E-commerce engineering that survives Black Friday.
Shopify Plus, custom commerce on Rails or Next.js, headless storefronts, inventory accuracy across DCs and POS. We have run BFCM traffic at 14x baseline with zero oversells.
What kills e-commerce engineering between Q3 and BFCM.
E-commerce engineering has one immovable deadline: the Friday after Thanksgiving. Every system that broke last year is going to break harder this year if nothing changes. Inventory is the single biggest source of post-BFCM regret across retailers we have audited: split across POS, two or three WMS instances, and the storefront, reconciled overnight, oversells in the dozens per minute during the peak hour. The other recurring losers are checkout, payment retry logic, and the email-marketing tools that hammer the database on Cyber Monday morning.
What changes when a Metafic pod is in your repo.
Inventory as event-sourced ledger
Each store, warehouse, and channel publishes events; a fast read model that POS and Shopify can query. Replay-from-zero is a first-class operation, not a heroics moment.
Checkout that does not depend on third-party flake
Payment retries with idempotency keys. Address validation cached. Cart-recovery emails async-triggered. The classic moves, done correctly.
Headless when the storefront warrants it
Next.js or Remix on Shopify Hydrogen or your own backend. We do not push headless on teams whose storefront is fine on classic Shopify.
Load testing against synthetic BFCM scenarios
k6, Locust, or Artillery against staging with realistic per-SKU concurrency. The traffic profile matters as much as the volume.
Returns and post-purchase done as carefully as checkout
Returns are 20-40% of revenue impact. Most retailers under-invest. We do not.
Who is on the pod for this work.
Pods scale up from here for Enterprise engagements.
Has shipped e-commerce production stacks. Knows the difference between a Shopify Plus customisation and a custom commerce platform.
5+ years, comfortable with both the storefront tier and the operational systems (OMS, WMS, payments).
Builds the BFCM-shaped synthetic test suite. Runs against real preview environments.
For the cutover weeks before BFCM. Auto-scaling, runbooks, on-call setup.
The bugs that bite this stack.
Shopify Plus webhook delivery without idempotency
Webhooks retry. Without idempotency keys, you process the same order twice.
Email marketing tools (Klaviyo, etc.) hitting your DB during peak
A common silent killer at peak. We add the rate-limit layer or move the data source.
Coupon stacking logic that double-applies during retry
Order placed → payment fails → retry → coupon applied twice. We add the constraint.
Stale inventory holds tying up units that could have sold
Cart abandoned but inventory held for 24 hours. We tune holds aggressively in the week before BFCM.
Honest about scope.
We will not migrate you off Shopify Plus to a custom commerce stack just because. Shopify is cheaper to operate than 95% of teams realise. We move off it only when the constraint is real.
On the US specialty retailer engagement, the inventory ledger rebuild meant zero oversells at 14.2x baseline BFCM traffic, down from 1.8% the prior year.
Read the full case studyCommon questions.
Shopify Plus, custom, or headless?
Shopify Plus for almost everyone under $100M GMV. Headless on Shopify when the storefront is the differentiator. Fully custom only when Shopify cannot model your business (uncommon).
Can you start before BFCM?
Yes, but the closer you get to October, the more we focus on shoring rather than rebuilding. The right BFCM work starts in Q2.
Do you do international expansion?
Yes. Multi-currency, multi-region tax, payment-method localisation. We have shipped to UK, EU, Singapore, UAE.
B2B commerce?
Yes. Net-terms, account-level pricing, approval workflows. Different shape than DTC; we have shipped both.
Ready to scope it?
A 25-minute call. We will tell you what we would do, what we would not, and whether a pod is the right shape.
Or stay in the loop. One engineering teardown a week.