5D Library › Order-Bump Friction Elasticity, One-Click Upsell Kinetics and Cross-Domain Recommendation Matrices · 中文

Preface: The Arithmetic of Adding to an Order, at the Sale and After It

Note

Please read this first. The book studies the mathematics of order bumps at checkout, one-click upsells after payment and cross-selling between sites: optimal price ratios under valuation distributions, the decay of acceptance over time, mutual-information routing and the sample sizes it needs, and the constraints of refunds and consent. The valuation distributions, acceptance rates, refund rates and the needs structure of 30 sites are illustrative or synthetic models, there to show structure, not data from any real shop or user; all data are generated by the programs printed in the book. The book is not legal advice.

The moment a buyer has decided and opened checkout is the moment of least friction in the whole sales flow: adding one more item takes a single tick. In the first second after payment, interest is still high, and upgrading by paying the difference needs no re-entry of payment details. And the needs a buyer reveals on one site are often related to what another site offers. Each has a computable question: what price for the bump? when to offer the upgrade? which two sites to recommend?

The book continues Volumes 69 and 77 and forms a set with Volume 79, keeping the same approach: write rules of thumb as runnable arithmetic, see when they hold, and report negative results as they are.

What the book contains

Chapter 1 derives the optimal price ratio of a checkout bump and its contribution to average order value, comparing one and two bumps. Chapter 2 compares upsells in the first second after payment and on day 7, counting refunds and regret. Chapter 3 uses mutual information on a synthetic 30-site needs model to find each site’s top two cross-fleet recommendations, and computes how many users are needed to estimate the routes stably. Chapter 4 decomposes the order-value multiplier and compares a checkout bump with a bundle price on the product page. Chapter 5 computes the cost of the consent needed to use personal data across sites, and how many recommendations to put at the foot of a report. Chapter 6 compares Thompson sampling with an even split for price tests and shows Simpson’s paradox. Chapter 7 reads the Shiji chapter on money-makers and the Guoyu. Chapter 8 gives incremental revenue, guardrail metrics and a pre-registered holdout experiment.

Claims corrected after checking

  1. The optimal bump price ratio is not a fixed 0.25 to 0.55. It depends on how buyers value the bump: in the illustrative model with zero friction the optimal ratio is 0.451, adding +22.5% to average order value; across medians of 0.30 to 0.60 and log standard deviations of 0.6 to 1.0, the optimal ratio ranges from 0.246 to 0.812 and the contribution from +15.0% to +31.0% (p01).
  2. “Order value up 22% to 48%” requires matching take rates. A single bump at a price ratio of 0.45 needs a take rate of 0.489 to add 22%; 48% is possible only at a ratio of 0.55 with a take rate of 0.873, and in practice needs two bumps plus an upsell (p07).
  3. A bundle price on the product page can beat a checkout bump. In the illustrative model, with uncorrelated valuations of the two items, the bundle earns RM 88.85 per visitor and the checkout bump RM 72.48 (p08).
  4. Upsells in the first second after payment are accepted more, but refunded more too. Under illustrative parameters the first-second upsell nets RM 33.56 per buyer and the day-7 offer RM 7.46; the first second stays better as long as its acceptance exceeds day 7’s by a factor of 1.09 (p03, p04).
  5. Choosing the “top two” routes by mutual information is very unstable. In the synthetic model, even with 60,000 users only 14 to 17 of 30 sites have the same top two as the “truth”, because the candidates differ very little (p06).

How to read the boxes

The boxes separate published evidence, runnable Python calculations (16 programs, standard library only, fixed random seeds, each run at build time with its real output printed beneath it), classical sources, notes and the author’s hypotheses; every hypothesis states what result would show it wrong. Programs and their output use ASCII characters only; sites are labelled S01 to S30.

Disclosure

The author uses the works of 5D Library, including this book, as the methodological basis of his community courses, the 5D Ascension Cards (Dimensional Shift Card Decks) and the Yafu AI knowledge base, and runs a number of websites and paid online reading software, reports and courses whose order bumps, upsells and cross-site recommendations are exactly what this book studies, so he has a direct commercial interest in its subject. The book promotes no software, course, card deck or service, uses no user data from the author’s businesses, and accepts no sponsorship. Its needs structure and conversion parameters are synthetic or illustrative. The experiment in Chapter 8 must follow a pre-registered protocol, and its results will be published whether positive or negative.

Fan, Y. (2026). Order-Bump Friction Elasticity, One-Click Upsell Kinetics and Cross-Domain Recommendation Matrices: Zero-Reentry Impulse Attachment, Conditional Mutual Information and Cart-Value Multipliers [结账顺手加购弹性、一键升单动力学与跨领域推荐矩阵] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090207 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).