Engineering & Architecture: Build Decisions This Week
Date: 2026‑08‑24
The last week’s headlines remind us that a single headline can ripple across the entire stack—from regulatory risk to physical grid stability, and even to how we hire. In this post I’ll surface three concrete build actions that align with the news from South Africa, the UK/EU, and the broader tech community.
---
Signal:
South African consumers were hit by an abrupt order cancellation after a significant price drop on Makro’s e‑commerce marketplace. The incident was highlighted in MyBroadband's article “Laws for online retailers cancelling orders due to incorrect prices in South Africa.”
Why it matters to you:
The Consumer Protection Act (CPA) applies to all consumer transactions, including digital ones. If your platform can’t guarantee that a displayed price is the one charged at checkout, you expose yourself to regulatory penalties and brand damage.
Build Decision – Enforce Immutable Pricing States
| Action | Trade‑offs |
|--------|------------|
| Schema‑level price validation (JSON schema + database constraints) | Adds compile‑time safety but increases developer friction when launching new SKUs. |
| Immutable pricing microservice that records every price change with a cryptographic hash and timestamp | Provides audit trail for compliance reviews, but introduces a latency cost at checkout if not cached aggressively. |
| Client‑side preview of confirmed price in cart UI + server‑side double‑check | Enhances UX confidence but requires extra round‑trip for the final confirmation step. |
Recommended approach: Adopt an immutable pricing microservice with a read‑through cache (e.g., Redis) so that checkout remains fast while you satisfy CPA requirements. Add automated unit and integration tests that run against the price service before each release.
---
Signal:
The Guardian reports “UK orders reviews of franchising sector after Vodafone allegations.” A court claim involving 62 former franchisees has triggered a government review, suggesting tighter oversight of partnership contracts and revenue sharing.
Why it matters to you:
If your platform serves franchise networks—be they retail, hospitality, or SaaS—the regulatory environment is shifting toward greater transparency and auditability. Failure to capture every financial transaction correctly can lead to class‑action litigation.
Build Decision – Adopt a Multi‑Tenant Data‑Isolation & Audit Architecture
| Feature | Benefit | Cost |
|---------|---------|------|
| Tenant‑level encryption (AWS KMS, Azure Key Vault) | Protects each franchisee’s data under SA POPIA and UK GDPR | Adds key‑management overhead; requires careful rotation policies |
| Immutable audit logs (WORM storage) | Enables post‑mortem investigations and compliance reporting | Higher storage costs but mitigates future liability |
| Contract lifecycle management UI that triggers smart‑contract style events on data changes | Automates royalty calculations, dispute flags, and revenue splits | Requires building a domain model that mirrors legal agreements |
Recommended approach: Implement a tenant‑aware schema with separate logical databases (or schemas) per franchisee. Use an event‑sourced audit layer so every change is logged in a write‑once repository. Expose compliance dashboards for legal teams to pull reports on the fly.
---
Signal:
Two stories converge here: Euronews’ “EDF to prepare site for two new nuclear reactors at Gravelines plant” underscores that even mature grids can be unexpectedly stressed, and City AM details HMRC’s plan to send mansion‑tax inspectors for property valuations.
Why it matters to you:
Both stories highlight the need for resilient infrastructure that adapts to fluctuating power availability (especially in regions with high renewable penetration) and robust data pipelines that can handle sudden spikes in request volume—think a valuation inspector calling your API at 10 pm while a nuclear plant shuts down.
Build Decision – Energy‑Aware Scheduler + Secure Valuation Microservice
| Component | Trade‑offs |
|-----------|------------|
| Real‑time power monitoring API integration (e.g., national grid feeds) | Adds dependency on third‑party APIs; latency in scheduling decisions if not cached |
| Dynamic throttling of non‑critical workloads during high‑stress periods | Improves overall uptime but can delay user requests unless priority queues are managed |
| Property valuation ingestion microservice that validates address data, interfaces with HMRC APIs, and logs access | Handles spikes via horizontal scaling; requires rigorous encryption (TLS 1.3) and GDPR/POPIA compliance |
Recommended approach: Build a scheduler service that ingests grid forecasts every 15 minutes, applies simple heuristics to flag high‑stress windows, and throttles background jobs accordingly. For property valuations, deploy the ingestion microservice in an isolated VPC with strict IAM roles; use an audit log on each valuation request so you can replay for compliance purposes.
---
Addy Osmani’s recent talk “From Chrome DevTools to AI Engineering” illustrates how AI agents are beginning to automate routine engineering tasks. While the article focuses on Chrome DevTools, the principle is broadly applicable: if you can automate compliance checks or code review patterns with LLM‑based tools, you free up engineers for higher‑value work.
Caveat: The security team must vet model outputs and embed strict input sanitisation to prevent injection attacks. This trade‑off (speed vs safety) should be managed via a staged rollout.
---
These decisions align with the week's headlines while keeping your architecture lean, compliant, and resilient.
---
Please review these assumptions against your existing infra and regulatory compliance teams before proceeding.