A/B Testing Website Load Time Per-User Calculator Examples
See worked examples of faster and slower website variants across different traffic splits, user volumes, and page-load frequencies.
These examples show how a per-user page-speed difference can be translated into cumulative waiting time for the users exposed to a test variant. They use consistent control and variant measurements and are illustrative estimates only.
Example 1: Faster product page at a 50/50 split
A retail site tests a lighter product page with 10,000 users and one measured product-page load per user.
Input Summary
Control load time
3.0 seconds
Variant load time
2.4 seconds
Total test users
10,000
Variant traffic share
50%
Average page loads per user
1
Calculation Breakdown
- 1Variant users10,000 * 0.505,000 users
- 2Per-user time saved(3.0 - 2.4) * 10.60 seconds
- 3Load-time change((3.0 - 2.4) / 3.0) * 10020.0%
- 4Total time saved0.60 * 5,0003,000 seconds
Result Summary
Total time saved
3,000 seconds
A/B Testing Website Load Time Per-User Calculator
The faster variant saves an estimated 0.60 seconds per exposed user and 3,000 seconds in total.
Example 2: Faster account dashboard with repeat loads
A software product sends 30% of 24,000 users to a dashboard variant, and each user generates four measured dashboard loads.
Input Summary
Control load time
2.8 seconds
Variant load time
2.5 seconds
Total test users
24,000
Variant traffic share
30%
Average page loads per user
4
Calculation Breakdown
- 1Variant users24,000 * 0.307,200 users
- 2Difference per load2.8 - 2.50.30 seconds
- 3Per-user time saved0.30 * 41.20 seconds
- 4Total time saved1.20 * 7,2008,640 seconds
Result Summary
Total time saved
8,640 seconds
A/B Testing Website Load Time Per-User Calculator
The variant saves an estimated 1.20 seconds per user and 8,640 seconds overall.
Example 3: Slower checkout variant
A checkout variant reaches 40% of 15,000 users, with one measured checkout load per user.
Input Summary
Control load time
1.9 seconds
Variant load time
2.2 seconds
Total test users
15,000
Variant traffic share
40%
Average page loads per user
1
Calculation Breakdown
- 1Variant users15,000 * 0.406,000 users
- 2Per-user time change(1.9 - 2.2) * 1-0.30 seconds
- 3Load-time change((1.9 - 2.2) / 1.9) * 100-15.8%
- 4Total time change-0.30 * 6,000-1,800 seconds
Result Summary
Total time change
-1,800 seconds
A/B Testing Website Load Time Per-User Calculator
The variant adds an estimated 0.30 seconds per user, totaling 1,800 seconds of additional wait time.
Example 4: Small improvement at high traffic
A content site tests a homepage variant with 200,000 users, a 75% variant allocation, and two measured homepage loads per user.
Input Summary
Control load time
2.1 seconds
Variant load time
2.0 seconds
Total test users
200,000
Variant traffic share
75%
Average page loads per user
2
Calculation Breakdown
- 1Variant users200,000 * 0.75150,000 users
- 2Per-user time saved(2.1 - 2.0) * 20.20 seconds
- 3Average test load time(2.1 * 0.25) + (2.0 * 0.75)2.03 seconds
- 4Total time saved0.20 * 150,00030,000 seconds
Result Summary
Total time saved
30,000 seconds
A/B Testing Website Load Time Per-User Calculator
The variant saves an estimated 0.20 seconds per user and 30,000 seconds across variant users.
How to Read Your Results
A positive time-saved result means the variant is faster across the entered number of measured loads per user.
A negative result means the variant adds waiting time relative to the control.
Load-time change is relative to the control, so the same number of seconds can represent different percentages at different baselines.
Average test load time is weighted by the traffic split and does not represent a full-variant rollout result.
Total time saved or added is an exposure estimate, not a measurement of conversion or revenue impact.
Assumptions & Important Notes
- Control and variant performance figures use the same measurement definition.
- Traffic assignment is representative of actual variant exposure.
- The average page-load count reflects the tested page or journey, not unrelated site activity.
- Examples use average values and do not model variation across users or devices.
Related Examples
Frequently Asked Questions
Can I use more than one page load per user?
Yes. Enter the average number of times each user experiences the measured page load during the test period.
Why can total time saved be large when per-user savings are small?
The total multiplies the per-user change by the number of users exposed to the variant.
Does a 75% variant split make the variant faster?
No. It only changes how many users receive the variant and therefore the size of the cumulative estimate.
Should a slower variant always be rejected?
Not from this calculation alone. Consider the full experiment design, measurement quality, and relevant outcome data.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.