Product comparison pages capture the highest-intent search in any category: someone who has narrowed it to two options and is deciding right now. They're also the page type AI assistants cite most when answering buying questions, because an assistant asked "A or B" needs a source that compares A and B. Your product page only argues for A.
The catch is that comparison pages only work if they're fair, which is exactly why most brands can't bring themselves to publish good ones.
Why the fair version outperforms the rigged one
The instinct is to build a comparison where you win every row. Everyone has seen that page. Readers spot it in about four seconds, bounce, and go read a third-party review instead.
Rigged comparisons fail three ways at once. Readers distrust them and leave. Assistants have other sources to choose from and prefer the ones that read as evaluation. And you surrender the ground entirely to affiliate sites, whose comparisons are fairer than yours because they don't care which product wins.
A comparison that occasionally recommends the competitor is more persuasive than one that never does, because it makes every other claim on the page credible. Conceding the cases where you genuinely lose costs you the customers who were going to be unhappy anyway.
The structure that works
- The verdict, in the first three sentences. Who should pick yours, who should pick theirs. This is the passage assistants quote and the one skimmers read.
- A specification table covering the attributes buyers actually weigh, not the ones you happen to win. Price, the two or three decisive features, support, and whatever your category argues about.
- "Choose [yours] if..." and "Choose [theirs] if..." as parallel sections, given equal weight and equal length. Unequal length is the tell that gives away a rigged page.
- Pricing stated plainly, both sides, including what's not included.
- Real limitations of your own product, named specifically. Vague self-criticism reads as false modesty; specific concessions read as confidence.
Keep the table narrow enough to scan. Twenty rows is a spec dump; six to nine rows is a decision aid.
The rules that keep it credible
- Don't load the criteria: picking only the attributes you win at is the same rigging, done quietly.
- Get their facts right and keep them current: an outdated competitor price is the fastest way to look dishonest, and a competitor can point at it.
- Don't editorialize in the table: facts in the table, judgment in the prose.
- Never fabricate a weakness they don't have: beyond being dishonest, it's the kind of claim that draws legal attention in a way vague marketing doesn't.
- Date the page and update it: comparisons rot faster than any other content type.
What to build, and in what order
Not every competitor deserves a page. Work from what people search:
| Page type | Targets | Priority |
|---|---|---|
| [You] vs [top competitor] | The head-to-head everyone runs | First |
| [Competitor] alternatives | People already leaving them | First, high intent |
| [You] vs [category leader] | Big volume, you may lose fairly | Second, still worth it |
| Best [category] for [use case] | Roundup framing, your own version | Second |
| [You] vs [minor competitor] | Thin demand | Skip until the others exist |
The alternatives pages deserve special attention. Somebody searching "[competitor] alternatives" has already decided to switch. That's the warmest traffic in your entire category, and the page is usually easy to rank because incumbents never write it about themselves.
How this feeds AI answers
The mechanics tie back to how LLMs pick which brands to name: assistants run searches, retrieve pages, and build answers from what reads as evaluation rather than promotion. A well-built comparison page gives them a quotable verdict, a structured table, and clear conditions for each recommendation. That's close to the ideal source format.
It also means your comparison pages get read in contexts you don't control, which is one more argument for playing it straight. A page that only makes sense inside your own funnel looks bad quoted anywhere else.
Pick your single most-searched head-to-head and write that page this week, verdict first, with a real "choose theirs if" section. If writing that section feels uncomfortable, that discomfort is the exact reason the page will outperform the ones your competitors published. Pair it with the collection pages that feed it traffic, and the two page types cover most of what ecommerce search work is made of.