Two case studies, written up in full.
Vasansi and Flawsome. Each has a page of its own below. You get what we changed, what we measured, and — where nothing was measured — a sentence saying so. Most agency portfolios are a wall of logos with a percentage stapled to each. This one is short. The short version is the honest one.
The Vasansi figures here are not screenshots. Both storefronts are still live, on separate domains. Fetch them yourself and you get the same answer we did.
- www.vasansi.com385,680 bytesThe original Shopify theme, still live
- www.vasansi.net3,084 bytesThe headless Next.js replacement
Initial HTML, before a single script runs. Open either domain and view source. The number is the page itself, not our account of it.
The same Shopify backend, behind a storefront over 100× lighter.
Vasansi sells apparel on Shopify. The storefront had grown the way storefronts do. A theme, then apps on the theme, then sections bolted onto the apps. We rebuilt the customer-facing layer as a Next.js application on the Storefront API. The admin, the checkout, the inventory and the payments stayed untouched.
The old storefront still answers on its original domain. That is unusual, and useful. The before state is not a claim we are making. It is a URL you can open right now.
Both figures are the initial HTML the server sends, measured with curl on 18 August 2026. We stand behind that number because you can reproduce it without our help. The full write-up covers the architecture, the image pipeline and what stayed on Shopify.
Read the Vasansi headless rebuildA product page rebuilt around the decision, not the description.
Flawsome sells organic cotton sanitary pads. The original product page was a stock theme layout. A very long product name, then a wall of description. The reasons to buy sat below the point where anyone had stopped reading. We rebuilt it around what a first-time buyer needs before they commit. What it is. Whether other people trust it. Which pack to choose.
The bundle selector, the trust signals and the shortened name are all in the before-and-after screenshots.
There is no conversion figure for this project. There will not be one here. The redesign shipped. We were never given access to the store’s analytics, so we hold no before-and-after of our own.
You will find ratings and product claims on the live page. Those are the merchant’s, and they describe the product, not our work. We do not repeat them as an outcome.
A design case study without a number is worth less than one with a number. It beats an invented one.
Read the Flawsome PDP redesignTwo more storefronts are live. Neither one is a case study.
Beat Guilt and Nobsci Nutrition are both trading on work we did. That makes them two live-store references. Open either and judge the storefront on its own merits. It does not make them case studies. We have not asked either client for permission to write the project up. We hold no measurement of what changed for them commercially.
The photography for both is shot and sitting in this site’s repository. Publishing it would take an afternoon. What is missing is consent. A client’s storefront going live is not a client agreeing to become a marketing asset.
What has to be true before a project appears on this page.
Four conditions. A project waits outside until it meets all four. That is why this page is short.
- 1
The client has agreed to be named
A storefront being live is not permission to write about it. Two have given it.
- 2
The before state is still observable
If the old site is gone, the comparison rests on our screenshot. That is not evidence.
- 3
There is an instrument and a date
A number without the tool that produced it and the day it was taken is decoration.
- 4
We took the measurement ourselves
We do not republish a figure from a client dashboard we cannot open.
Weighing us against other studios? The useful comparison is not who has more logos. It is which figures you can check without asking anybody. Ours is one fetch. The headless Shopify engagement page sets out scope and pricing. The apps we have built and run carry the same evidence rule as this page.
Questions we get about this page.
- Why are there only two case studies?
- Two clients have agreed to be written about. For each, there is something on the record. We have shipped more storefronts than that. A project appears here once the client signs off on being named and we hold a measurement of our own.
- Can I check the Vasansi numbers without asking you?
- Yes. Both storefronts are still live on separate domains. Open www.vasansi.com and www.vasansi.net and view source on each. Compare what the server sent before any script ran. We fetched both on 18 August 2026 and got 385,680 bytes and 3,084 bytes.
- Do you have a case study in my category?
- Probably not. Two published projects cannot cover apparel, supplements, snacking and cosmetics. An agency claiming otherwise is describing a portfolio it does not have. Ask us about the technical problem instead of the vertical.
- What will you not publish?
- A conversion lift we did not measure. A revenue figure we were told rather than saw. A client name without written agreement. A performance number from a screenshot instead of a fetch. Where one of those is missing below, the section says so.
Send us the URL that is losing you money.
We will fetch it and tell you what the server is actually sending. Then we will say whether it needs a rebuild or an afternoon. Often it is the afternoon.