The economics are different here
A checkout flow copied from an American storefront assumes payment at order.
When most orders are cash on delivery, the same flow produces inventory that is
committed but not sold, revenue that is recognised too early, and a courier bill
nobody reconciles. Everything downstream of checkout has to be designed for
that reality.
One stock pool, many channels
The single most common cause of angry customers in Bangladeshi retail is
overselling — the same unit sold on the website and again in a Facebook comment.
The fix is structural: one inventory service, reservations held for a defined
window, and every channel writing into the same order queue rather than a
spreadsheet per channel.
Making returns cheaper
| Stage |
What the system does |
| Failed delivery |
Courier status pulled automatically, order flagged |
| Return in transit |
Stock kept reserved, not silently restocked |
| Received |
Inspection step with condition recorded |
| Resolution |
Restock, repair or write-off, each with its own ledger effect |
| Settlement |
Courier statement matched line by line against records |
Getting this flow right typically recovers more margin than any amount of
storefront optimisation, because it attacks a cost that scales with order volume.
Built for peak weekends
Eid and campaign traffic arrive as a spike, not a curve. Systems are load-tested
against a stated target before the campaign, with caching and queueing where it
matters, so the checkout that is meant to take the year’s biggest orders does
not fall over while taking them.