UniFi* and MikroTik* networks engineered for SA SMBs, proper coverage, proper segmentation, proper monitoring. Server infrastructure (physical or virtualised on a modern hypervisor stack or Azure*) built to be understood six years from now by someone other than the person who installed it.
Networks fail for a short list of reasons. Consumer-grade routers trying to do business-grade work. Access points in the wrong physical location. Years of accumulated firmware debt. No segmentation between staff, guest, VoIP, and IoT traffic. ISP equipment with no visibility into what's happening. We've seen a lot of networks, and the pattern repeats.
The fix is almost never "throw more hardware at it." It's a proper survey, a proper design (segmented, with the right radio planning for the physical space), the right hardware for the job, and a monitoring stack so you can actually see when something's degrading before it breaks. Most of our network engagements take 1-2 weeks to install and years to pay back.
We walk the office with a laptop, test coverage, identify dead zones, measure interference. Review the existing kit, switches, access points, firewall, ISP connection. Look at the actual usage patterns: video calls, VoIP, file sync, printing, VPN.
Written design document covering: physical equipment layout, VLAN structure, firewall rules, Wi-Fi radio plan, monitoring approach, documented change management. You see the design before anything gets ordered or installed.
Hardware arrives. Scheduled installation, usually over an evening or weekend to avoid disruption. Cutover tested, old equipment decommissioned, documentation handed over. Users typically don't notice the change until someone comments that the Wi-Fi suddenly works in the far corner.
Monitoring in place. Firmware updated monthly. Quarterly review of performance against baseline. Immediate response if monitoring flags something degrading. Network management is available as part of a managed IT agreement, with network monitoring priced separately as an add-on. Standalone network maintenance is also available.
For most SA SMBs between 10 and 200 users, the right answer is a UniFi* network (Ubiquiti*) with a MikroTik* edge router or Fortinet* firewall if security requirements justify one. UniFi* gives you a controller-managed, cloud-visible stack of switches, access points, and routing that scales from 10 users to a few hundred without a replatform.
MikroTik* is the right choice for edge routing when you need serious routing capability, VPN termination, or multi-WAN failover without enterprise pricing. Avoid consumer-grade ISP-provided routers for anything above 5 users. They have no segmentation, no monitoring, and no reasonable firmware support beyond 2 years.
A proper UniFi* network for a 25-50 person office in South Africa costs between R35,000 and R90,000 in hardware (switches, access points, firewall, cabling if needed), plus R20,000 to R45,000 in installation and configuration labour. Larger offices or multi-floor environments scale from there, usually linearly with user count and square meterage.
Ongoing cost is minimal, software updates are free, support contracts aren't required. Compared to the productivity cost of unreliable Wi-Fi, dropped video calls, and VPN that doesn't work from meeting room 3, the investment pays back inside 12 months.
The most common reasons office Wi-Fi fails: (1) router in the wrong place, behind a cupboard or in a server room, losing 50%+ of signal; (2) one consumer-grade router expected to cover 500m² when it was designed for a 100m² house; (3) 2.4GHz congestion from neighbouring businesses and devices (3 non-overlapping channels, dozens of networks); (4) outdated firmware not updated since 2019, leaving both security holes and performance problems unfixed; (5) ISP router that was fine for 5 users but can't handle 30 simultaneous video calls.
A proper Wi-Fi survey takes us 2-3 hours and the resulting design usually involves 2-5 access points, strategically placed, with proper 5GHz coverage and seamless handoff between APs as people walk around.
Usually no, but sometimes yes. The default answer in 2026 is cloud. Microsoft 365* for email and collaboration, SharePoint for file storage, Azure* or AWS* for any custom workloads. On-premise still makes sense when: you have very large datasets that would be expensive to move to the cloud at bandwidth rates (video production, scientific data), you have latency-sensitive line-of-business software (some legacy ERPs), you have data sovereignty requirements that your cloud provider can't satisfy, or you have intermittent connectivity that makes cloud reliance risky (some remote sites, mining operations).
Where on-premise is still right, we virtualise on a modern hypervisor stack. It is licence-free, well supported, and avoids the licence-cost shocks of proprietary hypervisors.
Yes. We use MikroTik* extensively for edge routing, firewall, VPN termination, and multi-WAN failover scenarios. It's the right tool when you need serious routing capability without enterprise pricing, a RouterOS* v7 device with sensible configuration will outperform most SMB-grade firewalls at a fraction of the cost. We also handle configurations other MSPs find intimidating: BGP, OSPF, complex routing tables, PPPoE and MPLS carrier handoffs.
* Products named on this page are examples of the stack we deploy. Actual recommendations depend on your environment, scale, and existing licensing. We have zero commission arrangements with any vendor. Recommendations are made on technical fit alone.