A/B Testing Bandwidth Cost (Per-User) Calculator
Estimate the extra bandwidth and delivery cost per exposed user when an A/B test variant has a different page or asset size than the control.
Overview
Estimate the incremental bandwidth cost of an A/B test when the variant transfers more data than the control. Enter average transfer sizes, expected test traffic, variant allocation, expected page views, and your delivery rate per GB to see the cost per exposed user and for the full experiment.
How it works
The calculator first finds the positive difference between the variant and control transfer sizes. It multiplies that difference by expected page views per variant user, converts the result from KB to GB, and applies your bandwidth rate. The per-user result is then multiplied by the number of users allocated to the variant to estimate total extra transfer and cost. If the variant is smaller than the control, this calculator reports no additional bandwidth cost.
How to use this calculator
- 1Enter the average transferred size of the control experience in KB.
- 2Enter the average transferred size of the variant in KB.
- 3Add the total number of users expected in the experiment.
- 4Set the percentage of users allocated to the variant.
- 5Enter expected page views per exposed variant user and your bandwidth price per GB.
- 6Review the estimated incremental cost per user and total test cost.
Example Calculation
Control transfer size
250
Variant transfer size
400
Total test users
100000
Variant traffic allocation
50%
Page views per exposed user
1
Bandwidth rate
$0
Extra cost per exposed user
$0.000011
With 50,000 users exposed to a variant that is 150 KB larger, the test adds about 7.15 GB of transfer and costs about 0.57 in the selected currency, or roughly 0.000011 per exposed user.
Frequently asked questions
What is bandwidth cost per user in an A/B test?
It is the estimated incremental data-delivery cost caused by one user receiving a larger variant instead of the control.
Why does the calculator use variant traffic allocation?
Only users assigned to the variant create the additional transfer counted here, so the allocation determines how many users contribute to the total cost.
Does a smaller variant reduce the result below zero?
No. This calculator focuses on additional cost, so it reports zero extra bandwidth when the variant is the same size as or smaller than the control.
Should I use compressed or uncompressed page size?
Use the average transferred size seen by users, ideally after compression and delivery, because it is closer to billable data transfer.
Are CDN request charges included?
No. The estimate covers data-transfer charges only. Add separate estimates for requests, storage, serverless execution, or other provider fees if needed.
Why might billed bandwidth differ from this estimate?
Caching behavior, regional pricing, protocol overhead, redirects, asset reuse, provider rounding, and actual user behavior can change billed usage.
Explore Related Calculators
Assumptions and warnings
Assumptions
- The control and variant transfer sizes represent average data delivered for one page view or experience load.
- Only the positive size difference between the variant and control is counted as additional bandwidth.
- Each user assigned to the variant generates the entered average number of tested page views.
- The bandwidth rate is constant and excludes requests, storage, cache, platform, and other delivery charges.
- Results are estimates and can differ from billed usage because of compression, caching, and provider measurement rules.
Warnings
- This calculator provides a bandwidth cost estimate only; actual CDN or cloud invoices may include minimum fees, request charges, regional pricing, taxes, and other costs.