How to migrate to the cloud without wrecking your business

Every second pitch deck in South Africa mentions cloud. Fewer mention that a botched migration is more expensive than staying on-premise. The cloud is not a destination — it is an operating model, and switching operating models mid-flight takes planning. Here is the order that works, the costs people miss, and the South African specifics that change the equation.

1.Why SA businesses hesitate — and when they are right to

Three concerns come up in every conversation: connectivity, cost in a weak-rand environment, and data sovereignty. The first two are solvable. The third is a legal requirement, not a preference, and any migration plan that treats it as an afterthought is storing up a compliance problem.

If your business runs happily on a single on-premise server with a good backup, cloud is not automatically better. Cloud wins when you need elastic scale, remote access, disaster recovery you cannot afford to build yourself, or when your on-premise hardware is approaching end-of-life and the refresh quote is eye-watering.

2.The migration order that minimises risk

Phase 1: Collaboration. Email, file storage, identity (Microsoft 365 or Google Workspace). Low risk, immediate productivity gain, and it gives your team cloud experience before anything critical moves.

Phase 2: Non-critical workloads. Development environments, internal tools, archival data. These are safe to experiment with and cheap to roll back.

Phase 3: Line-of-business apps. ERP, CRM, finance systems. Move these only after Phase 1 is stable and you have validated your connectivity. Consider SaaS replacements where they exist — migrating a legacy app to a VM in the cloud is often the worst of both worlds.

Phase 4: Latency-sensitive and regulated workloads. Real-time systems, anything with strict data residency requirements, systems with complex integrations. These go last because they need the most planning and the consequences of failure are highest.

3.Costs vendors quote vs. costs you will actually pay

The compute and storage line is the one on the proposal. The real cost includes: egress charges (data leaving the cloud — these add up fast), connectivity upgrades (you will need more bandwidth and probably a backup link), migration labour (someone has to move the data and test everything), training (your team needs to manage cloud infrastructure), licensing changes (some vendors charge more for cloud-hosted instances), and the first month of parallel running where you pay for both environments.

Budget in rand, stress-test the number at R20 to the dollar, and add 25% for things nobody predicted. If the business case still works, proceed.

4.POPIA and data residency

POPIA does not ban cross-border data transfer, but it requires that the destination country provides adequate protection, or that you have binding agreements in place. Both Azure and AWS have South African regions, which simplifies compliance. But check that your specific services actually run in the SA region — some features default to overseas data centres unless you explicitly configure them.

For regulated industries (financial services, healthcare), additional requirements apply. Get a written opinion from your compliance team before moving personal data offshore, not after.

5.Connectivity — the South African wildcard

A cloud-first business on a single fibre line with no failover is one cable theft away from a full outage. Plan for: a primary fibre connection, a secondary connection on a different route (ideally a different provider), and a mobile data fallback for critical systems.

Load shedding affects your ISP, not just your office. Ask your provider what their uptime guarantee actually means during extended outages, and whether their exchanges have generator backup.

6.The hybrid reality

Most South African businesses end up hybrid — some workloads in the cloud, some on-premise, and a growing number in SaaS platforms that are cloud by default. That is fine. The goal is not to be 100% cloud; it is to have each workload in the environment where it runs best, costs least, and complies with the rules.

The mistake is not planning the hybrid state deliberately. An accidental hybrid — some things moved, some forgotten, no central view — is harder to manage and more expensive than either pure model.

The bottom line

Cloud migration is an operating-model change, not a server move. Get the order right, budget honestly, sort your connectivity, and confirm your data residency before you start. Businesses that treat it as a phased project with rollback options migrate successfully. Businesses that treat it as a weekend cutover end up paying for two environments indefinitely.

General guidance based on experience. Your circumstances may differ — validate with qualified professionals.