Thor Commerce

GraphQL commerce infrastructure for complex B2B, DTC and multi-market stores. We share implementation details, edge cases and build notes.

[Thor Commerce @ThorCommerce]

@KeisukeIshikawa Reduced-motion support is a good filter for these libraries. Curious whether the transitions fall back to instant state changes under prefers-reduced-motion or just shorten the duration.

[Thor Commerce @ThorCommerce]

@amifideles_ Per-country caps are another useful lever. Pumping traffic concentrates on a few expensive prefixes, and numbers rotate faster than a per-number limit can catch.

[Thor Commerce @ThorCommerce]

@HeySukhwinder I’d love to see revenue per recipient beside the 2.86%. That would separate list growth from how much each subscriber actually contributed.

[Thor Commerce @ThorCommerce]

@jackorgbuild Asking what else shipped that same month usually helps. Redesigns rarely land alone, and the honest candidates will say so instead of claiming the whole lift.

[Thor Commerce @ThorCommerce]

@MaheshPawaar The awkward case is the same key with a different payload. Storing a request hash lets you reject that second call instead of returning a successful response for the wrong operation.

[Thor Commerce @ThorCommerce]

@metalbear One assertion I’d add: one reservation row after two transport attempts with the same key. A handler can dedupe the charge and still reserve inventory twice.

[Thor Commerce @ThorCommerce]

@SrujanGowda_10 The response looks identical whether that’s one join or fifty queries, so tests stay green either way. Logging query count per request is the cheapest way to tell.

[Thor Commerce @ThorCommerce]

Thor Commerce is the GraphQL backend we're building for complex B2B, DTC and multi-market stores.

This account is where we share queries, architecture decisions, messy integration details and the parts that took a few tries.

[Thor Commerce @ThorCommerce]

@zuruApp Worth recording each payment as its own row and deriving the balance from them. Three instalments then leave dated records for aging and disputes.

[Thor Commerce @ThorCommerce]

@delimit_ai A schema diff in CI catches the rename before the tests do. Validating checked-in client queries against the new schema shows which consumers actually break.

[Thor Commerce @ThorCommerce]

@catmanyau @jose_murillo_d One approach: every write is a proposed patch carrying the record version it was based on. Approval only applies it if nothing changed. Keep the old values and rollback becomes another patch.

[Thor Commerce @ThorCommerce]

@AsphaltCowb0y Are you keying the dismissal off the head SHA? Curious whether a new comment wakes it back up or only new commits do.

[Thor Commerce @ThorCommerce]

@BaigSamee The Rs 1 opening is doing a lot of work there. Nobody feels silly placing the first bid.

[Thor Commerce @ThorCommerce]

@0xlelouch_ Carrier feeds are rough that way. A delivered scan arriving before out for delivery is the classic one.

[Thor Commerce @ThorCommerce]

@HeySukhwinder Loud week is a good name for it. The week itself usually looks fine in the numbers, which is why it keeps happening.

[Thor Commerce @ThorCommerce]

@yo Seeing 30% and 2.9% next to each other makes the rush to web checkout very easy to understand.

[Thor Commerce @ThorCommerce]

@MakeBookHQ Interesting concept! Seen it countless times.. Demand until the customer has to open up the wallet. Even a small refundable deposit gives a much more honest number.

[Thor Commerce @ThorCommerce]

@DsMatie Bogus Singapore traffic has fooled a lot of app devs into thinking a listing was taking off. Good to have it filtered at the source.

[Thor Commerce @ThorCommerce]

@Olu_Abiodun01 Going all the way through to checkout is the part most of these skip. Nice work, Abiodun.

[Thor Commerce @ThorCommerce]

@HeySukhwinder Pre-checked boxes are the worst. You end up emailing people who never meant to join, and the welcome flow takes the blame - In the end you just end up tricking yourself.