How many access points does your venue need?

It is Friday, the room is full, and the WiFi has died. The broadband is fine: at nine on a Tuesday it flies. What changed is the number of people in the room, and that single fact is what most owners get wrong about WiFi.

"How many access points do I need?" is not a one-number question. It depends on how big the space is, how many bodies are in it on a busy night, and what the walls are made of. You can get a solid working answer yourself with a tape measure, a phone and an afternoon, but a proper survey beats any rule of thumb, and we will be straight about when you need one.

The one big AP in the corner

An access point (AP) is the box, usually on a ceiling, that broadcasts your WiFi. Almost every owner buying their first one gets the biggest, most powerful AP they can afford and tucks it out of the way, behind the bar or in a back office, then wonders why the table in the far corner cannot hold a connection.

Here is the problem. Range is not the same as a strong signal, and a strong signal from far away is not the same as a fast connection. WiFi is a two-way conversation: your AP can shout loudly enough to reach the corner, but the phone there has a tiny aerial and cannot shout back, so it keeps resending and everyone in between slows down. One powerful AP in a corner covers a fraction of the room well and the rest badly.

Walls make this worse. A pub with a solid bar, a kitchen wall and a cellar is a maze of brick, plaster and stainless steel that absorbs the signal. Aruba's own site-survey guidance puts it plainly: brick, cinderblock and plaster walls "should be tested to see to what degree they will attenuate RF transmissions," and concrete "can have a wide range of different propagation characteristics" (Aruba/HPE Indoor Site Survey VRD). Two cheaper APs spread across a space almost always beat one expensive one straining through brickwork.

Coverage and capacity are two different problems

This is the heart of it: you are solving two problems at once, and they pull in different directions.

Coverage is about area. Can a guest get a usable signal everywhere they sit? You solve it by placing enough APs so their signal reaches every table, with a bit of overlap so a phone hands over cleanly as someone walks from the bar to the garden.

Capacity is about crowds. Can the WiFi cope when fifty phones all want it at once? You solve it by adding more APs, even in a space a single AP could technically cover, so the load is shared. WiFi airtime is a queue: every device waits its turn, so the more devices on one AP, the longer everyone waits.

A space can have perfect coverage and still fall over on a busy night. That is a capacity failure, not a coverage failure, and you fix it by adding APs and shrinking each one's cell, not by buying a bigger box.

Cisco's high-density design guide says it plainly: "you cannot cover 400 clients on a single AP... so you're going to need a couple more access points." Its fix is a section titled "Smaller Cells, More Radios, More Bandwidth": turn each AP's power down a little so its cell shrinks, add more of them, and let each handle fewer phones at higher quality. That is the opposite of the one-big-AP instinct, and it is why a Friday-night failure is almost always a capacity one.

The rules of thumb (and where they come from)

Two numbers will get you most of the way. Treat them as a starting point, not gospel.

Coverage: about one AP per 95 square metres

Ubiquiti's UniFi design training states that most UniFi APs cover 1,000 to 2,000 ft² (95 to 185 m²) and recommends at least one AP per 1,000 ft² (95 m²) to get the best performance and connectivity (UniFi Academy). Cisco's high-density guidance lands in a similar zone, around one AP per 93 to 139 square metres for general indoor coverage. So a clean working figure is one AP per 95 square metres of usable floor, leaning denser if the space is full of walls.

The other half of coverage is overlap. Ubiquiti recommends neighbouring APs overlap at about -70 dBm so phones roam between them without dropping. In plain terms, do not place APs so far apart that there is a dead patch between them: a little overlap is the goal, not a fault.

Capacity: design for about 50 devices per AP

Ignore the marketing maximums on the box; a datasheet that says "300+ clients" is a lab figure, not a comfortable real-world load. The number that matters is the design target vendors use for high-density spaces. Cisco Meraki recommends "around 25 clients per radio or 50 clients per AP in high-density deployments." Ubiquiti agrees: "no more than 50-100 active client devices per radio." Aruba plans at 60 percent of a radio's hardware maximum to leave headroom "for inrush/outrush and normal crowd density variations."

For a venue, the safe design figure is around 50 active devices per AP. Count the bodies, not the seats: most adults walk in with a phone, plenty also carry a watch and staff have devices too. A 60-cover restaurant at peak might be running 80 or more connected devices, which is two APs on capacity grounds alone even if one would cover the floor.

Why 5 and 6 GHz do not reach as far

Modern APs broadcast on more than one frequency band, and the higher bands are fast but short-legged. A phone will show a healthy 5 GHz signal at the AP and a useless one two rooms away.

The physics is simple: higher frequencies fade faster over distance and through walls. Extreme Networks measured the free-space loss in the first metre at about 40 dB for 2.4 GHz versus about 47 dB for 5 GHz, "about 7 dB more," and concludes that "5 GHz radios will have a shorter effective range and a smaller coverage area than... 2.4 GHz radios." The 6 GHz band (used by Wi-Fi 6E and Wi-Fi 7) loses only about 2 dB more than 5 GHz, so the real gap is 2.4 GHz versus the higher bands, not 5 versus 6.

This matters for sizing because the fast 5 GHz band carries the load, so plan coverage off it. Your effective per-AP area is then smaller than the longer-reaching 2.4 GHz signal suggests. The 2.4 GHz band travels further but is slower and far more congested, sharing airspace with every baby monitor and microwave nearby, so design for usable 5 GHz coverage and the slower band looks after itself. There is more on the band trade-offs in our Wi-Fi 6 vs 6E vs 7 explainer.

Worked example: cafe, pub and hotel

Here is how the two rules combine for three common venues. Take the larger of the coverage and capacity numbers, then round up. These are a planning starting point, not a quote.

Read the pattern. The cafe is tiny and quiet, so one AP wins on both counts. The pub is coverage-light but crowd-heavy, with a beer garden the signal has to cross an exterior wall to reach, so capacity pushes it to three or four. The hotel flips that: dominated by area and walls, it is coverage-led, needing one AP roughly per room or pair of rooms because plasterboard and corridors carve the signal up. A single AP per hotel floor is a classic, expensive mistake. The pub limited by capacity and the hotel by coverage is exactly why one number cannot answer the question.

The do-it-yourself site survey

You do not need a consultant to get a decent first answer, just an afternoon and a method.

1. Pace out the space. Step out each room's rough floor area in square metres, including the upstairs, garden, function room and cellar bar, then divide the total usable area by 95 for a first coverage count.
2. Walk it and note the walls. Mark every solid wall, the kitchen, the stairwell, lift shafts, big mirrors, fish tanks and metal shelving. The signal does not care about your floor plan, only what sits between the AP and the phone, so each significant barrier is a reason to add an AP or move one closer.
3. Count peak bodies, not seats. Think about your busiest realistic hour, count people, assume a bit more than one device each, then divide by 50 for a capacity count.
4. Take the larger number and round up. Whichever of coverage or capacity is higher wins, then add one for headroom if the budget allows. An extra AP is cheaper than a reputation for bad WiFi.
5. Test with a free signal app. If you already have WiFi, install a free analyser app and walk the room watching the signal strength. Anywhere it drops below roughly -70 dBm on 5 GHz needs a closer AP. Do this empty and, if you can, again when busy, because bodies absorb signal.

Then plan power and data to each spot, because a ceiling AP needs a cable. Most business APs are powered down that same network cable using Power over Ethernet, so you rarely need a separate mains socket up there. The cable run has a hard 100-metre limit and your switch shares its power budget across all ports, so settle this before you drill. We cover it in the PoE and cabling guide.

When a proper survey beats any rule

Every rule of thumb above is an approximation, and the manufacturers who publish them say so. Ubiquiti treats its own coverage plots as "a general guide" and tells designers to plan with floor plans rather than a fixed radius, because coverage "depends on environment, materials and interference." Cisco advises a site survey before any real coverage estimate or bill of materials, and Aruba calls for "active testing to measure how far signals travel" in spaces full of shelving, freezers and walls.

So when do you stop guessing and pay for one? The rules of thumb are fine for a single-room cafe, a small salon or a typical pub. Bring in a surveyor with proper kit when:

• The building is large, multi-floor or made of concrete, brick or stone (hotels, warehouses, listed buildings).
• You run high-density events, like a packed bar with live music or a conference space, where capacity is the whole game.
• Getting it wrong is costly, because retro-fitting cable through a finished hotel is far more expensive than surveying first.

A predictive or on-site survey models your exact walls and tells you where to put each AP and how much power to run. For anything bigger than a couple of rooms it pays for itself, stopping you buying APs you do not need or wiring a dead zone you would only find on opening night. Whatever you land on, decide whether to wire each AP back to the switch or lean on a wireless mesh link: wire it if you can, as we explain in mesh vs access points, with a brand-by-brand steer in how to choose an access point for guest WiFi.

Check the answer with your own WiFi data

The rules of thumb get you to opening night. After that, the WiFi itself can tell you whether you guessed right, because every guest who signs in does so through a particular access point. If CaptiFi runs your guest WiFi, three views in its analytics (on the Growth plan and above) do most of the work.

• Popular Access Points ranks your APs by visits, with an average dwell time for each. Rename each AP after the area it covers (Front Bar, Garden Terrace) and the ranking reads like a list of rooms, and clicking a row filters the other charts to that area.
• Peak occupancy shows the most guests online at once and exactly when it happened, in your venue's own timezone.
• The Venue Map takes a photo or image of your floor plan. Drop each AP where it sits and the map draws a heat blob per AP, biggest at your busiest, with a badge counting the guests connected there right now, refreshed every minute.

Read them against the capacity target. If one AP carries most of the room at your peak while its neighbours sit quiet, that cell is doing too much: move it, turn it down a little so phones spread out, or add one beside it. If an area where you know people sit barely registers, guests there are probably connecting to a further AP, which is a coverage gap in disguise.

One caution on the numbers. These views count guests who signed in, and plenty of devices on a guest network never do: across CaptiFi venues on UniFi in September 2026, 55.4% of new devices on guest WiFi signed in. Add staff phones and card terminals and the real load on each AP is higher than the guest count, so check the connected-client figure in your controller as well before you decide an AP has headroom.

Once coverage is right, the portal does the work

Sorting the access points is the plumbing. It has to be right before anything else works, but it is not the bit that makes you money. A guest connecting to flawless WiFi and then leaving with you knowing nothing about them is a missed opportunity.

That is the layer CaptiFi adds. We are an external captive portal: the branded sign-in page a guest sees when they join your WiFi. It runs on top of whatever network you have just sized, authorising guests through the controller (no separate RADIUS server to stand up on UniFi, Omada or Meraki, and CaptiFi's own RADIUS server on Aruba, Ruckus, Fortinet and Cisco), and it supports UniFi, TP-Link Omada, Cisco Meraki, Aruba, MikroTik, Ruckus, Fortinet and DrayTek, or CaptiFi sends its own plug-and-play device, on a refundable deposit, if your hardware will not cooperate. The hardware and marketing decisions stay separate, which is the point.

In CaptiFi's venue data, 55% of the new devices that joined CaptiFi guest WiFi in September 2026 signed in. The automated welcome email arrives while the visit is fresh, review automation asks each guest who agreed to hear from you to rate their visit afterwards, and win-back campaigns give lapsed guests a reason to return. Marketing consent is asked separately from getting online and recorded for each guest, so you know exactly who agreed to hear from you. Size the network with this guide, buy the kit from your installer, then connect CaptiFi and test the whole guest flow on a 30-day free trial, from $69/mo.

Sources: Ubiquiti UniFi Academy ("Fundamentals of UniFi WiFi Design"); Cisco High Density and High Client Density design guides (cisco.com); Cisco Meraki "High-Density Wi-Fi Deployments"; Aruba/HPE Indoor Site Survey and Very High-Density VRDs; Extreme Networks "How Far Will Wi-Fi 6E Travel in 6 GHz?". The CaptiFi sign-in figure was measured from paying venues on 22 October 2026 (method on the guest WiFi benchmarks page) and is not a guarantee. Specifications and prices were correct at the time of writing (June 2026); confirm current figures with the manufacturer before you buy.