How we shipped Bango Power's e-commerce platform in 18 days

When Bango Power’s founder rang us on a Friday afternoon, the brief was sharp: “We have 47 SKUs ready, three suppliers lined up, and Eid in five weeks. We need a real e-commerce site, not a Facebook page.”

By Tuesday of the next week, we’d shipped a working store. By the end of week two, bKash and Nagad were live. By day 18, the courier integrations were tested with real shipments. Here’s how that happened.

The decision tree on day one

The founder wanted everything custom-coded — no Shopify, no WooCommerce, no Wix. He’d been burned twice before by template builders that crashed under load. We agreed Laravel 12 was the right call: real backend, real admin, real ownership of the data.

The opening question wasn’t “what tech stack” — it was “what can we cut from MVP?” We negotiated three things off the list:

• Loyalty points (Phase 2)
• Multi-language toggle (Phase 2)
• Advanced filters by 6+ attributes (week-4 add-on)

What stayed:

• Mobile-first catalog
• bKash + Nagad checkout (both required — BD shoppers split roughly 60/40)
• Pathao + Steadfast courier APIs
• Admin to add products / mark sold-out / update prices
• Customer accounts + order history

The payment integration nobody talks about

bKash’s payment gateway documentation is functional. Nagad’s is less so. The trick is realizing that both expect HTTPS callback URLs with valid SSL — which means you can’t fully test on a .test domain. We deployed to a staging subdomain with Let’s Encrypt on day 4, before the design was even finalized, just to start testing the payment loops.

This single decision saved us a week. Most agencies leave payment integration until after design freeze — then discover the sandbox callback issues with three days to go.

Day 12: Pathao API throws a 401

Pathao’s merchant API authentication uses a token-refresh flow that’s poorly documented for the first call. The fix took 45 minutes after we finally got hold of their dev team via Discord — turns out the merchant ID in the JWT payload is required even on the initial auth call. Once cleared, courier label printing worked the first time.

What the launch numbers showed

Day 1 after launch: 127 visitors, 4 orders. Day 7: 1,840 visitors, 31 orders, average order value ৳1,250.

The founder messaged us on day 10: “আমার Facebook page-এ এক বছর ৩৫ orders ছিলো। ১০ দিনে ৩৪ orders করেছি এই site-এ।” (My Facebook page got 35 orders in a year. This site did 34 orders in 10 days.)

That’s not because the site is magic. It’s because a real product page with real photos, real prices, real shipping ETAs, and a real checkout converts better than a chat-with-the-page-admin flow. Every time.

What we’d do differently

In hindsight, we should have set up Meta Pixel + GA4 conversion tracking on day 1 instead of day 14. The first two weeks of organic traffic data is now harder to attribute back to specific channels.

Lesson learned, baked into our launch checklist now.