Every analysis in this library is about a brand we have never worked with. They are independent breakdowns of publicly documented growth and infrastructure decisions at well-known D2C companies — reconstructed from official Shopify case studies, SEC filings, press coverage, and vendor-published performance data, all cited on each page. Carryup did not do this work, does not claim credit for these outcomes, and says so plainly at the top of every piece. Our own client engagements live separately, on the work pages, with our own numbers attached.
That separation is the point of the format. A conventional agency case study is a sales document about its author. These are closer to teardowns: what a company actually did, why the mechanism worked, and which part of it generalises to a store that isn't theirs. Below is how they're selected, sourced, and written — and how to read a growth number without letting it mislead you.
What these are — and what they are not
This library currently holds 23 analyses of brands including Gymshark, Liquid Death, Allbirds, Brooklinen, Mejuri, Fashion Nova, Kylie Cosmetics, Sugar Cosmetics, and Libas. None of them are Carryup clients. Each piece is written from public information about decisions those companies made and outcomes they or their vendors reported. We publish them because the useful knowledge in ecommerce is disproportionately locked inside a handful of well-documented stories that are usually retold as slogans — "Gymshark migrated to Shopify Plus" — with the actual mechanism stripped out. If you're looking for what we have built and measured ourselves, that's a different page and a different standard of evidence: our client work, with the metrics we're accountable for. Conflating the two would make both less useful, so we keep them structurally apart.
How we choose which brands to analyse
Three filters decide what gets written. The story has to be documented well enough that we can reconstruct a mechanism, not just an outcome — a headline number with no explanation of what produced it is unusable regardless of how impressive it is. The decision has to be one a smaller brand could plausibly face: a platform migration, a channel bet, an email discipline choice, a fulfilment fix, an international expansion. And the lesson has to survive removal of the brand's unfair advantages. Kylie Cosmetics' launches are not replicable by a store without a celebrity founder, but the infrastructure question underneath them — what it takes to survive a traffic spike that arrives in seconds — applies to any brand running a drop. Where a story only works because of scale or fame, we say so rather than pretending it transfers.
Where the numbers come from
Every figure is attributed. The strongest sources are official Shopify case studies and merchant-published data, where the company has put its name to the number; next are regulatory filings and earnings disclosures, which are the only figures subject to legal accountability; then reputable trade and business press; then vendor case studies from tools like Klaviyo or Postscript, which are reliable on mechanics but naturally selective about which customers they publish. Every source is listed on the article page so you can weigh it yourself. What we don't do is model, estimate, or round a number into something more impressive and present it as reported. Where public figures conflict between sources, we say so rather than quietly picking the higher one.
Why we explain the mechanism, not just the outcome
"Conversion doubled after migrating to Shopify Plus" is not a lesson, it's a headline — and taken at face value it implies the platform did the work. In the ARMEDANGELS analysis, as in most migration stories, the improvement traces to specific things the migration made possible: a checkout with fewer failure points, faster page loads, and payment methods that previously didn't work properly in certain markets. Those are the transferable parts, and some of them can be captured without migrating at all. Every analysis is therefore structured around the causal chain rather than the result. If we can't establish from public information why something worked, we say the mechanism is unclear instead of inventing a plausible-sounding one, because a confident wrong explanation is more damaging than an acknowledged gap.
How to read a growth number without being misled
Four questions defuse most misleading statistics. What's the baseline — a 300% increase from a very small starting point is a different event from the same percentage at scale, and press coverage rarely distinguishes them. Over what period, and what else changed in it: a brand that migrated platforms while also launching a new range and increasing ad spend cannot cleanly attribute the result to any one of those. Who published it, and what were they selling — a vendor case study is evidence of what's achievable, not of what's typical. And is it a rate or an absolute: conversion rate improvements mean little if traffic quality shifted underneath them. We try to supply this context inside each analysis, but it's worth applying to any case study you read, including ours.
What actually transfers to your store
The recurring finding across this library is that the transferable lesson is usually smaller and less exciting than the headline. Brooklinen's retention performance comes from restraint in email volume, not from a tool. Taylor Stitch grew revenue per recipient by sending less, not more. Mejuri's fulfilment saving came from fixing one shipping loop rather than renegotiating carrier contracts wholesale. Death Wish Coffee converted a two-hour traffic spike into a subscription base by having the capture mechanism in place before the spike arrived. None of these require the budget the brands had. What they require is a specific operational discipline applied consistently, which is why we end each analysis with the version of the lesson that a store doing a fraction of the revenue could act on this quarter.
How this differs from our own client work
Two libraries, two standards of proof, deliberately not mixed. These case studies are third-party analysis, sourced from public material, about companies with no relationship to us. Our client work is first-party: real Carryup engagements with named brands including Obeetee, The Last Ape, Motodrift, The Basics Woman, WUD Homes, Furnishka, Anantham Silks, and World of Toji, with the metrics we're directly accountable for. The reason for keeping them apart is that agency marketing routinely blurs this line — a page of famous logos with no statement of what the agency actually did leaves you assuming a relationship that may not exist. Every page in this library states in its opening that Carryup did not do the work, because a case study you have to interrogate for provenance isn't worth reading.
Corrections and updates
These are analyses of live companies, so they age. Reported figures get restated, acquisitions change ownership and strategy, and platform capabilities that made a decision necessary in 2019 sometimes ship as standard features later, which changes what the lesson means today. We revise pieces when a cited source is corrected or superseded, and we'd rather remove a claim than leave a number standing that its original publisher has since walked back. If you find something here that's wrong, out of date, or attributed to the wrong source, tell us and we'll correct it — including where the correction makes the story less interesting than the version we published.
