A/B Testing Database Storage Calculator Examples
Worked examples show how experiment traffic, event volume, replication, overhead, and retention affect A/B testing database storage.
These examples show how different A/B testing programs can generate very different storage volumes. Each calculation includes experiment participation, records per participant, average record size, replication, and database overhead.
Lean experiment tracking for a product launch
A product team runs a launch experiment for 25,000 monthly visitors. Eighty percent participate, and each participant produces six compact records.
Input Summary
Monthly visitors
25,000 visitors
Experiment traffic
80%
Records per experiment visitor
6 records
Average record size
0.25 KB
Replication factor
2 copies
Index and metadata overhead
20%
Retention period
6 months
Calculation Breakdown
- 1Experiment visitors25,000 × 0.8020,000 visitors
- 2Tracked records20,000 × 6120,000 records
- 3Raw storage120,000 × 0.25 KB30,000 KB
- 4Storage with replicas and overhead30,000 × 2 × 1.2072,000 KB
- 5Monthly storage72,000 ÷ 1,048,5760.07 GB
- 6Six-month retained storage0.0687 × 60.41 GB
Result Summary
Six-month retained storage
0.41 GB
A/B Testing Database Storage Calculator
The program generates about 0.07 GB per month and about 0.41 GB across six retained months.
Standard website A/B testing program
A website receives 100,000 monthly visitors, enrolls half in experiments, and stores 10 records per participant.
Input Summary
Monthly visitors
100,000 visitors
Experiment traffic
50%
Records per experiment visitor
10 records
Average record size
0.5 KB
Replication factor
3 copies
Index and metadata overhead
30%
Retention period
12 months
Calculation Breakdown
- 1Experiment visitors100,000 × 0.5050,000 visitors
- 2Tracked records50,000 × 10500,000 records
- 3Raw storage500,000 × 0.5 KB250,000 KB
- 4Storage with replicas and overhead250,000 × 3 × 1.30975,000 KB
- 5Monthly storage975,000 ÷ 1,048,5760.93 GB
- 6Annual retained storage0.9298 × 1211.16 GB
Result Summary
Annual retained storage
11.16 GB
A/B Testing Database Storage Calculator
The program generates about 0.93 GB monthly, or about 11.16 GB when 12 monthly cohorts are retained.
High-traffic experimentation with rich event tracking
Two million monthly visitors generate rich experiment data for 35% of the audience, including 24 records per participant.
Input Summary
Monthly visitors
2,000,000 visitors
Experiment traffic
35%
Records per experiment visitor
24 records
Average record size
0.8 KB
Replication factor
3 copies
Index and metadata overhead
40%
Retention period
18 months
Calculation Breakdown
- 1Experiment visitors2,000,000 × 0.35700,000 visitors
- 2Tracked records700,000 × 2416,800,000 records
- 3Raw storage16,800,000 × 0.8 KB13,440,000 KB
- 4Storage with replicas and overhead13,440,000 × 3 × 1.4056,448,000 KB
- 5Monthly storage56,448,000 ÷ 1,048,57653.83 GB
- 618-month retained storage53.8330 × 18968.99 GB
Result Summary
18-month retained storage
968.99 GB
A/B Testing Database Storage Calculator
The rich tracking program produces about 53.83 GB per month and nearly 969 GB across 18 retained months.
Full-population feature experiment with detailed telemetry
A feature team includes 400,000 monthly visitors in an experiment and captures 50 records per visitor for detailed behavior analysis.
Input Summary
Monthly visitors
400,000 visitors
Experiment traffic
100%
Records per experiment visitor
50 records
Average record size
1.2 KB
Replication factor
3 copies
Index and metadata overhead
50%
Retention period
3 months
Calculation Breakdown
- 1Experiment visitors400,000 × 1.00400,000 visitors
- 2Tracked records400,000 × 5020,000,000 records
- 3Raw storage20,000,000 × 1.2 KB24,000,000 KB
- 4Storage with replicas and overhead24,000,000 × 3 × 1.50108,000,000 KB
- 5Monthly storage108,000,000 ÷ 1,048,576103.00 GB
- 6Three-month retained storage102.9968 × 3308.99 GB
Result Summary
Three-month retained storage
308.99 GB
A/B Testing Database Storage Calculator
The program generates about 103.00 GB each month and about 308.99 GB with three months of retention.
How to Read Your Results
Monthly storage is the estimated capacity generated by one month of experiment data after replicas and overhead are included.
Retained storage estimates the accumulated data held online for the selected retention period at a steady monthly rate.
Raw storage is useful for understanding the underlying record volume, but it is not the full database capacity requirement.
A higher replication factor increases storage because each additional copy stores the same underlying data.
Use the result as a planning baseline and separately account for backups, logs, temporary migration space, and growth headroom.
Assumptions & Important Notes
- Each example assumes consistent traffic and tracking behavior for every month in the retention period.
- The stated average record sizes are pre-replication and pre-index-overhead values.
- Replication factors include the primary database copy.
- The examples do not subtract storage savings from compression, aggregation, archival, or early deletion.
- All GB results use 1 GB equal to 1,048,576 KB.
Related Examples
Frequently Asked Questions
Can I use these examples to estimate feature-flag storage?
Yes, if you enter the visitor volume, evaluation or event records per visitor, average stored record size, replicas, and overhead for your feature-flag data.
Why do the examples include an overhead percentage?
Stored database capacity commonly includes indexes, metadata, and internal structures in addition to the raw record fields.
Which input has the biggest effect on storage?
The largest driver depends on the setup, but records per visitor, record size, replication, and experiment traffic can each materially change the result.
Should I multiply the result by 12 for annual data generation?
Yes, multiplying monthly storage by 12 estimates annual generation if monthly traffic and tracking remain stable. Retained storage may differ if retention is shorter or longer than 12 months.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.