A/B Testing File Size (Monthly) Calculator Examples
See worked examples of how page size, traffic volume, and allocation change monthly A/B test data transfer.
These examples show how to estimate test-period data transfer against an all-control baseline. Each uses 1 GB = 1,000,000 KB and assumes the specified page size is transferred on every tested pageview.
Lighter variant in an even traffic split
A team tests a streamlined product page that is 300 KB smaller than the control in a 50/50 experiment.
Input Summary
Monthly visitors
100,000
Pageviews per visitor
2
Variant traffic share
50%
Control file size
1,500 KB
Variant file size
1,200 KB
Calculation Breakdown
- 1Tested pageviews100,000 × 2200,000
- 2Traffic per experience200,000 × 50%100,000 control and 100,000 variant pageviews
- 3Test transfer(100,000 × 1,500 + 100,000 × 1,200) ÷ 1,000,000270 GB/month
- 4All-control transfer200,000 × 1,500 ÷ 1,000,000300 GB/month
- 5Monthly difference270 − 300-30 GB/month (-10%)
Result Summary
Monthly difference
-30 GB/month (-10%)
A/B Testing File Size (Monthly) Calculator
The test transfers an estimated 270 GB per month instead of 300 GB with all control traffic.
Heavier variant in a limited rollout
A team tests an 1,800 KB interactive variant against a 1,500 KB control, with only 10% of traffic assigned to the variant.
Input Summary
Monthly visitors
250,000
Pageviews per visitor
1.2
Variant traffic share
10%
Control file size
1,500 KB
Variant file size
1,800 KB
Calculation Breakdown
- 1Tested pageviews250,000 × 1.2300,000
- 2Variant pageviews300,000 × 10%30,000
- 3A/B test transfer(270,000 × 1,500 + 30,000 × 1,800) ÷ 1,000,000459 GB/month
- 4All-control transfer300,000 × 1,500 ÷ 1,000,000450 GB/month
- 5Monthly difference459 − 450+9 GB/month (+2%)
Result Summary
Monthly difference
+9 GB/month (+2%)
A/B Testing File Size (Monthly) Calculator
The limited rollout adds an estimated 9 GB of monthly transfer compared with all-control delivery.
Large traffic volume with a compressed variant
A publisher tests a 650 KB variant against a 900 KB control, assigning 75% of tested traffic to the variant.
Input Summary
Monthly visitors
2,000,000
Pageviews per visitor
1.5
Variant traffic share
75%
Control file size
900 KB
Variant file size
650 KB
Calculation Breakdown
- 1Tested pageviews2,000,000 × 1.53,000,000
- 2Experience allocation3,000,000 × 75% and 3,000,000 × 25%2,250,000 variant and 750,000 control pageviews
- 3A/B test transfer(750,000 × 900 + 2,250,000 × 650) ÷ 1,000,0002,137.5 GB/month
- 4All-control transfer3,000,000 × 900 ÷ 1,000,0002,700 GB/month
- 5Monthly difference2,137.5 − 2,700-562.5 GB/month (-20.8%)
Result Summary
Monthly difference
-562.5 GB/month (-20.8%)
A/B Testing File Size (Monthly) Calculator
The experiment is estimated to use 562.5 GB less data per month than an all-control experience.
How to Read Your Results
A negative monthly bandwidth difference means the test uses less estimated data than the all-control baseline.
A positive monthly bandwidth difference means the variant increases estimated transfer at its current traffic share.
A/B test data transfer is the combined total for control and variant traffic during the month.
The percentage change puts the data-volume difference in context relative to all-control transfer.
The estimate concerns the tested pageviews, not necessarily the entire website's bandwidth use.
Assumptions & Important Notes
- Each example treats the entered file size as the average KB transferred per tested pageview.
- One GB equals 1,000,000 KB for all calculations.
- Traffic allocation is assumed to remain constant throughout the month.
- The all-control baseline uses the same visitor and pageview volume as the test.
Related Examples
Frequently Asked Questions
Can I use these examples with a traffic split other than 50/50?
Yes. Replace the variant allocation with your planned percentage and recalculate variant and control pageviews.
Why can a 75% variant split save more data than a 50% split?
When the variant is lighter, assigning more traffic to it applies the file-size saving to more pageviews.
Will a heavier variant always increase monthly transfer?
Compared with an all-control baseline, yes, if it receives any traffic and its average transferred size is larger than the control.
What is a useful file-size input for a page with dynamic assets?
Use an average transferred page size measured across representative pageviews, devices, and content states where possible.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.