Role
Lead, The Edit and Wishlists
Team
Founder (PRD, type and colour system), four engineers
Timeline
Aug 2026 to present
Tools
Figma, Airtable, Claude Code, React, Supabase, Vercel
The Edit on desktop, tablet and phone, recorded on staging.
The short version
Problem
What I decided
Result
The brief
The Edit is where DADO shows taste before it sells anything. I lead it: the luxury layouts, the pages, and keeping every product and gift guide fresh as the year turns.
Premium layouts
Always current
In season, everywhere
The Calendar on production
The Calendar menu on production. It runs through the whole year, from Halloween, Diwali and Thanksgiving to Hanukkah, Christmas, New Year and wedding season, so members who celebrate different things all find their season.
The problem
The founder’s PRD puts six franchises on one item model, so a story can end on something you can wish for, and sets one test: a List reads in ninety seconds. By September a colleague’s ten seasonal editions sat on staging. Production held one, and an automated check that only asks whether a page loads called it clean.
The Edit on desktop and on a phone, from the homepage through the seasonal Edits and the city Editions.
The catalogue
Products, brands, gift guides and lists live in Airtable, so anyone on the editorial side with access can add or change them without touching a page. The admin console pulls Airtable in on a schedule and shows what it synced. Affiliate products are managed the same way, through the console and Airtable.
The console’s Catalogue on staging: products from Airtable with a Sync now button, the sync screen and its history, Missing links and Card brands.
59
gift guides
479
brands
678
gift guide items
665
products, 494 of them live
Staging counts from Airtable at the time of recording, 24 September 2026.
Decisions
Decision 01
Why
The PRD sets The List, DADO Presents, Editions and gifts on one platform and one item model. The Edit sits in the global nav beside Browse and One Big Wish, and every card carries the same actions.
What it cost
The homepage reads its own data file, and two files can disagree about what is current. The September catch-up was one such disagreement, fixed by hand.
The Calendar on production: ten seasons in order, the featured one above. Before the move, this index rendered the wedding edition.
Decision 02
Why
The PRD describes City Guides as numbered issues. The first six cities shipped with three sections each: The Hotels, The Tables and The Addresses. A colleague later added five more cities as data entries, with no new pages.
What it cost
A city that wants its own layout has to earn a template change, and none has yet. One template bug reaches all eleven pages at once.
The Paris Edition, one of eleven cities rendered by the same template.
Decision 03
Why
An automated crawl said everything was clean. Reading the pages found 18 dead links on the Gift & Give hub, 7 more in the mobile menu, 33 images that loaded a web page instead of a photo, and 11 pages giving search engines the wrong address.
What it cost
My link checker only compared the start of each address. It now checks every part and lists its six known false alarms. The image check opens every file, so it runs slower.
Ten seasonal pages that all loaded without an error, and what reading them found. Everything was fixed before it shipped.
Decision 04
Why
Collections and wishlists had drifted into two separate stores for the same idea, and a filter sent every birthday list to Collections. A delete could report success while the list stayed on screen, because the database said yes even when nothing was removed.
What it cost
The old tables stay so a rollback is possible, and merged lists start private. Every delete makes one extra check.
Before and after: one list with a type, and a delete that says when nothing changed.
Also decided
01
A pasted link brings its colours and sizes to a wishlist, through the same product page a wish uses.
02
The option picker collapses clones: H&M’s 58 rows became 46 real choices, and Dior’s 17 identical tiles one size axis.
03
The picker answers in 10.8 seconds, down from 39.4, and says what it is doing after six seconds.
04
Each edition’s chapters take its own photograph. 25 look-alike pages fixed; covers went from 4.5MB of JPEG to 1.9MB of WebP.
05
One line fixed in the city template gave all eleven city pages the right address for search engines.
06
The Calendar index came to production with the season, with 32 automated checks that flag it when it falls out of date.
Full decision traces available on request.
Outcome
10
seasonal editions on production, up from one
11
city Editions on one template
0
dead links in the seasonal nav after review
1
product page for a wish and a wishlist pick
Shipped in the private beta. Reading is not measured yet, so whether a List reads in ninety seconds is still untested.
Reflection
Every real problem in the September move was invisible to an automated check and obvious on the page.
The type and colour system is the founder’s, the List template is Somava’s and the chapters were written by a colleague. Mine is the lead, the pages I built and the state they shipped in.
What I kept
The image check that opens every file should have existed before the first chapter shipped, before thirty-three images broke.
