What AI believes about Chase Sapphire.
When different kinds of shoppers ask an AI which travel card to get, the answer comes back with a shortlist of names and a recommendation based on what the models understand about the brands. That is measurable. And it is changeable. We looked at how five models understand Sapphire Preferred, what they say, and why.
Chase is not an Upmode client. This is a measurement example of how AI understands the brand, produced with Upmode's platform from public materials only.
Chase Sapphire
How often Sapphire Preferred is the answer.
A query battery and scoring system designed against personas measured ChatGPT, Gemini, Claude, Perplexity and Google AI Mode.
Dozens of travel cards compete for the recommendations of these AI models. Sapphire Preferred scored highly for being recommended once mentioned, and very well for converting that mention into a primary recommendation.
Where Sapphire Preferred is the primary recommendation
Mentioned at all, same order
Models can vary in their conversion rates from mention to recommendation. In this case, Gemini names it most often at 99% and recommends it in 45%, while ChatGPT recommends it most at 56% on 86% mentions.
The answer depends on who asks.
Models were asked through three kinds of card shopper lenses. How often Sapphire Preferred is the primary recommendation for each.
19 points separate Premium buyer from New card shopper.
The shopper opening their first travel card hears Sapphire Preferred. The premium buyer hears it too, in 95% of answers, but the recommendation goes elsewhere in nearly two thirds of them. Amex Platinum appears in 72% of those answers and Venture X in 61%.
Seven brand elements and how AI understands them.
How strongly each belief is attached to the card for the models. Strong means the models state it without being asked. Partial means it shows up with a hedge. Thin means it is rarely attached at all.
The two strongest beliefs are the ones a shopper hears first: the points are worth more, and the fee is fair. The welcome bonus, the thing the brand spends the most to promote, scores 14%: stated outright in 5% of answers and missing from 77% of them.
Recommended, not just described.
Multiple asks of the same query battery across each model produce different contextual outcomes. Context matters when measuring how AI is shaping customer perceptions and actions beyond these AI conversations.
“A very good first travel credit card for most people is the Chase Sapphire Preferred®.”
ChatGPT · Recommended
“If you’re fine with a $95 fee and want something simple but stronger than VentureOne, look at Capital One Venture Rewards.”
Where the recommendation goes instead · Perplexity names Capital One Venture
“If you want absolutely NO annual fee: Start with the Bilt Mastercard®.”
And again, on another model · Gemini names Bilt Mastercard
Who else in the competitive set is on the shortlist.
How often each other card is named anywhere in the answers.
Venture X is the card the answer reaches for when it wants something simpler. Amex Platinum is the card it reaches for when it wants something richer.
This is the read for Chase Sapphire. Yours is next.
Reach out and we can get started decoding how AI understands your brand, and what to do to get recommended more often.
Ready to talk live? Book a walkthrough.
