Direct answer
"X or Y, which should I choose?" is one of the questions people ask AI most often before buying. In this head-to-head matchup, AI doesn't just name names: it picks a side, or leans one way, and above all it gives arguments for each. Knowing what AI answers when you're compared to your direct competitor — and on which criteria it makes you win or lose — is directly actionable: you discover the exact argument AI puts in your rival's mouth.
The problem
In the past, comparing two brands meant reading reviews, tests, and forums. Today, more and more people ask AI directly: "[you] or [your competitor], which is better?" And AI answers — often by naming a winner, or by splitting the use cases ("one for this, the other for that").
The problem is that you don't know what it answers. Maybe it recommends you. Maybe it recommends the other one, with a specific argument you could counter if you knew it. Until you've measured it, this matchup plays out behind your back.
The idea to grasp
A purchase comparison is a very different moment from mere presence. Here it's no longer "am I cited among others," it's "in the head-to-head, who does AI pick, and why."
Three things are at stake in a comparative answer:
- The verdict. Does AI lean toward you, toward the other one, or call it a tie? And how clearly?
- The deciding criteria. What AI bases its comparison on — price, quality, durability, service, innovation? These criteria reveal how AI structures the choice in your category.
- The argument given to the other side. This is the most valuable part: the exact sentence by which AI justifies preferring your competitor. You can work on this argument, qualify it, counter it — but only if you know it.
It's this last point that makes the comparison so actionable. A presence measurement tells you "you're there or you're not." A comparison tells you "here's the exact argument that's making you lose this matchup" — information of an entirely different value for marketing and positioning.
And as always with AI, the answer varies: depending on the wording, the mode (web or memory), the provider. A matchup can swing in your favor with one model and against you with another. Hence the need to measure, not to look just once.
What you hear everywhere
"People don't ask AI that." They do, on a massive scale. "X or Y" is one of the most natural uses of an assistant: people use it as a buying advisor. Ignoring it means ignoring a key decision moment.
"If I'm better, AI will say so." Not necessarily. AI compares based on what circulates about you and on its associations, not on an absolute judgment of quality. A competitor with a better story told online can win the matchup inside AI.
"Knowing the verdict is enough." The verdict alone only serves to reassure or worry you. What you can actually work on is the argument — why AI decides the way it does. Without the "why," there's no action.
My stance: facts only. Assuming AI prefers you is comfortable. Measuring the matchup — verdict, criteria, opposing argument — is useful.
My vision: measure the matchup, and above all its "why"
From here on, the register changes: we describe the instrument.
Measuring a purchase comparison means:
- Setting up the real matchups (you vs. your direct competitors), in the wording people actually use.
- Repeating each matchup to measure a trend, not a one-off shot.
- Recording the verdict (who AI picks, or how it splits use cases) and its consistency.
- Extracting the criteria and arguments given to each side — particularly the opposing argument, word for word.
- Comparing modes and providers, and dating/sealing to track over time.