Comment réparer un WiFi lent sans améliorer votre forfait Internet

How to Fix Slow WiFi Without Upgrading Your Internet Plan A Purple WiFi Intelligence Briefing [INTRO — approx. 1 minute] Welcome back. I'm speaking today as a senior solutions architect, and the brief I want to tackle is one that lands on my desk constantly: slow WiFi. Specifically, how to fix it without writing a cheque to your ISP for a faster pipe. This matters because in most enterprise and venue deployments I've reviewed — hotels, retail estates, conference centres, stadiums — the internet connection itself is rarely the bottleneck. The problem is almost always in the local network. The radio frequency environment, the access point placement, the QoS policy, the client density management. These are all things you can fix this quarter, with the infrastructure you already own. So in the next ten minutes, I want to walk you through the diagnostic framework, the key technical levers, the implementation priorities, and the pitfalls I see teams fall into repeatedly. Let's get into it. [TECHNICAL DEEP-DIVE — approx. 5 minutes] Let's start with the most common culprit: RF interference and channel overlap. In the 2.4 gigahertz band, you have 13 channels in the UK, but only three of them — channels 1, 6, and 11 — are non-overlapping. If your access points are all auto-selecting channels, there's a good chance several of them are transmitting on overlapping channels, which causes co-channel interference. Every packet collision forces a retransmission. Throughput drops. Latency climbs. Users complain. The fix is straightforward: run a spectrum analysis using a tool like Ekahau, NetSpot, or even the built-in diagnostics on enterprise controllers from Cisco, Aruba, or Ruckus. Identify which APs are competing with each other, and manually assign non-overlapping channels. In high-density environments, I'd also recommend reducing transmit power on the 2.4 gigahertz radio — counterintuitively, turning it down reduces the interference footprint and improves overall network performance. Now, the 5 gigahertz band is your friend here. It offers significantly more non-overlapping channels — up to 24 in the UK with DFS channels enabled — and far less congestion from consumer devices and neighbouring networks. If your APs support 802.11ac Wave 2 or Wi-Fi 6 — that's 802.11ax — you should be steering clients aggressively toward 5 gigahertz using band steering policies. Most enterprise controllers support this natively. The second major lever is client density management. This is the one that catches venue operators off guard. An access point rated for 500 Mbps aggregate throughput will deliver a very different experience when it's serving 8 clients versus 80. The IEEE 802.11 protocol is a shared medium — every client on the same AP is competing for airtime. The solution is proper AP density planning. In a conference centre or hotel ballroom, you should be targeting no more than 25 to 30 concurrent clients per AP in a high-density scenario. That means deploying more APs at lower power, rather than fewer APs at full power. This is a fundamental design principle that many organisations get backwards. You also need to look at your minimum data rate settings. By default, most APs will still allow clients to associate at legacy rates — 1 megabit per second, 2 megabit per second. A single client operating at 1 Mbps consumes a disproportionate share of airtime. Raising the minimum data rate to 12 or even 24 Mbps forces legacy clients to either connect at a higher rate or associate with a closer AP. It's a blunt instrument, but it works. Third: Quality of Service, or QoS. In a mixed-use environment — a hotel where guests are streaming video, staff are processing POS transactions, and the conference suite is running video calls — you need traffic classification and prioritisation. Without QoS, a guest downloading a software update can degrade the latency for a VoIP call or a card payment terminal. The framework I recommend is a three-tier model. High priority for latency-sensitive traffic: VoIP, video conferencing, POS. Medium priority for general business traffic: web browsing, email, cloud applications. Low priority, rate-limited, for bulk transfers: software updates, peer-to-peer, large file downloads. This is implemented at the controller level using DSCP markings and traffic shaping policies. Fourth: SSID proliferation. Every SSID you broadcast consumes airtime through beacon frames. I've walked into venues running eight or ten SSIDs — one for guests, one for staff, one for IoT, one for POS, one for CCTV, and so on. Each SSID broadcasts a beacon every 100 milliseconds by default. At scale, this overhead is measurable. Best practice is to keep it to four SSIDs maximum, and use VLANs to segment traffic rather than separate SSIDs. Fifth: Roaming behaviour. In a multi-AP environment, clients don't always roam to the nearest AP — they tend to hold onto their current association until the signal degrades significantly. This is called sticky client behaviour. The result is a client at the far end of a corridor still connected to an AP three rooms away, operating at a low data rate. 802.11r fast BSS transition, 802.11k neighbour reports, and 802.11v BSS transition management are the standards that address this. Together they're called the 802.11 RRM suite. Enabling these on your controller dramatically improves roaming behaviour and average client throughput. And finally: the backhaul. Even if your RF environment is clean and your AP placement is optimal, a congested uplink switch or a misconfigured trunk port will create a bottleneck that looks like a WiFi problem. Verify that your APs are connected to gigabit ports, that PoE budgets aren't being exceeded, and that your uplink aggregation is sized correctly for peak concurrent load. [IMPLEMENTATION RECOMMENDATIONS AND PITFALLS — approx. 2 minutes] So how do you sequence this work? I'd recommend a four-phase approach. Phase one: baseline measurement. Before you change anything, capture your current state. Run a WiFi analyser to document channel utilisation, signal strength, and noise floor across the venue. Record baseline throughput and latency from multiple client locations. This gives you the before-and-after data you'll need to demonstrate ROI. Phase two: RF optimisation. Address channel assignment, transmit power, and minimum data rates. This is zero-cost if you have an enterprise controller, and typically delivers the fastest improvement. In my experience, venues see a 30 to 50 percent improvement in average throughput from RF optimisation alone. Phase three: policy configuration. Implement QoS, band steering, SSID consolidation, and 802.11r/k/v roaming. This requires controller access and some testing, but it's still within the scope of a standard maintenance window. Phase four: analytics and continuous optimisation. This is where a platform like Purple adds significant value. Purple's hardware-agnostic analytics layer sits across your existing infrastructure and gives you visibility into client density, dwell time, session duration, and throughput trends — without requiring a forklift upgrade of your hardware. That data feeds back into your capacity planning and helps you identify emerging bottlenecks before they become user complaints. Now, the pitfalls. The most common one I see is making changes in production without a rollback plan. Always test channel and power changes during off-peak hours, and document the previous configuration. The second pitfall is over-relying on auto-RF features. Cisco's RRM, Aruba's ARM, and Ruckus's ChannelFly are all good, but they're not infallible in complex RF environments. Manual oversight is still required. And the third pitfall is ignoring the client side. A misconfigured roaming aggressiveness setting on a Windows laptop or an Android device can undermine all your network-side optimisation. Client-side diagnostics are part of the picture. [RAPID-FIRE Q&A — approx. 1 minute] A few questions I get asked regularly. "Should I enable Wi-Fi 6E?" If your client devices support it and you're in a high-density environment, yes — the 6 gigahertz band is essentially interference-free right now and offers 1200 megahertz of clean spectrum. But verify client device support before deploying. "How many APs do I need per floor?" For a standard office environment, plan for one AP per 1,000 to 1,500 square feet. For high-density venues like conference rooms or hotel lobbies, one AP per 500 square feet or fewer. "Is WPA3 worth deploying?" Yes, particularly in guest WiFi environments where GDPR and data protection obligations apply. WPA3's Simultaneous Authentication of Equals protocol eliminates the offline dictionary attack vulnerability in WPA2-Personal. For enterprise deployments, 802.1X with WPA3-Enterprise is the gold standard. "What's the quickest win?" Raise your minimum data rates and fix your channel plan. You can do both in under an hour and the impact is immediate. [SUMMARY AND NEXT STEPS — approx. 1 minute] To summarise: slow WiFi in enterprise and venue environments is almost never an internet capacity problem. It's an RF environment problem, a network design problem, or a policy configuration problem — and all three are solvable without upgrading your internet plan. The five levers are: channel optimisation, client density management, QoS policy, SSID rationalisation, and roaming configuration. Address them in that order, measure the impact at each stage, and you'll have a compelling ROI case for your next board review. If you want to go deeper on any of these topics, Purple has a full library of technical guides covering WiFi analyser tools, network design for hospitality and retail, and how to use analytics data to drive continuous network improvement. Links are in the show notes. Thanks for listening. Until next time.