Competitive Compass Competitive Compass Signal for Financial Leaders
Mystery Agent  ›  Credit Cards  ›  Capital One
Credit Cards · August 2026

Capital One publishes full card comparisons but pricing details await scripts

An AI agent ran five ordinary credit cards tasks on the public pages of Capital One and scored what each page delivered.

Run 6 August 2026 · five tasks · public pages, logged out · Issuer
67/100
Mystery Agent score
Credit Cards · rank 17 of 25
Purchase APR
90Served as text
Card comparison
90Served as text
Pricing and terms
50Partly served
Welcome offer
92Served as text
Application page
15Automated reader declined
Summary
Capital One also appears in the Deposits study, where it scores 36 and ranks 22 of 25. See that scorecard.

Task by task

Each task is scored out of 100 on completion, answer quality and how directly the answer was reached. Each note describes what the page delivered, with a link to the page the answer was found on.

Task 01Purchase APRServed as text90 out of 100

The Venture Rewards card page states its purchase APR range of 19.49% to 28.49% and a $95 annual fee in a visible stat strip outside any script tag.

Task 02Card comparisonServed as text90 out of 100

The credit cards landing page lists Venture Rewards at a $95 annual fee with 2X miles and a 75,000 mile bonus, next to Quicksilver Rewards at a $0 annual fee with 1.5% cash back and a $200 bonus.

Task 03Pricing and termsPartly served50 out of 100

The Venture page states its purchase APR and $95 annual fee as visible text, while the fuller rates table loads its cash advance and penalty APR figures after the page arrives.

Task 04Welcome offerServed as text92 out of 100

The Venture page states the welcome bonus in full: earn 75,000 bonus miles after spending $4,000 on purchases within the first 3 months from account opening.

Task 05Application pageAutomated reader declined15 out of 100

The application link leads to a page that returns 84,923 bytes yet only 197 characters of visible text, with the application form and its requirements loading entirely through client script.

Earn 75,000 bonus miles once you spend $4,000 on purchases within the first 3 months from account opening.

The application page

What the account opening page states to a reader that runs no scripts.

84,923Bytes delivered
197Characters of visible text
0Form fields in the first response
After scripts runWhat an applicant needs

Strengths and opportunities

What is working

  • The credit-cards landing page (the creditcards page) is fully server-rendered and gives an agent a complete, accurate comparison of annual fees, rewards rates, and welcome offers across more than two dozen cards in one fetch, no script execution required.
  • The Venture flagship card page states its purchase APR range, annual fee, and welcome bonus terms including the exact spend requirement and time window in plain visible text, a genuinely complete answer for two of the five jobs.
  • robots.txt places almost no restrictions on agent access to card and marketing pages, and every page in this run delivered with a clean 200 on the first attempt with no challenge screens.

Opportunities

  • Publish the Rates and Fees Schumer box as static delivered text or a linked PDF rather than behind a javascript:void(0) modal; the underlying template fields are already named and present in the page, they simply are not resolved before the response ships.
  • The Quicksilver card page itself shows only 5,650 characters of visible text and never states its own annual fee or purchase APR outside of a script-embedded JSON blob; publishing the same server-rendered hero stats used on the Venture page would close that gap.
  • The application entry point serves an essentially empty JavaScript shell (197 characters of text, no form fields). Adding a lightweight server-rendered summary of what is required, SSN, income, address, would let an agent confirm the application path while staying clear of execute client script.
  • Publishing a real llms.txt, even a short one pointing to the card comparison and disclosures pages, would give agents a documented starting point instead of a 404 redirect.

Access

Every page the agent requested was delivered on the first attempt. Publishing an llms.txt file would give agent traffic a documented entry point.

Back to the credit cards leaderboard  ·  Read the report  ·  How the study works