already talk about in a good way.
How Ross measures AI search
Siege uses paid trackers, but Ross's favorite measurement is free. Add an open-ended "How did you hear about us?" field to your signup or contact form. Make it a text field, not a dropdown, because people pick random options in dropdowns. When Siege did this for themselves, Claude showed up as a source out of nowhere. If a text field hurts conversion, put a multiple-choice version after the conversion instead.
His two quick wins for any agency or in-house team:
- Drop Perplexity from your tracker unless your own data shows it above 1% of AI referrals. In Siege's data it has less than 1% of influence, but many trackers weight it the same as ChatGPT or Gemini. That can push you to spend on citations that change almost nothing.
- Segment your tracking by topic instead of one visibility score. A single score hides the prompts that matter. His example is home insurance, where two head terms carry almost all the value and the long tail of city pages can make a site look healthier than it is.
If you are choosing a tracker, we compared the main AI visibility tools and what each one measures. For the on-page side, our guide on how to optimize for AI search goes step by step.
→ Ask your customers where they found you before you trust any dashboard.
What I'm taking from Ross
Versus pages first. Of all the content types Siege tested, this one had the strongest link to AI traffic. For a SaaS company it is also the page closest to a sale.
Build them like sales pages. Landing page layout, a table near the top, a clear call to action, accurate pricing, and a fair word about the competitor.
Let sales calls choose the list. The competitors your prospects mention are the pages to build. Stop when the list runs out.
Choose a segment. "Best overall" gives an AI model nothing to work with. "Best for teams of five" does.
Measure what customers tell you. A free text field on a form told Siege more than a tracker score.
Watch the full interview
The full episode is long, about an hour and 36 minutes. The versus page part is in the first 15 minutes, and the measurement advice starts around the 43-minute mark.
FAQ
What is a versus page?
A versus page compares your product directly with one competitor, usually targeting "your brand vs competitor" searches. It typically has a headline naming both brands, a feature and pricing table near the top, an honest section on who each product suits, and a call to action. Ross Hudgens recommends putting it on a landing page instead of the blog.
Do versus pages help with AI search?
Yes, according to Siege Media's research. Across 116 B2B sites, versus pages had the strongest correlation with AI search traffic of any page type, about double the next one. Content explained roughly 28% of the difference between sites, so brand authority and category fit still matter a lot.
How many versus pages should a SaaS company make?
Make one for every competitor that comes up in sales calls or has search volume for "you vs them", then stop. Ross says competitor-versus-competitor pages are hard to rank. Three-way comparisons can work in markets with three dominant players, especially if you are the fourth brand trying to get into that conversation.
Is a branded subreddit worth it for SEO?
Only for larger brands. Ross's qualifiers are repeat purchases, around 20,000 brand searches a month, open questions about the brand, and mostly positive sentiment. Siege estimates the search value at about 2 to 4% of a site's traffic value. With negative sentiment, a branded community can make things worse.
How do you measure traffic from AI search?
Ross's first recommendation is an open-ended "How did you hear about us?" field on your forms, since customers report sources like ChatGPT or Claude that analytics miss. On top of that, use a tracker segmented by topic rather than one overall score, and weight engines by their real share of your traffic.
Start with one versus page
Ross's advice for tomorrow morning was to put up one versus page. If you want the comparison pages, the alternatives roundups and the backlinks behind them to keep going out without a content team, see how Distribb runs SEO for SaaS companies and start the free trial.