Talk about your next project
Napkin.AI landing page

The lesson

Run a competitor comparison as a disclosed, repeatable test with a published rubric, and let the rival win the rows it really wins.

The move

Under the title and byline, a "Quick takeaways" box gives the verdict up front: Napkin scored 23 out of 30, Claude 21 out of 30, and the advice is "Use both." The body discloses "we make Napkin" and then lays out the test. It introduces a made-up firm, Hamlet and Hamilton Inc., quotes the prompt word for word, lists the models used, and shows a scoring table for Accuracy, Finding the Point, Visual Judgment, Memorability, Clarifying Questions and Editability. Sections then walk through "The First Pass," "Shaping the Presentation" and "Editing Afterwards" with product screenshots. A final score breaks out each criterion, including Accuracy: Napkin 4, Claude 5. A sticky sidebar offers "Try Napkin Slides for Free."

Why it works

  • Our readStating the scores and a "Use both" verdict in the first box gives a skimming reader the answer before any method detail.
  • Our readPublishing the exact prompt, the invented data set and the rubric lets a skeptical reader see how the result was reached and judge it for themselves.
  • Our readConceding accuracy and finding the point to Claude makes Napkin's wins on clarifying questions and memorability easier to believe.

What not to copy

Don't copy the fictional company, the specific model settings or Napkin's scores, and don't run this format unless you are willing to publish the rows you lose. Keep the order: quick verdict, disclosure, fixed test setup, visible rubric, step-by-step walkthrough, per-criterion score, with a persistent trial CTA beside it.

More from Napkin.AI

More landing pages

Steal it for your brand.

Tell us what you want to make. We’ll talk through how to put these ideas to work for your brand.