A/B Testing File Size Calculator Examples
Worked examples showing how file size changes, traffic allocation, and cache hits affect A/B test data transfer.
These examples use different experiment sizes, file-size changes, traffic splits, and cache assumptions. They show the estimated transfer difference relative to serving the control file to every uncached request.
50/50 test of a larger asset
A product team expects 100,000 test visitors, with two file downloads per visitor and a 30% cache hit rate.
Input Summary
Control file size
250 KB
Variant file size
300 KB
Test visitors
100,000
Downloads per visitor
2
Variant traffic share
50%
Cache hit rate
30%
Calculation Breakdown
- 1Total requests100,000 × 2200,000
- 2Uncached requests200,000 × 70%140,000
- 3Variant requests140,000 × 50%70,000
- 4Extra transfer(300 − 250) × 70,000 ÷ 1,048,576+3.34 GB
Result Summary
Total requests
200,000
A/B Testing File Size Calculator
Estimated A/B test transfer is 36.72 GB versus 33.38 GB for an all-control baseline.
25% rollout of a smaller image
An image optimization test uses a 400 KB variant instead of a 500 KB control for 25% of traffic.
Input Summary
Control file size
500 KB
Variant file size
400 KB
Test visitors
50,000
Downloads per visitor
1.5
Variant traffic share
25%
Cache hit rate
40%
Calculation Breakdown
- 1Total requests50,000 × 1.575,000
- 2Uncached requests75,000 × 60%45,000
- 3Variant requests45,000 × 25%11,250
- 4Transfer reduction(400 − 500) × 11,250 ÷ 1,048,576−1.07 GB
Result Summary
Total requests
75,000
A/B Testing File Size Calculator
Estimated A/B test transfer is 20.38 GB versus 21.46 GB for the all-control baseline.
High-traffic test with strong caching
One million visitors request a 210 KB variant instead of a 150 KB control, but 80% of requests are served from cache.
Input Summary
Control file size
150 KB
Variant file size
210 KB
Test visitors
1,000,000
Downloads per visitor
1
Variant traffic share
10%
Cache hit rate
80%
Calculation Breakdown
- 1Total requests1,000,000 × 11,000,000
- 2Uncached requests1,000,000 × 20%200,000
- 3Variant requests200,000 × 10%20,000
- 4Extra transfer(210 − 150) × 20,000 ÷ 1,048,576+1.14 GB
Result Summary
Total requests
1,000,000
A/B Testing File Size Calculator
Estimated A/B test transfer is 29.75 GB versus 28.61 GB for the all-control baseline.
How to Read Your Results
A positive bandwidth difference means the test transfers more data than serving the control to all uncached requests.
A negative result means the variant lowers estimated transfer relative to the control baseline.
The A/B test transfer is the combined transfer for control and variant during the experiment.
The control-only transfer is a comparison baseline, not an additional amount of transfer.
File size difference is per downloaded file request; total impact depends on uncached volume and traffic allocation.
Assumptions & Important Notes
- Examples use the entered file size as the transferred size for each uncached request.
- Requests are distributed according to the stated variant traffic share.
- Cache hits are assumed not to require a new file transfer.
- All GB values are calculated using 1,048,576 KB per GB.
Related Examples
Frequently Asked Questions
Can I model a 90/10 A/B test split?
Yes. Enter 10% as the variant traffic share when 10% of test traffic receives the variant.
How should I estimate downloads per visitor?
Use a representative average from analytics, logs, or a reasoned forecast for the tested asset.
What if my variant is the same size as the control?
The estimated bandwidth difference is zero because the file-size difference per variant request is zero.
Do these examples include cached file transfer?
No. The cache hit rate removes the assumed cached share before transfer is calculated.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.