Fair usage policies explained

What a fair usage policy is

A fair usage policy (FUP) is a set of rules that defines the actual limits on plans marketed as "unlimited," ensuring that a small number of heavy users do not consume so much bandwidth that everyone else on the network suffers. DataRadius is ISP billing and management software that enforces FUP thresholds automatically — when a subscriber crosses the limit, speed changes take effect mid-session without manual intervention.

Every ISP oversubscribes bandwidth. A provider with a 1 Gbps backhaul link might sell 10 Mbps plans to 200 subscribers, knowing that they will not all use their full allocation simultaneously. This works because most subscribers browse, stream, and check email — activities that use bandwidth in bursts, not continuously. Peak concurrent usage on a well-managed residential network typically sits between 20% and 40% of total sold capacity.

The problem arrives when a few subscribers run 24/7 downloads, host servers, or redistribute their connection. One subscriber saturating a 10 Mbps link around the clock consumes 3.2 TB per month — roughly 50 to 100 times what a typical residential subscriber uses. On a shared backhaul, three or four subscribers like this can fill the link during peak hours and slow everyone else to a crawl.

A fair usage policy addresses this by setting a threshold — a data volume or usage pattern — beyond which the subscriber's connection is adjusted. The adjustment might be a speed reduction, deprioritization in the traffic queue, or a notification with a request to reduce usage. The point is not to punish heavy users but to prevent one subscriber's consumption from degrading the service that 195 other subscribers are paying for.

The term "fair usage" rather than "data cap" is deliberate. A data cap implies a hard cutoff — use exceeds the limit, and the connection stops. A fair usage policy is more nuanced. The subscriber is not cut off. Their service continues, but at a reduced level, for a defined period. The goal is proportional response, not punishment.

Why ISPs need fair usage policies

Without a fair usage policy, an ISP selling "unlimited" plans has no mechanism to protect the majority of its subscribers from the consumption patterns of a few, and the math of shared bandwidth makes this inevitable.

Consider a specific scenario. An ISP has a 500 Mbps backhaul serving 150 subscribers on 10 Mbps "unlimited" plans. Peak demand is about 200 Mbps — well within capacity. Then five subscribers begin running torrents continuously, adding 50 Mbps of constant demand. More heavy users join. When 20 are saturating during peak hours, every other subscriber experiences buffering and dropped video calls. Subscribers who use 50 GB per month — 90% of the customer base — start leaving.

Fair usage policies solve three problems simultaneously:

Network quality. By throttling or deprioritizing the heaviest users during congestion, the ISP maintains usable speeds for the majority. The heavy user's 10 Mbps might drop to 2 Mbps during peak hours, but the other 145 subscribers keep their full speed.

Financial sustainability. Backhaul bandwidth has a cost. An ISP paying per-megabit for transit cannot afford to deliver truly unlimited usage at residential prices. FUP aligns actual cost with actual revenue by limiting the outliers who consume disproportionately.

Plan differentiation. FUP thresholds can vary by plan tier. A basic plan might throttle after 200 GB, while a premium plan throttles after 1 TB. This gives subscribers a reason to upgrade without the ISP needing to sell explicitly metered plans.

Common fair usage models

ISPs implement fair usage through several models, each with different trade-offs in complexity, subscriber perception, and network impact. Most ISPs use one or combine two depending on their infrastructure and subscriber expectations.

Data cap then throttle

The most common FUP model. The subscriber gets full speed up to a defined data volume per billing cycle. After reaching the cap, their speed is reduced for the remainder of the cycle.

Example: A 20 Mbps plan with a 500 GB FUP. For the first 500 GB, the subscriber gets full 20 Mbps. After 500 GB, speed drops to 2 Mbps until the billing cycle resets.

Advantages. Simple to explain to subscribers. Simple to implement — the billing platform tracks cumulative data from RADIUS accounting records and sends a CoA when the threshold is reached. Subscribers understand exactly when and why their speed changes.

Disadvantages. Heavy users hit the cap early in the cycle and spend the rest of the month throttled. If the cap is too low, even moderate users are affected and feel the plan is not truly unlimited.

Tiered speed reduction

Instead of a single threshold, this model applies progressive reductions at multiple data levels. It feels less punitive than a single hard throttle.

Example: A 50 Mbps plan reduces to 25 Mbps at 500 GB, then to 10 Mbps at 1 TB, then to 2 Mbps at 2 TB.

Advantages. Moderate heavy users experience a gentle reduction rather than a cliff. The subscriber who uses 600 GB gets 25 Mbps — still usable for streaming — rather than 2 Mbps.

Disadvantages. More complex to communicate. The billing platform needs to track multiple thresholds and send CoA packets at each stage. Some subscribers find it confusing: "What speed am I getting right now?"

Time-of-day shaping

This model applies speed reductions only during peak hours and allows full speed during off-peak times, regardless of total data consumed.

Example: Between 6 PM and midnight, subscribers who have consumed more than 300 GB this month are throttled to 5 Mbps. Outside peak hours, they get full speed.

Advantages. Addresses the actual problem — congestion during peak hours — rather than penalizing total consumption. A subscriber who downloads heavily at 3 AM does not affect other subscribers because the backhaul is not congested.

Disadvantages. Harder to implement. Requires time-aware policy enforcement, which means either scheduled CoA packets or RADIUS attributes that encode time-based rules. Also harder to explain: subscribers may not understand why their speed changes at certain hours.

Deprioritization

Instead of reducing the subscriber's speed to a fixed rate, this model lowers their traffic priority in the queue during congestion. If the network is not congested, the subscriber gets full speed. If it is congested, their packets are served after regular-priority subscribers.

Example: Subscribers who exceed 500 GB are moved to a lower QoS queue. During peak congestion, they get whatever bandwidth is left after normal-priority traffic is served. During off-peak, they get full speed because there is no contention.

Advantages. The most network-aware approach. Heavy users are only affected when their usage actually causes congestion. If the backhaul is lightly loaded, nobody notices the policy.

Disadvantages. Requires queue management on the router (queue trees or simple queues with priorities). Harder to communicate to subscribers because the impact varies. A subscriber might get 18 Mbps during off-peak deprioritization and 3 Mbps during peak — the experience is inconsistent, even if the reason is sound.

How to set fair usage thresholds

Setting the threshold too low catches normal subscribers and generates complaints. Setting it too high fails to protect the network. Data from your own network — not guesses or industry averages — should drive the decision.

Measure your subscriber base

Before choosing a threshold, measure what your subscribers actually use. Pull accounting data for the last three billing cycles and calculate per-subscriber monthly data consumption. You need three numbers:

• Median usage — the midpoint. Half your subscribers use less, half use more. On most residential ISPs, this is 50 to 150 GB per month.
• 95th percentile — the level below which 95% of subscribers fall. This is your "normal" ceiling. Users above this are outliers.
• Maximum usage — the highest individual consumption. This tells you how extreme the outliers are.

A threshold set at the 95th percentile affects only the top 5% of subscribers — the heavy users — while leaving 95% completely unaffected. This is a defensible, data-driven position.

Adjust for plan tier

Higher-tier plans should have higher FUP thresholds. A subscriber paying twice as much for a 50 Mbps plan reasonably expects a higher data allowance than a subscriber on a 10 Mbps plan.

A common approach is to set the FUP threshold proportional to the plan speed:

Plan speed	FUP threshold (monthly)	Throttled speed
10 Mbps	300 GB	1 Mbps
20 Mbps	500 GB	2 Mbps
50 Mbps	1 TB	5 Mbps
100 Mbps	2 TB	10 Mbps

The throttled speed should still allow basic browsing and email. Throttling to 256 Kbps makes the connection effectively unusable and generates support calls. Throttling to 1 to 2 Mbps for entry plans and 5 to 10 Mbps for premium plans maintains a usable baseline.

Account for peak vs off-peak

If your congestion problem is limited to peak hours, consider time-of-day thresholds rather than monthly volume caps. Measure your backhaul utilization hourly over a week. If utilization exceeds 80% only between 7 PM and 11 PM, a peak-hour-only FUP is more proportionate than a 24/7 data cap.

Review and adjust quarterly

Usage patterns shift over time. Streaming quality increases (4K is 15 to 25 Mbps, up from 5 Mbps for HD), operating system updates get larger, and cloud backup adoption grows. A threshold set today may catch normal subscribers within a year. Review accounting data quarterly and adjust thresholds upward as median consumption increases.

Communicating fair usage to subscribers

The most technically sound fair usage policy fails if subscribers feel deceived. Transparency — telling subscribers what "unlimited" means before they buy, during the sale, and while they use the service — is what separates a fair policy from a bait-and-switch.

Terms of service

The FUP must be documented in the terms of service that subscribers agree to during sign-up. The language should be specific, not vague:

Bad: "We reserve the right to manage network traffic and may adjust speeds for heavy users."

Good: "Plans include a fair usage threshold. On the 20 Mbps plan, speeds are reduced to 2 Mbps after 500 GB of data transfer in a billing cycle. Speeds reset to 20 Mbps at the start of each cycle."

BEREC's Open Internet guidelines require ISPs in the EU to describe traffic management practices in contract summaries using clear, non-technical language. Even outside the EU, this standard is worth following — specific terms generate fewer complaints than ambiguous ones.

Plan descriptions

On the plan selection page, list the FUP threshold alongside speed and price. Do not bury it in footnotes:

Plan	Speed	Monthly data at full speed	After threshold
Basic	10 Mbps	300 GB	1 Mbps
Standard	20 Mbps	500 GB	2 Mbps
Premium	50 Mbps	1 TB	5 Mbps

Subscribers who see this before buying are making an informed choice. Subscribers who discover the limit only when they are throttled feel cheated.

Subscriber dashboards

Give subscribers visibility into their consumption relative to the threshold. A progress bar showing "342 GB of 500 GB used" is more useful than raw byte counters. When a subscriber is within 80% of their threshold, show a notification: "You have used 400 GB of your 500 GB fair usage allowance. After 500 GB, your speed will be reduced to 2 Mbps until your next billing cycle starts on November 1."

This warning gives the subscriber a chance to adjust their usage. Without it, the speed drop is a surprise — and surprised subscribers call support.

Throttle notifications

When the FUP threshold is reached and throttling begins, notify the subscriber immediately via SMS or email. Include:

• What happened: "Your fair usage threshold of 500 GB has been reached."
• What changes: "Your speed is now 2 Mbps until your next billing cycle starts on November 1."
• What they can do: "Upgrade to the Premium plan for a 1 TB threshold, or purchase a data top-up for 50 GB at full speed."

The notification converts a negative experience into a sales opportunity. Subscribers who understand the situation and have clear options are far less likely to churn than subscribers who simply notice their internet is slow and assume the ISP is unreliable.

Enforcing fair usage with RADIUS

The technical enforcement of FUP happens through the same RADIUS infrastructure used for subscriber authentication. The billing platform monitors data consumption through accounting records and sends CoA packets to adjust speed limits when thresholds are crossed.

RADIUS Interim-Update packets contain cumulative byte counters for each session, sent every 5 to 10 minutes. The billing platform aggregates these across all sessions within a billing cycle to maintain a running total per subscriber.

When consumption crosses the FUP threshold, the billing platform calculates the new speed, sends a CoA packet with the updated rate-limit attribute, logs the event, and notifies the subscriber. The router applies the new speed to the active session immediately — no disconnection required.

At cycle reset, the billing platform zeroes the data counter and sends a CoA restoring the original plan speed. If the subscriber is offline at that moment, the next Access-Accept on reconnection contains the unrestricted speed.

Edge cases to handle: if a subscriber is offline when their cumulative data crosses the threshold, mark them as throttled so the next Access-Accept includes the reduced speed. If a subscriber disconnects and reconnects while throttled, the new session receives throttled attributes. Data top-ups extend the threshold and trigger an immediate CoA restoring full speed.

Balancing network protection and subscriber satisfaction

A fair usage policy is a trade-off. Too aggressive, and subscribers leave because they feel deceived. Too lenient, and the network degrades for the majority.

Three principles help find the balance:

Affect as few subscribers as possible. The 95th percentile threshold means 95% of subscribers never see the policy in action. If your threshold catches 20% of subscribers, it is too low.

Keep the throttled speed usable. Throttling to 256 Kbps makes the connection unusable and generates support calls. Throttling to 2 Mbps still allows browsing, email, and standard-definition streaming.

Make the path to full speed clear. Whether it is waiting for the cycle to reset, purchasing a data top-up, or upgrading to a higher plan, the subscriber should always know how to restore full speed. Throttling with a countdown ("Full speed resumes in 8 days") or an action ("Add 50 GB for $5") feels manageable rather than punitive.

Fair usage policies are necessary for shared networks. The ISPs that implement them well treat the policy as a communication challenge, not just a technical one. The enforcement — RADIUS, CoA, accounting — is straightforward. The hard part is giving subscribers enough visibility that the policy feels fair.