Jun.
Works

Case Study

Repeat
Work

For a proxy-shopping team: product posts that used to be rebuilt by hand every time are now generated from a pasted link.

Client
A team that buys Japanese goods on behalf of its customers
Business model
Whenever an official Japanese store releases new products, the team publishes a thread of posts on social media listing each item's photo, Chinese product name and price, and customers order by replying.
Pain point
For every release, that thread had to be built by hand: each item copied, translated and formatted from the Japanese site, one at a time.
Solution
A tool: paste the store's URL, the whole thread is generated automatically, and a person checks it before it is published.
01 — Before

Moved one item at a time, from scratch for every release

Every item goes through these four steps by hand, and the next release starts over.
  • SlowPre-orders in Japan are open for only a few days; a late post means fewer days for customers to order.
  • Error-pronePosts were reused from the previous release, and the price or release date sometimes stayed unchanged.
  • Veterans onlyHow each product name should be translated lived only in senior staff's memory and old posts.
02 — Now

Paste a URL, handle the whole release at once

The same page of items goes to the tool as one batch; a person checks once, right before publishing.

The format and translations were worked out backwards from posts the team had already published, so the output looks just like before and nobody has to learn a new way of working.

03 — Results

It saves more than time

New staff can publish on day oneThe rules live in the system, so there is nothing to memorize first.

Consistent product namesNo longer depends on who happens to be working that day.

Mistakes are stopped before publishingA missing field or an expired order deadline cannot go out.

04 — After

Who maintains it afterwards

Source websites get redesigned all the time, and a redesign stops the tool from finding the data. So maintenance went live together with the tool: the system runs regular automatic checks, hands anything unusual to AI to fix and verify, and a person reviews the fix before it goes live.

A person appears only at the last step.

Does your company have a task like this?

Data copied by hand from several sources, the same document rebuilt every time, rules only senior staff know — if any of these sounds familiar, let's talk.