A/B Testing Database Storage Calculator Examples
Explore worked annual storage examples for small, growing, and high-volume A/B testing programmes.
These examples show how traffic, experiment frequency, event detail, and database overhead change annual A/B testing storage estimates. Each result is a capacity-planning estimate for primary database storage.
Small product team with lightweight event tracking
A product team has 20,000 monthly visitors, runs 6 tests each year, includes 40% of traffic, and retains 10 events per participant.
Input Summary
Monthly visitors
20,000
Tests per year
6
Traffic included per test
40%
Events per participant
10
Average event size
0.5 KB
Assignment record size
150 bytes
Average variants
2
Storage overhead
30%
Calculation Breakdown
- 1Annual visitors20,000 × 12240,000 visitors
- 2Participant records240,000 × 6 × 0.40576,000 records
- 3Event records576,000 × 105,760,000 events
- 4Raw storage2.949 GB events + 0.086 GB assignments + 0.0006 GB metadata3.036 GB
- 5Annual storage3.036 × 1.303.9 GB
Result Summary
Annual storage
3.9 GB
A/B Testing Database Storage Calculator
Estimated annual primary database storage is about 3.9 GB.
Growing SaaS experimentation programme
A SaaS company has 100,000 monthly visitors, runs 12 tests per year, includes 50% of traffic, and stores 20 one-KB events per participant.
Input Summary
Monthly visitors
100,000
Tests per year
12
Traffic included per test
50%
Events per participant
20
Average event size
1 KB
Assignment record size
200 bytes
Average variants
3
Storage overhead
40%
Calculation Breakdown
- 1Annual visitors100,000 × 121,200,000 visitors
- 2Participant records1,200,000 × 12 × 0.507,200,000 records
- 3Event records7,200,000 × 20144,000,000 events
- 4Raw storage147.456 GB events + 1.440 GB assignments + 0.002 GB metadata148.898 GB
- 5Annual storage148.898 × 1.40208.5 GB
Result Summary
Annual storage
208.5 GB
A/B Testing Database Storage Calculator
Estimated annual primary database storage is about 208.5 GB, or 0.21 TB.
High-traffic programme with detailed events
A consumer platform has 2 million monthly visitors, runs 24 tests yearly, includes 60% of traffic, and retains 30 events of 2 KB per participant.
Input Summary
Monthly visitors
2,000,000
Tests per year
24
Traffic included per test
60%
Events per participant
30
Average event size
2 KB
Assignment record size
250 bytes
Average variants
4
Storage overhead
50%
Calculation Breakdown
- 1Annual visitors2,000,000 × 1224,000,000 visitors
- 2Participant records24,000,000 × 24 × 0.60345,600,000 records
- 3Event records345,600,000 × 3010,368,000,000 events
- 4Raw storage21,233.664 GB events + 86.400 GB assignments + 0.005 GB metadata21,320.069 GB
- 5Annual storage21,320.069 × 1.5031,980.1 GB
Result Summary
Annual storage
31,980.1 GB
A/B Testing Database Storage Calculator
Estimated annual primary database storage is about 31.98 TB.
How to Read Your Results
Annual event records show the expected number of retained experiment-related event rows, not all product analytics events.
Storage before overhead is the estimated payload, assignment, and metadata requirement before local database costs.
Estimated annual database storage adds the configured allowance for indexes, row overhead, and capacity headroom.
The TB figure uses decimal units: 1 TB equals 1,000 GB.
Treat the result as primary database capacity only unless backup and replica storage is explicitly added elsewhere.
Assumptions & Important Notes
- Visitor volume is steady enough for the monthly average to represent the year.
- Every participant produces the stated average number of retained events for each test entered.
- A visitor can appear in more than one experiment during the year.
- The selected overhead represents local database overhead rather than independent backup copies.
Related Examples
Frequently Asked Questions
Can I use these examples for feature flag experiments?
Yes, if the feature flag system stores assignment or exposure records and related events in a database. Adjust inputs to reflect the records you retain.
Why does the high-traffic example use terabytes?
It combines a large annual audience, many test exposures, 30 events per participant, and 2 KB event payloads. Those factors create billions of event records.
What input usually reduces storage most?
For event-heavy programmes, reducing the number of retained events per participant or the average stored event size often has the largest direct effect.
Should I use average or peak monthly visitors?
The calculator is designed for an annual estimate using an average. Peak periods may still matter when planning active capacity and ingestion throughput.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.