Deep-dive guide
Cloud Egress Shock Simulator: When Data Transfer Out Eats Your Margin
Executive industry analysis
Cloud pricing teaches founders to look at compute, and compute is cheap. The line that surprises people is data transfer out, usually called egress: the fee for every gigabyte that leaves the provider's network. Putting data in is generally free, but sending it to users is metered, and the per-gigabyte price looks tiny until you multiply it by video, images, backups or file downloads. A product that earns pennies per user can pay more to deliver its content than it earns from it, and the bill arrives weeks after the traffic that caused it.
The structure of the problem is a ratio between two numbers that grow at different speeds. Revenue per user is flat or grows slowly, while gigabytes per user tends to creep up with every new feature, higher image resolution and autoplaying video. Egress price is almost a fixed tax per gigabyte, so cost per user rises in step with traffic, and the margin per user falls even if the user count is doubling. Growth then makes the problem worse instead of better, which is the opposite of what a software business is supposed to do.
The domino model shows how it cascades. Data served per user and the number of users decide the total gigabytes. A CDN decides how much of that reaches the origin, and the origin is where the expensive egress price applies. The delivery bill then becomes a share of revenue, and that share determines the final profit. A modest drop in cache hit rate, caused by a changed cache header or a new URL pattern, can push several times more traffic to the origin overnight, which is how teams get a surprise bill from a deployment rather than from growth.
The iceberg separates the line everyone watches from the ones nobody does. The visible part is the egress charge itself and the rest of the infrastructure. Below the waterline sit CDN delivery fees, per-request and API charges that scale with every object fetched, egress wasted on retried and aborted downloads, and payment fees on the revenue. Each looks small alone, but together they add a meaningful extra layer on top of the headline egress number. Use the simulator to find the traffic per user at which profit reaches zero and to compare it with where you are today.
Formula and methodology blueprint
The model works per month. Total gigabytes are users times gigabytes per user. The CDN offload share is delivered through the CDN, and the remainder is origin traffic. Origin traffic above a small free allowance is billed at the egress price, CDN traffic at a lower delivery rate, and every gigabyte also triggers request fees. A small share of origin traffic is re-sent after failures and billed again. Revenue is users times revenue per user, minus payment fees and other infrastructure.
- Total traffic
- G = users × GB per user
- All gigabytes delivered in the month, before any caching.
- Origin traffic
- O = G × (1 − CDN offload)
- The gigabytes that still leave the provider network and are billed at the egress price.
- Egress bill
- E = max(0, O − free allowance) × egress price
- The headline charge, which falls only when traffic or price falls.
- Delivery bill
- D = E + CDN GB × CDN rate + G × request fee + O × retry share × price
- Everything you pay to get bytes to users, including the parts that are easy to forget.
- Net profit
- Net = users × ARPU × (1 − fee) − D − other infra
- The monthly result. The breaking point is where it reaches zero.
Cascading elasticity
Gigabytes per user and egress price act on the bill linearly, but their effect on profit is large because the bill is a big share of low-priced revenue. CDN offload is the most powerful lever once traffic is high, because each point moved off the origin saves the full egress price on that traffic. Revenue per user matters most for low-monetisation products such as free apps and media sites.
Three real-world case studies
Bootstrapped starter
A small app with light downloads
A new product serves a few thousand users with small images and modest files. Most traffic is cached and revenue per user is healthy, so the delivery bill is a minor line. Loading it shows how comfortable the early stage looks and how little warning you get before heavy features change the maths.
Scaling mid-market
A large audience with a solid cache
A product with a hundred thousand users and a high cache hit rate delivers a lot of data but pays the low CDN rate for most of it. Profit is large in absolute terms. Loading it shows why cache hit rate is worth engineering time, and how a drop would multiply origin traffic.
Black-swan crisis
A video feature goes live without a plan
A new video feature multiplies the gigabytes per user while the cache is cold and most traffic hits the origin at full egress price. Revenue per user stays low. Loading it shows the bill overtaking income and which lever, offload or traffic per user, brings the month back to profit.
Industry benchmark matrix
| Metric | Danger zone | Industry average | Top 10% elite |
|---|---|---|---|
| CDN offload (cache hit share) | Below 40% | 60–80% | Above 90% |
| Delivery bill as share of revenue | Above 40% | 10–25% | Below 8% |
| Egress price per GB (list, first tier) | Above $0.10 | $0.05–0.09 | Below $0.02 with commitments |
| Data served per active user per month | Above 20 GB for low-ARPU products | 2–8 GB | Under 2 GB with good compression |
| Share of origin traffic that is retried | Above 8% | 3–5% | Under 2% |
Variables and levers
Active users per month
People who download or stream something from your product in the month.
Lever: Do not chase user growth until the cost per user is known, because every extra user multiplies the same delivery bill.
Data served per user (GB)
Average gigabytes delivered to each user per month, including images, video, downloads and API payloads.
Lever: Compress media, lazy-load heavy assets and cap default quality, since gigabytes per user is the biggest lever you control.
Egress price per GB
What your cloud provider charges per gigabyte leaving its network once the free allowance is used.
Lever: Negotiate committed-use discounts, move bulk files to a low-egress storage provider or use a private interconnect for large flows.
CDN offload
Share of traffic served from a CDN cache instead of your origin storage or servers.
Lever: Raise cache hit rates with longer cache lifetimes, versioned file names and a tiered cache, so fewer bytes ever reach the origin.
Revenue per user per month
Average monthly revenue you earn from one active user, including ads, subscriptions and usage fees.
Lever: Price the heavy users: add a fair-use limit, a paid tier for large downloads or an overage fee tied to data.
Other infrastructure per month
Compute, databases, monitoring and tooling that do not depend on how much data you send out.
Lever: Right-size the rest of the stack, because fixed infrastructure decides how much egress you can absorb before profit disappears.
How to use this simulator
- 1Start from the realistic preset or enter last month's users, traffic and bill.
- 2Read the five domino boxes to see whether traffic, offload or revenue per user turns red first.
- 3Press Find my breaking point to learn how many gigabytes per user your plan can bear.
- 4Compare the current plan with the CDN-first plan on the 24-month chart.
- 5Set a budget alert at the traffic level the tool shows, and review cache hit rate monthly.
Mistakes to avoid
- Looking only at compute cost and ignoring the egress line, which for media and file products is often the largest part of the invoice.
- Assuming the cache hit rate stays constant, when a single header or URL change can send most traffic back to the origin.
- Counting only user-visible downloads and forgetting retries, health checks, bots, backups and cross-region copies that also leave the network.
- Setting one price for all users when the heaviest few percent generate most of the traffic and should pay for it.
Frequently asked questions
What exactly is egress?
Egress is data leaving a cloud provider's network, for example a user downloading a file or a video stream. Providers usually charge per gigabyte for it, while data coming in is typically free. Check your provider's current price sheet.
Is a CDN always cheaper than direct egress?
Often, but not always. A CDN is cheaper per gigabyte at volume and also cuts latency, yet it adds request fees and can have minimum commitments. Compare the blended cost per gigabyte for your traffic.
Why did my bill jump without more users?
Common causes are a changed cache header, a new asset size, an unversioned file that cannot be cached, a bot crawling large files or a feature that streams more data. Check offload rate and top URLs first.
Can I avoid egress fees entirely?
Rarely entirely, but some providers and storage services offer very low or zero egress prices, and CDNs or private interconnects reduce them. Rules change, so confirm current terms before deciding.
How do I find the breaking traffic level?
Press Find my breaking point. The tool moves each input towards its worse end and reports the value at which monthly profit reaches zero, ranked by how close it is to your current setting.
How accurate is this?
It is an educational planning model with the stated assumptions, not financial or technical advice. Replace the constants with your own price sheet and check against a real invoice.
This is an educational planning model with stated assumptions. It is not financial, tax or legal advice. Replace the inputs with your own figures and verify important decisions with a qualified professional.