A/B Testing Memory Requirement Per-User Calculator Examples
Worked per-user storage examples for lightweight, event-heavy, and profile-heavy A/B testing designs.
These examples show how changes in experiment enrollment, event retention, record size, profile data, and overhead affect estimated storage for a single user.
Lightweight Feature Experiment
A team runs two active feature tests per user and retains five 250-byte events for each test.
Input Summary
Active experiments
2
Events per experiment
5
Event size
250 bytes
Assignment size
80 bytes
Base profile
1,000 bytes
Overhead
20%
Calculation Breakdown
- 1Event data2 × 5 × 2502,500 bytes
- 2Assignment data2 × 80160 bytes
- 3Raw storage1,000 + 2,500 + 1603,660 bytes
- 4Estimated storage3,660 × 1.204,392 bytes (4.29 KB)
Result Summary
Estimated storage
4,392 bytes (4.29 KB)
A/B Testing Memory Requirement Per-User Calculator
Estimated retained storage is 4,392 bytes, or about 4.29 KB per user.
Event-Heavy Product Experimentation
Each user is active in six experiments, with 50 retained events of 750 bytes per experiment.
Input Summary
Active experiments
6
Events per experiment
50
Event size
750 bytes
Assignment size
120 bytes
Base profile
3,000 bytes
Overhead
30%
Calculation Breakdown
- 1Event data6 × 50 × 750225,000 bytes
- 2Assignment data6 × 120720 bytes
- 3Raw storage3,000 + 225,000 + 720228,720 bytes
- 4Estimated storage228,720 × 1.30297,336 bytes (290.37 KB)
Result Summary
Estimated storage
297,336 bytes (290.37 KB)
A/B Testing Memory Requirement Per-User Calculator
Estimated retained storage is 297,336 bytes, or about 290.37 KB per user.
Profile-Heavy Personalization Test
A personalized service stores an 80 KB profile for each user and tracks limited experiment activity.
Input Summary
Active experiments
3
Events per experiment
10
Event size
400 bytes
Assignment size
150 bytes
Base profile
80,000 bytes
Overhead
25%
Calculation Breakdown
- 1Event data3 × 10 × 40012,000 bytes
- 2Assignment data3 × 150450 bytes
- 3Raw storage80,000 + 12,000 + 45092,450 bytes
- 4Estimated storage92,450 × 1.25115,562.5 bytes (112.85 KB)
Result Summary
Estimated storage
115,562.5 bytes (112.85 KB)
A/B Testing Memory Requirement Per-User Calculator
Estimated retained storage is 115,562.5 bytes, or about 112.85 KB per user.
How to Read Your Results
The bytes result is the full estimated retained-storage footprint for one user after overhead.
The KB result uses 1 KB = 1,024 bytes; it is useful for comparing user-level scenarios.
Compare the event-data component with the total to identify whether event retention is the primary driver.
To create a rough dataset estimate, multiply the byte result by a defined user population, then separately account for architecture-specific copies.
Treat results as scenario estimates and validate key assumptions against representative stored records.
Assumptions & Important Notes
- Each example uses one average value for all users and experiments.
- Event records are assumed to be retained independently for each active experiment.
- The overhead allowance covers general storage overhead but not every operational copy of data.
- The figures estimate storage, not service memory or processing capacity.
Related Examples
Frequently Asked Questions
What should I use for average event record size?
Use a measured serialized size from representative retained events where possible, including identifiers, timestamps, properties, and metadata.
Can I use zero stored events per experiment?
Yes. In that case, the estimate consists of profile data, assignment data, and overhead.
Which example is most sensitive to retention changes?
The event-heavy scenario is most sensitive because retained event data is the largest component.
Can I multiply the example result by all registered users?
You can create a rough scenario that way, but use a user count aligned with your retention and enrollment definition.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.