B BROCENT

Guests Complained About Wi-Fi, Not Rooms: Managed Wireless at a Hong Kong Serviced Office Operator

A composite scenario from Hong Kong: a serviced-office operator running four locations keeps hearing the same tenant complaint — Wi-Fi, not the fit-out — and finds each site's wireless was set up separately with no shared standard and nobody able to see all four at once. Why the real question is who runs the UniFi controller, what Brocent's hosted controller covers at a per-access-point monthly price, where the scope boundary sits, and why the managed IT plan is still the thing that solves the underlying problem.

Coworking-space members at shared tables with laptops and coffee, representing tenants at a Hong Kong serviced-office operator's multi-site locations
TL;DR: A Hong Kong serviced-office operator running four locations kept hearing the same complaint on every tenant survey — Wi-Fi, not the fit-out, not the coffee — and traced it to a simple root cause: each site's wireless had been set up separately by whichever contractor built that location, with no shared standard and no one able to see all four networks at once. The fix wasn't better access points. It was putting every site's access points onto one managed UniFi Network controller — Brocent runs the controller, the operator keeps owning the hardware, and the price is per access point, per month.

Hong Kong's Serviced-Office Boom Runs on Wi-Fi, Not Square Footage

Hong Kong's flexible-workspace sector has grown past the point where "we have desks and a receptionist" is a competitive offer. Operators running anywhere from a handful of floors in Central to a portfolio spread across Kowloon, Wan Chai and Kwun Tong are selling something closer to a serviced product than raw square footage — meeting rooms that book themselves, printers that just work, and a network that a tenant's own IT team never has to think about. For a serviced-office or flexible-workspace tenant, the physical desk is table stakes. The thing they are actually renting judgment on, every single day, is whether the internet holds up during a client call.

That makes Wi-Fi quality unusually load-bearing for this industry in a way it isn't for, say, a retail chain or a warehouse operator. A shop can have a slow back-office network and the customer never notices. A serviced-office tenant notices immediately, because their own business runs on that connection for the hours they are paying to occupy the desk — video calls with their clients, cloud-based accounting and CRM tools, file transfers, VoIP handsets. When the Wi-Fi in a meeting room drops mid-call, it isn't an internal inconvenience for the operator; it's a visible failure happening in front of the tenant's own customer. Renewal decisions get made on exactly that kind of moment.

Growth compounds the exposure. An operator opening a third or fourth site isn't just adding square footage — it's adding another independent thing that can go wrong with the network, on top of the ones it already has. Without a shared standard, every new site is a fresh roll of the dice on whoever gets hired to wire it up.

One Site, One Contractor, One Undocumented Wi-Fi Setup

This is a composite scenario, not a named client, but it's a familiar shape across Hong Kong's flexible-workspace operators once they pass three or four locations. Each site was fitted out at a different time, often by a different contractor, sometimes years apart. The access points at the Central location are a different make and firmware generation than the ones in Kwun Tong. Nobody currently on staff was involved in specifying either deployment — the person who oversaw the original build has since moved on, and the "network setup" now lives as a folder of invoices and a vague memory of which electrician subcontracted the cabling.

Each site's wireless was configured in isolation, tuned to whatever that contractor considered good enough at the time, with no shared SSID naming convention, no shared guest-network policy, and — critically — no single place that shows all four locations at once. If the operations manager wants to know how the network at the Wan Chai site is performing today, there is no way to check without either calling the site's on-duty staff and asking them to describe what they're seeing, or physically going there. There is no single login, no single view, no single alert feed. Four sites means four separate, disconnected stories about whether Wi-Fi is working right now.

Tenants roam constantly within a single site, too — out of their private office, into a meeting room, past reception, into a shared lounge — and a wireless setup stitched together ad hoc rarely handles that roaming cleanly. Access points installed without a real design plan tend to have coverage gaps between rooms, overlapping channels that interfere with each other, and no consistent authentication handoff, so a tenant's laptop drops the connection and has to rejoin the network mid-meeting. It reads to the tenant as "the Wi-Fi is bad." It's actually "the network was never designed or managed as one system."

What Four Different Networks Actually Cost an Operator

Three real problems come out of this pattern, and they compound each other the longer the operator keeps growing on the same ad hoc basis.

The first is an inconsistent tenant experience across locations. A prospective tenant touring the Central site and then the Kwun Tong site — because a growing client wants two locations, or a broker is showing both — can have a materially different Wi-Fi experience at each, with no obvious reason why one location feels more reliable than the other. That inconsistency undercuts the operator's own pitch, which is usually built around a consistent, premium brand experience regardless of which site a tenant chooses. A serviced-office brand that can't guarantee the same Wi-Fi quality across its own portfolio is quietly working against its own sales pitch.

The second is the complete absence of early warning. Because nobody has central visibility into all four networks, the operator finds out about a problem the same way every time: a tenant complains. There's no view showing an access point rebooting in a loop, no alert when a switch port stops delivering enough power, no way to catch a problem building over days before it becomes a full outage during someone's investor call. By the time anyone at head office knows something is wrong, a tenant has already had the bad experience that prompted the complaint — and, often, has already mentioned it to two other people touring the building.

The third is the growth trap. Every new site the operator opens means repeating the exact same ad hoc process: hire a contractor, let them make site-specific decisions with no reference to what the other three sites are doing, and hope it holds up. There's no template to hand a new contractor, no standard bill of materials, no documented configuration to replicate. The operator isn't just carrying four disconnected networks today — it's on track to carry five, then six, each one an independent risk with its own quirks, unless something changes the pattern before the next lease is signed.

There's a quieter cost underneath the three above: the time the operations team spends firefighting instead of running the business. Every complaint about a dropped call or a slow connection eventually lands on someone's desk at head office, who then has to chase down which site it happened at, call around to figure out what's actually going on, and follow up with the tenant once — hopefully — it's resolved. None of that is the operations manager's actual job. It's a tax the ad hoc model quietly charges every week, and it scales with the number of sites exactly the way the underlying risk does.

The Real Question Isn't the Access Points — It's Who Runs the Controller

Here is the part most operators only discover after they have already bought hardware twice.

Business-grade wireless of the kind these operators typically end up with — Ubiquiti's UniFi range is the common case in Hong Kong fit-outs — isn't managed on the access points themselves. It's managed by a piece of software called the UniFi Network controller, which holds the SSIDs, the VLANs, the per-site policy, the topology and the statistics for every access point registered to it. Adopt an access point into a controller and it becomes part of a managed estate. Leave it without one and it's an isolated box someone configured once.

The controller software is free. That is exactly what misleads people. The hard part is not the licence — it's who runs the thing 24×7, backs it up, upgrades it, and answers for its downtime. That is the whole question, and for a four-site operator there are only three real answers to it.

Comparison: Three Ways to Run a UniFi Controller

  • Self-host it. The software costs US$0, and the server, the people and the risk are yours. In practice that means providing a Java and MongoDB environment — the official minimum is MongoDB 3.6, bundled with the installer — where even a small network needs roughly 1–1.5 GB of RAM just for the controller. It means opening and hardening TCP 8443 / 8080 / 8843 and UDP 3478 / 10001. It means owning upgrades, backups, hardening and availability yourself; Ubiquiti's own guidance is not to self-host if you can't keep it running continuously. And a self-hosted install runs UniFi Network only. This fits teams with a dedicated network engineer who are happy to own that server. A serviced-office operator whose IT function is one operations manager is not that team.
  • Buy another Cloud Key. A UniFi Cloud Key Gen2 Plus retails around US$239.99, one-time, and you generally need one per site. It works out of the box, but its resources are fixed and it does not scale horizontally, and it is a single point of failure: if the unit fails, loses power or loses its storage, the management plane goes with it. Upgrades and backups still need a person. For a four-site portfolio that's four boxes to buy, four to power, four to remember to back up. Ubiquiti's own hosted controller is a further option, from US$29/month for up to 1,000 devices — but the operations still sit with you. The one genuine advantage of the hardware path: if you also want UniFi Protect (video) or Access (doors), it is the only path, because those applications run only on Ubiquiti's own UniFi OS hardware.
  • Use a controller someone already runs (Brocent's model). Nothing to build: no server, no Cloud Key, no ports opened on the operator's own network. The access points stay the operator's assets — only the control plane moves, and it can move back. Upgrades, configuration backups, change execution and alert response are carried by a NOC. Priced per access point, per month.

*Third-party figures above are public retail and vendor list prices observed 2026-09-10, shown for comparison only — they are not Brocent quotes. Cloud Key and the official hosted controller are Ubiquiti products.*

What Brocent Actually Provides, and What It Costs

Brocent already runs a UniFi Network controller on its own BCS platform. For an operator in the four-site scenario, the change is narrow and specific: the access points at every location get adopted into that controller, and everything that used to be four separate undocumented setups becomes one managed estate with one panel over it.

The question that changes the price is only ever who owns the access points. Everything behind them — the controller, the NOC watching it, the change process, the monthly report — is identical either way.

Model A — bring your own access points. The operator has already bought UniFi access points, possibly across three different fit-out generations. They get registered into the controller Brocent already runs: adoption, SSID and VLAN, guest portal and enterprise authentication, 24×7 monitoring with alerts turned into tickets, firmware upgrades inside a maintenance window, daily configuration backup with three-year retention, and a monthly availability report. The hardware stays the operator's, and the configuration can be exported if they ever leave. Pricing is US$1.20 per access point per month on a 12-month contract, excluding tax — that is the current year-end offer, quoted with promo code YE26-UNIFIAP and honoured in writing through 30 November 2026. The standing list price for controller hosting is a custom quote; the per-access-point offer above is what is published right now.

Model B — hardware included. For an operator that would rather not put capital into hardware at all — particularly one opening a new site where nothing has been bought yet — Brocent supplies the access points as part of the subscription and keeps them running. No capital investment; the network is ready within an hour of subscribing. Hardware maintenance and replacement are included: a spare device is delivered and swapped within the agreed SLA, at no extra cost for the repair. The controller, the NOC monitoring, the change process and the monthly reporting are identical to Model A. This model is custom-quoted, because it scopes by site count and access-point count rather than by a single per-unit rate.

For a growing operator, the honest answer is often a mix: existing sites run Model A on hardware already bought and paid for, and the next location opens on Model B so there's no capital outlay in the same quarter as the fit-out. A consultant will price both against a real site and access-point list. The published rate and what it includes sit on the network, server and wireless pricing page, and the service itself is documented on the managed wireless network page.

What the monthly price covers

  • Access-point adoption and zero-touch provisioning
  • SSID, VLAN and per-site wireless policy
  • Guest portal, plus RADIUS / 802.1X enterprise authentication
  • 24×7 monitoring — offline access points, reboot loops and PoE faults raised as tickets
  • Firmware upgrades executed in batches inside an agreed maintenance window
  • Daily configuration backup, retained three years — roll back to any day
  • Monthly availability report, delivered through Report Central
  • Site-policy cloning when a new office opens

The commitments behind it

Four things are stated rather than implied, and they are the ones worth checking against any alternative quote. Availability is a 99% controller-service commitment, running highly available rather than depending on a single host — which is precisely the property a Cloud Key on a shelf in the Wan Chai site does not have. Configuration backup runs daily with three-year retention, so a mis-click or a device swap always has a way back. Scale is elastic: capacity is added as the access-point and site count grows, with no fixed access-point ceiling, so opening a fifth location never triggers a "the controller can't take it" conversation. Isolation is logical: the operator's sites, devices, data, credentials and logs are scoped to their own tenant, their accounts resolve only that scope, and Brocent engineer access is granted per client and logged. The underlying platform and runtime are shared — which is also part of why a per-access-point price can be what it is — and that boundary is stated plainly rather than dressed up as "completely separate". An operator whose own compliance requires a dedicated instance or a specific data jurisdiction should raise it before the quote so a consultant can scope it.

What Controller Hosting Doesn't Cover

This matters more for a serviced-office operator than for most buyers, because the same building often wants cameras and door access too — and it is worth knowing the boundary before someone promises a tenant something that can't be delivered.

UniFi is not one piece of software; it's several applications. Only UniFi Network — the wireless and switching controller — can be hosted as software at all, and that is exactly what this service provides. UniFi Protect (video), Access (doors) and Talk (telephony) cannot be installed on Brocent's servers, or on the operator's own: Ubiquiti allows those three to run only on its own UniFi OS console hardware, such as a Cloud Key Gen2 Plus, a UNVR or a Dream Machine. Ubiquiti's own self-host package, UniFi OS Server, doesn't change this — it currently covers Network, InnerSpace and Identity, and Protect, Access and Talk are not among them. Access has a physical constraint on top: every door needs an Access Hub on site to actually release the lock, which cannot be moved to the cloud.

So if a site needs camera coverage of a reception area or badge access on a tenant suite door, a UniFi OS console goes into that office. Brocent can specify, supply, deploy and maintain it — that is a hardware purchase, not a change to the controller-hosting rate. The wireless estate and the console can coexist at the same site perfectly well; they're just billed and delivered as two different things.

What This Looks Like Across Four Sites in Practice

The practical shift for an operator moving off the per-site ad hoc model happens in a few concrete pieces.

Adoption replaces configuration. Power an access point onto the network and the controller discovers and adopts it — nobody logs into individual units any more. The four sites stop being four independent configurations and become one estate with per-site policy: wireless networks and VLAN separation pushed per site or per floor, defined once and applied in bulk. Guest authentication runs through a branded portal with voucher and other methods, keeping visiting guests off the internal network; staff and resident tenants can connect through RADIUS / 802.1X with an account instead of a shared password that leaks to every departed member.

Monitoring replaces the complaint loop. Offline access points, reboot loops and PoE faults are watched 24×7 by the NOC and raised as tickets, so somebody is accountable for closing them rather than an alert dying in an inbox. Client counts, band distribution and congestion per site become the basis for capacity decisions — how many people that Kwun Tong lounge can actually take before it degrades, answered with data rather than a guess. Channel utilisation and interference are visible, so "full bars but no internet" finally gets an actual cause instead of a shrug.

Change becomes a process, not a favour. Adding an SSID for a new anchor tenant, changing a VLAN, upgrading a batch of firmware — none of it is agreed verbally in a chat group. A change is raised at the service desk or automatically by a monitoring alert; an engineer confirms the blast radius and the rollback plan and proposes a window; the operator approves and it gets scheduled, so business-affecting changes are not forced through during working hours; a configuration backup is taken first, then the change runs in batches inside the window and is verified; and the outcome is written to the change record and appears in that period's service report. For an operator whose tenants include regulated firms that get audited, that trail is the difference between "we think guest Wi-Fi is separated" and being able to show it.

Opening a site becomes a copy, not a rebuild. A mature site's policy clones to a new office instead of being configured from scratch, and pre-provisioned access points inherit it as soon as they are powered on. Under Model B, the network is ready within an hour of subscribing. The design work happens once, centrally, and gets replicated.

The report arrives whether or not anyone asks. Each period, an availability monitoring report states what happened, what it means and what is recommended — not a monthly export of graphs. It's delivered through Report Central, the same set of reports every managed client receives, and none of them is billed separately. The kind of finding that shows up is concrete: an access point at one site logging repeated unexpected restarts, traced to power-over-Ethernet delivery sitting at the edge of budget on the switch feeding it, with a recommendation to move it to a supported port and replace the unit under warranty if the restarts persist. That is the early warning the ad hoc model never produced.

And none of this removes the value of doing the physical design properly. A wireless site survey before hardware goes up — measuring how a space's layout, materials and expected tenant density actually behave — is still what determines whether a new site starts clean or starts accumulating the same coverage gaps and roaming problems the operator is trying to escape. The controller manages what's installed well; the survey determines whether the right things were installed in the right places.

Comparison: Three Ways to Run Wi-Fi Across Multiple Sites

  • Per-Site Ad Hoc Setup — Each location is fitted out by whichever contractor is available, with no shared standard, no central view and no documentation beyond an invoice folder. Problems surface only when a tenant complains, and every new site repeats the same undocumented process from scratch.
  • Consumer-Grade Mesh Systems — Easy to buy off the shelf and reasonable for a single small space, but built around one household or one office, not a multi-site portfolio. No central management across locations, no VLAN separation worth the name for guest versus tenant traffic, no change trail, and someone still configures each site's units by hand.
  • A Managed UniFi Controller (Brocent's model) — Every site's access points adopted into one controller that Brocent runs, priced per access point per month, with 99% availability, daily configuration backup retained three years, logical tenant isolation, a monthly availability report and no fixed access-point cap. The hardware stays the operator's, or comes included under Model B. New sites clone an existing site's policy instead of starting over.

Frequently Asked Questions

How is a managed UniFi controller priced across multiple sites?

Per access point, per month — not per site and not per user. That is deliberately the unit that scales with the actual estate: a four-site operator with twenty-four access points pays for twenty-four access points, whether those are spread over four floors or four buildings. The current published offer is US$1.20 per access point per month on a 12-month contract, excluding tax, quoted with promo code YE26-UNIFIAP through 30 November 2026; the standing list price is a custom quote. There is no fixed access-point cap, so adding a fifth site adds line items rather than triggering a platform change. Separately, operators already on a Managed IT Support plan can buy a Network & Wireless add-on that maintains switches, gateways and access points together at published per-market rates for Hong Kong, Singapore and mainland China — that is a different SKU from controller hosting, and a consultant will tell you which is cheaper for a given estate.

Do we need to buy access points from Brocent?

No. UniFi access points the operator already owns can be adopted directly, and controller hosting is priced per access point regardless of where the unit came from. If new units are needed, Brocent publishes per-model pricing and can handle deployment — but that's a separate hardware purchase. The alternative is Model B, where the access points come as part of the subscription instead of as a purchase, with maintenance and replacement included; that one is custom-quoted by site and access-point count.

The controller is in the cloud. If a site loses its internet circuit, do tenants lose Wi-Fi?

No. UniFi access points keep forwarding traffic using the configuration already pushed to them when the controller is unreachable. What's affected is management and statistics, not tenant connectivity — which is a meaningfully better failure mode than a Cloud Key dying on site and taking the management plane with it. The controller service itself runs highly available with a 99% availability commitment, and its configuration is backed up daily with three-year retention.

Will a firmware upgrade take a site down during business hours?

Upgrades run in batches inside an agreed maintenance window, with a configuration backup taken beforehand, and never across a whole estate at once during working hours. Each upgrade goes through the same change process as anything else: assessed, confirmed with the operator, executed, verified and recorded — and the outcome appears in that period's report.

Is our network data separated from other clients on the same platform?

Yes, by logical isolation. The platform runtime is shared — which is part of why a per-access-point price can be what it is — but the tenant is a hard boundary: sites, devices, data, credentials and logs are scoped to the tenant, operator accounts resolve only that scope, and Brocent engineer access is granted per client and logged. If a specific compliance obligation requires a dedicated instance or a particular data jurisdiction, raise it before the quote rather than assuming it's included.

Can you host our cameras and door access on the same controller?

No, and neither can anyone else. Ubiquiti allows UniFi Protect (video), Access (doors) and Talk (telephony) to run only on its own UniFi OS console hardware; they are not available as self-hosted or third-party-hosted software, and Ubiquiti's UniFi OS Server package covers Network, InnerSpace and Identity rather than these. Access additionally needs an Access Hub on site for each door. If a building needs either, a UniFi OS console goes in that office, and Brocent can specify, supply, deploy and maintain it — a hardware purchase, not a change to the controller-hosting rate.

Do we still need a site survey for every new location?

Yes, and that's design, not upsell. A wireless site survey is what tells the installer where access points actually need to go given that location's layout, construction materials and expected tenant movement — a plan that worked for the Central site won't automatically work for a different building with different walls and a different meeting-room layout. The controller makes a well-designed estate manageable; it does not compensate for access points mounted wherever a ladder happened to reach.

Can we move off the service later?

Yes. The access points are the operator's assets throughout under Model A. On termination the configuration can be exported and the devices adopted back into a controller the operator runs itself, or into a hardware Cloud Key. Brocent hosts the control plane; it does not hold ownership of the network.

Where This Leaves an Operator Planning the Next Site

None of this is really a story about access points. It's about whether an operator running multiple sites has one wireless standard with one accountable owner, or four separate ones that happen to share a company name on the lease — and every new lease signed on the old model just adds a fifth. The fix that actually holds isn't a better access point at the site with the loudest complaints; it's moving the control plane somewhere that's watched, backed up, upgraded and reported on by people whose job that is.

That's why managed wireless sits as one of the infrastructure capabilities inside Brocent's managed IT support plans, rather than as a wireless-only purchase an operator has to separately shop for, contract and manage alongside everything else running its sites. The same per-user plan that covers help-desk support, security monitoring and infrastructure management for a serviced-office operator's own back-office team is where the accountability actually lives — one service desk, one change process, one monthly report, one team that already knows all four sites. Controller hosting is a real, separately purchasable service and doesn't require a plan; but an operator deciding how to run IT across a growing portfolio is better served by deciding the plan first and letting the wireless estate sit inside it, than by buying a wireless subscription and still having nowhere to send the tenant whose laptop won't join the guest network.

The managed wireless network page documents the controller service in detail, and the published per-access-point rate and what it includes are on the network, server and wireless pricing page. The plan tiers and what each one includes are on managed IT support, with every other published rate on the pricing hub — and for an operator scoping this against a real site and access-point count before the next lease is signed, Brocent's team will price both buying models and send a sample availability report with the quote.

Share:

Ready to take action?

Turn these insights into a roadmap for your business.

Book a 15-minute no-obligation consultation with our APAC IT experts. We'll review your current setup and provide a tailored IT roadmap within 24 hours.

Explore all services
📋

Free Checklist

10 Critical Checks Before Expanding IT to Greater China

PIPL compliance, network segmentation, bilingual helpdesk setup, and more — everything your IT team needs before Day 1 in China.

Request the checklist →

📬 Monthly Asia IT Insights

China compliance updates, cybersecurity alerts, and IT tips for APAC teams — once a month.

No spam. Unsubscribe anytime.