← Back to Resources 🏆 Composite / anonymised

Hong Kong office, Pearl River Delta factory: using SOPA to make housewares RFQs and back-to-back confirmation controllable

📅 2026-09-28 · English case · Composite / anonymised

Composite adaptation. Some details are anonymised.

This story combines common situations from Hong Kong and Pearl River Delta kitchen, housewares, and small-appliance OEM export teams. It is not an audited result for a named client. Figures are illustrative, not a guarantee.

1. Context

Picture a mid-size OEM export team of about 40–90 people: a Hong Kong office handling customer RFQs, price, and documents, and a Pearl River Delta factory handling BOM, capacity, packaging, and shipment. The lines are kitchen electrics and housewares — kettles, small appliances, storage — sold to overseas brands and buying offices.

Before the pilot, basic accounting or inventory software existed. The path from inquiry to first quote to customer PO to supplier lock still lived in email attachments, WhatsApp, and personal spreadsheets.

2. Challenge

The same RFQ arrived in a buyer email, a salesperson’s WhatsApp, and a factory group. Voltage, plug, colour, inner box, and barcode changes sat in long threads.
Old prices, old FX, and old MOQs were forwarded as if they were new. Customers sometimes saw three prices.
First quotes were slow because someone retyped the attachment into Excel. Even stable SKUs often waited more than a working day for a draft.
Back-to-back confirmation was the sharp risk: a customer PO was treated as accepted before materials or capacity were locked, or the PO quantity and pack version did not match the factory order until inspection.

3. Approach: a 90-day SOPA pilot

The team did not replace the ERP. For about ninety days they ran a narrow pilot on ManuTrade AI SOPA-style quotation and order assist. SOPA is the hero: extract line items from PDF, Excel, or images into a reviewable quote draft and an order draft, and keep back-to-back status explicit. A light Message Pool-style tag on inbound email or WhatsApp was optional, so the source was cleaner. The quote-to-lock chain stayed in SOPA.

Scope. Two higher-frequency lines first (for example one kettle family and one storage item). About two Hong Kong sales or merchandising people, one factory planning window, and one packaging or purchasing counterpart. The boss reviewed exceptions weekly. Out of scope: every SKU, an ERP cutover, and shop-floor vision.

Working rules. Extraction produces a draft, not an outbound quote. One current quote view holds price, MOQ, lead-time wording, validity, and version notes. Status runs inquiry → clarifying → quoted → customer PO → supplier or capacity locked → hold. A formal lead-time promise needs a named confirmation after lock. Mismatched quantity, date, or pack version is flagged for a person to judge.

Window	Focus
Weeks 1–2	Map RFQ inlets, pick two SKU lines, define back-to-back states, baseline first-quote time and “promised before lock”
Weeks 3–6	In-scope inquiries go through SOPA drafts; a short daily standup covers exceptions only; a person confirms before anything is sent
Weeks 7–10	Compare with baseline, tighten “no formal lead time before supplier lock”, cut retyping; add light inbound tags only if needed
Weeks 11–13	Decide: widen the SKU set, stay narrow, or adjust fields and who may promise a date

4. Results (illustrative only)

These figures are illustrative. They show a direction and a rough magnitude for discussion. They are not an audited result for this composite, and they are not a ManuTrade AI guarantee.

Signal	Illustrative before	Illustrative after the pilot	Read with care
First quote draft on the narrow lines	Median about 1.5–3 working days, depending on retyping	When the spec was complete, most reviewable drafts were ready the same day or the next working morning	Spec ping-pong is not “first draft”
Attachment to draft	Often about 25–40 minutes of retyping	Extract plus human review often under about 10 minutes	Custom and certification packs were outside the pilot lines
Old price treated as new	A few “three prices” disputes a month	Parallel versions on the pilot lines dropped noticeably	Needs one current quote view
Lead time promised before supplier lock	In peak season, almost every week someone said “basically confirmed”	Unlocked orders were less often treated as promise-ready	Discipline matters more than a percentage
Customer PO vs factory order	Mismatches often showed up at scheduling or inspection	Quantity, date, or pack exceptions were flagged at draft stage	A person still makes the call

Season, SKU complexity, and customer mix move these numbers a lot. Use your own pilot baseline. Illustrative ranges are not a savings promise.

5. Lessons

Measure first-quote time on a narrow SKU set before talking about the whole catalogue.
Back-to-back needs states — inquiry, quoted, customer PO, supplier locked — not one shared spreadsheet.
SOPA drafts; people promise. Extraction and mismatch flags speed the work. Price and lead time still need a named confirmation.
One current quote view is what stops old prices from being forwarded as new.
Inbound tagging can stay light. The hero for this pain is the quote draft and the lock, not another chat window.
The first month of logging and arguing about status is part of the pilot.

6. A quiet next step

If your team also sits between a Hong Kong window and a Pearl River Delta factory, with RFQs spread across email, WhatsApp, and Excel, a ninety-day narrow-SKU check is a practical way to see whether a SOPA-style quote and order assist fits. Learn the rhythm here. Measure your own baseline.

A short conversation, not a hard sell:

If funding or a digital subsidy comes up: approval is not guaranteed.

Read the full Traditional Chinese case

Again: composite and anonymised. Results are illustrative.

ManuTrade AI · 2026-09-28