Published
The shape of the market
Bangladesh's business-to-consumer e-commerce market passed four billion dollars by early 2026, growing at around 22% a year. More than sixty million people now shop online with some regularity.
The internet population has grown from roughly thirty million in 2016 to more than 130 million, and 93% of those users reach the web through a mobile phone. More than 85% of online transactions begin in a mobile app or a responsive site.
Daraz dominates structured marketplace retail, with Chaldal in urban grocery, AjkerDeal among mid-tier sellers and Pickaboo in electronics. But a substantial share of trade never touches a marketplace at all, and that share is where most independent brands operate.
Selling through conversation
Facebook and Instagram in Bangladesh function as full marketplaces rather than as discovery channels feeding a website. Direct sales through messaging, and increasingly through live streams, are ordinary rather than experimental.
The commercial consequence is that the negotiation, the product questions, the delivery arrangement and often the price are settled in a chat thread. Cash on delivery closes a large share of it.
The measurement consequence is significant. A transaction beginning with a comment, continuing in Messenger and settling in cash leaves almost no trace in the analytics that Western e-commerce practice is built around. Businesses here are often described as having weak digital measurement when what they actually have is a sales process the standard tools were never designed to observe.
Where the website still earns its place
It would be a mistake to read this as evidence that a website is unnecessary. The site does a different job here than in a marketplace-led economy, and it is not a lesser one.
It is the verification layer. A buyer who found a business through a Facebook page checks whether it is real — whether there is an address, whether prices match what the page said, whether anyone credible has written about it. That check happens before the message is sent, and it decides whether the message is sent at all.
It is also the only asset the business owns. Page reach, algorithm changes and account suspensions sit outside the seller's control in a way a domain does not. Businesses that have lost a Facebook page overnight understand this better than those that have not.
And it is what machines read. AI systems and search engines cannot see a Messenger conversation. If the company's expertise, range and legitimacy exist only inside a social account, they are invisible to the systems buyers increasingly ask first.
The imported playbook and its assumptions
The digital marketing industry serving these businesses is estimated at $150m to $200m annually and growing at 35% to 40% a year, which means a rapidly expanding supply of advice imported wholesale from markets that behave differently.
That advice assumes a checkout, a pixel-based attribution model, a desktop consideration phase and a customer who prefers self-service to conversation. This market has a chat thread, cash on delivery, a phone and a customer who expects a reply.
The mismatch shows up as strategies that optimise a funnel most buyers never enter, and reporting that measures a conversion event most purchases never trigger.
What adaptation looks like
The changes are not complicated, but they have to be deliberate.
Build mobile-first without qualification — not responsive as an afterthought, but designed for a mid-range Android phone on an uneven connection, because that is the entire market.
Treat response time in messaging as a conversion-rate variable and resource it accordingly. It is the equivalent of checkout speed, and it is usually the largest uncontrolled variable in the business.
Make trust signals explicit and verifiable on the page: address, registration, real photographs, delivery terms, prices that match the social listing.
Measure the enquiry, not only the transaction. A tagged conversation count, a consistent question about how someone found the business, and a record of which posts produced threads will describe the funnel more accurately than any pixel.
Common questions
How big is e-commerce in Bangladesh?
The business-to-consumer market passed four billion dollars by early 2026, growing at around 22% a year, with more than sixty million regular online shoppers and 130 million internet users.
Why is social commerce so dominant in Bangladesh?
Facebook and Instagram operate as full marketplaces rather than discovery channels. With 93% of users mobile-first and cash on delivery widespread, buying through a Messenger conversation is the default rather than the exception.
Does a business selling on Facebook still need a website?
Yes, for three reasons: buyers verify legitimacy on the site before messaging, the domain is an owned asset that a page suspension cannot remove, and search and AI systems cannot read a Messenger conversation.