A/B Testing Bandwidth Cost Per-User Calculator Examples
Worked examples showing how variant size, traffic allocation, repeat views, and delivery rates affect A/B test bandwidth cost.
These examples show the difference between a small page-weight change and a heavier variant used at scale. Each uses incremental transfer only, so a variant that is no larger than its control produces no additional bandwidth cost in this calculator.
Small UI experiment with a 20 KB increase
A product team tests an added recommendation module that increases the transferred page size slightly.
Input Summary
Control size
300 KB
Variant size
320 KB
Test users
20,000
Variant allocation
25%
Views per variant user
2
Bandwidth rate
0.08 per GB
Calculation Breakdown
- 1Variant users20000 × 25%5,000 users
- 2Extra data per user(320 − 300) × 240 KB
- 3Total additional transfer40 × 5000 ÷ 10485760.19 GB
- 4Total additional cost0.19 × 0.080.02
Result Summary
Total additional cost
0.02
A/B Testing Bandwidth Cost Per-User Calculator
The experiment adds about 0.19 GB of transfer and 0.02 in the selected currency.
Large image variant with repeat visits
An ecommerce team tests richer product imagery for half of 500,000 experiment users.
Input Summary
Control size
450 KB
Variant size
750 KB
Test users
500,000
Variant allocation
50%
Views per variant user
4
Bandwidth rate
0.10 per GB
Calculation Breakdown
- 1Variant users500000 × 50%250,000 users
- 2Extra data per user(750 − 450) × 41,200 KB
- 3Total additional transfer1200 × 250000 ÷ 1048576286.10 GB
- 4Total additional cost286.10 × 0.1028.61
Result Summary
Total additional cost
28.61
A/B Testing Bandwidth Cost Per-User Calculator
The larger variant adds about 286.10 GB and 28.61 in incremental bandwidth cost.
High-volume feature experiment with a 1 MB increase
A service exposes 10% of 8 million users to a feature variant that is 1,024 KB larger.
Input Summary
Control size
600 KB
Variant size
1,624 KB
Test users
8,000,000
Variant allocation
10%
Views per variant user
3
Bandwidth rate
0.08 per GB
Calculation Breakdown
- 1Variant users8000000 × 10%800,000 users
- 2Extra data per user(1624 − 600) × 33,072 KB
- 3Total additional transfer3072 × 800000 ÷ 10485762,343.75 GB
- 4Total additional cost2343.75 × 0.08187.50
Result Summary
Total additional cost
187.50
A/B Testing Bandwidth Cost Per-User Calculator
The test adds about 2,343.75 GB of transfer and 187.50 in the selected currency.
Smaller optimized variant
A team tests a compressed layout that reduces transfer size for 100,000 users.
Input Summary
Control size
500 KB
Variant size
420 KB
Test users
100,000
Variant allocation
50%
Views per variant user
2
Bandwidth rate
0.08 per GB
Calculation Breakdown
- 1Variant users100000 × 50%50,000 users
- 2Additional KB per viewmax(420 − 500, 0)0 KB
- 3Additional transfer per user0 × 20 KB
- 4Total additional cost0 ÷ 1048576 × 50000 × 0.080.00
Result Summary
Total additional cost
0.00
A/B Testing Bandwidth Cost Per-User Calculator
The calculator reports 0 additional GB and 0.00 additional bandwidth cost.
How to Read Your Results
Extra cost per exposed user applies only to users assigned to the variant, not every user in the test.
Extra data per exposed user accumulates the positive page-size difference across expected tested page views.
Total additional bandwidth combines per-user transfer with the estimated number of variant users.
Total additional cost is a transfer-only estimate in the currency used for the bandwidth rate.
Very small per-user amounts can become significant when multiplied by high traffic or repeated views.
Assumptions & Important Notes
- Inputs represent average delivered transfer sizes rather than source-file or uncompressed sizes.
- The entered bandwidth price applies uniformly to all added GB.
- The expected page views per exposed user are representative of actual experiment behavior.
- Only data transfer caused by a larger variant is included.
Related Examples
Frequently Asked Questions
Can I use these examples with another currency?
Yes. Enter a bandwidth rate in your preferred currency per GB, and the cost outputs use that same currency.
Why is the cost per user so small in many examples?
A single user's extra transfer is often a fraction of a GB. Total cost grows when that amount is multiplied across many variant users and views.
What happens if users see the variant multiple times?
Increase views per exposed user. The added KB and estimated cost per exposed user rise proportionally.
Should I include every page on my site?
Include only loads of the tested experience that receive the changed transfer payload.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.