Signal, Not Superstition: Replacing Wi-Fi Guesswork With a Measured Site Survey

The Ticket That Reopens Every Afternoon

Most UK SMEs do not decide to have a Wi-Fi problem. They inherit one. An access point is added here, a range extender there, a new meeting room is partitioned off over a bank holiday, and the wireless estate accumulates quietly in the background like paperwork. Then a pattern emerges: at roughly the same hour each day, when the office fills and the video calls begin, the complaints arrive. A call drops. A tablet at the goods-in desk loses its connection. Someone in accounts discovers that a file transfer takes nine minutes rather than nine seconds.

The usual response is a reboot, a firmware update, or the purchase of another access point in the general vicinity of the unhappiness. Occasionally this works, briefly. More often it moves the problem, because nobody has established what is actually wrong. Wi-Fi is treated as weather: something that happens to a business rather than something that is designed, documented and measured.

A site survey is the remedy, and it is a far less exotic exercise than the term suggests. It is an inspection with instruments, carried out by someone who is not selling a particular brand of hardware.

Start With the Complaints, Not the Kit

Before anyone walks a floor with a laptop and a dongle, assemble the evidence the business already holds. Service desk tickets tagged to connectivity, with time stamps and locations. The names of the applications affected — Teams calls, the finance package, the warehouse scanners, the EPOS terminals. Whether the failure is a dropout, a slowdown, or a refusal to reconnect. For a small firm this may be a dozen tickets; for a mid-sized one it may be three months of them.

Three weeks of that record will usually expose the shape of the problem: a single room, a shift pattern, a floor that behaves differently from its neighbour. That context matters because a survey measures the radio environment, and the radio environment is only half the story. A surveyor who knows that complaints cluster in the same place each afternoon will look for capacity pressure rather than simply hunting for weak signal.

Three Questions a Survey Must Answer

A credible survey addresses three separate planes, and confusing them is the most common cause of wasted capital.

Coverage is where the signal reaches and at what strength. Interference is what else is competing for the same air. Capacity is whether the network can carry the number of devices and the volume of traffic actually present at the busiest moment. A site can score well on the first and fail entirely on the third — which is precisely the profile of an office that works at 9am and collapses at 3pm.

There are several survey methods. A predictive survey models the building from floor plans and is useful at design stage. A passive survey listens to the environment. An active survey associates a client device and measures the experience the network actually delivers, including throughput and roaming behaviour. For an existing site with live complaints, the sensible combination is a passive walk of all occupied areas, repeated during a quiet period and during peak occupancy, supplemented by active testing in the rooms where users struggle.

The deliverable should be a heat map set, an access point schedule, a channel plan, and a written findings document. If a supplier will only provide a screenshot of a signal map, that is not a survey.

Reading Coverage Properly

Signal strength alone is a crude measure. For general data work, a received level around minus 67 dBm is a reasonable floor; for voice and video, minus 65 dBm with a signal-to-noise ratio of at least 25 dB is the more useful target. Surveyors should also record the secondary signal — the second-strongest access point in each location — because a device that can hear one access point clearly and nothing else will cling to it long past the point at which it should hand over. Sticky client behaviour of that kind produces the maddening symptom where a laptop works in the corridor and fails at the desk.

Minimum data rates, band steering and roaming thresholds are configuration settings rather than hardware, and they frequently explain more than the floor plan does.

Interference and the Neighbours

In the 2.4 GHz band there are only three non-overlapping channels, and in most UK business parks all three are in use by somebody. Co-channel contention, adjacent-channel overlap, and non-Wi-Fi sources — microwave ovens, DECT handsets, wireless video links, poorly shielded equipment in a neighbouring unit — all reduce the airtime available to legitimate traffic. The 5 GHz band offers far more room, and 6 GHz where hardware supports it, but it comes with its own considerations: shorter range through dense walls, and DFS channels that must vacate when radar is detected. A survey that ignores the 5 GHz plan is only half a survey.

The output here tends to be configuration and channel planning rather than spending: reducing channel width where it is not needed, disabling legacy rates, rationalising the number of SSIDs broadcast, and ensuring the channel plan is fixed rather than left to automatic selection that reacts to the neighbours rather than to the business.

Capacity Is Where Most Estates Fall Down

A modern office may have three or four connected devices per person: laptop, mobile, headset, a desk booking panel, a meeting room display. Add handheld scanners, printers, CCTV, and building systems, and the device count climbs faster than headcount ever will. Capacity design therefore works from peak concurrency, not from the number of desks.

The measurements that matter are airtime utilisation at peak, retry rates, and the number of clients associated with each access point. Utilisation sustained above roughly half the available airtime during busy periods is a warning sign, not a curiosity — it means queues are forming and latency-sensitive traffic will suffer first. Meeting rooms deserve particular attention, because a room rated for eight people may host twenty connected devices once a hybrid call begins.

From Findings to a Funded Plan

A survey is only worth commissioning if it converts into a sequence of decisions with costs attached. The practical structure is a tiered plan: configuration changes that cost time rather than money; relocation of existing access points, which carries cabling and containment cost per drop; additional access points where density or structure demands them; and, occasionally, switching or cabling work where Power over Ethernet budgets or Cat5e runs have become the constraint.

Ask for the remediation quote to be itemised — survey and validation re-survey, cabling per drop, hardware unit price, licensing, commissioning, and a post-change verification walk. A second survey after the work is not an optional extra; it is the only proof that the problem has been solved rather than relocated.

A Score the Board Can Track

Few finance directors will engage with decibels, but most will engage with a single figure. Agree a simple coverage score: the percentage of occupied areas meeting the target signal and signal-to-noise thresholds during peak occupancy, reported alongside peak airtime utilisation and retry rate. Published monthly, that index turns a recurring grumble into a managed service line, and it gives the business a defensible trigger for capital spend rather than an argument conducted entirely in anecdote.

The same discipline should extend to change control. New partitioning, racking, a warehouse mezzanine or a relocated server room all alter the radio environment. If Wi-Fi is not on the change checklist, the estate will drift back towards its original condition within a couple of years.

What Ninety Days Should Show

Within a quarter, a business that has surveyed, remediated and validated should expect a measurable drop in repeat connectivity tickets, a shorter average resolution time because engineers are no longer guessing, and a wireless estate documented well enough to hand to any competent supplier. That last point is worth more than it appears: an accurate as-built pack — access point schedule, channel plan, cabling records, warranty details — is what makes a competitive tender possible and what protects the business if it ever changes provider.

The survey itself costs a fraction of a speculative hardware refresh. That is the argument to make internally, and it is the reason so many organisations find, once the heat maps are on the table, that they never needed a new network at all — only a documented one.