A/B Testing Database Storage (Annual) Calculator
Estimate the annual database storage needed for A/B test participant assignments, tracked events, indexes, and test metadata.
Overview
This A/B testing database storage calculator estimates the annual space needed to retain experiment assignments and event data. Enter your visitor volume, expected number of tests, test traffic allocation, event volume, typical event size, and database overhead allowance.
How it works
The calculator estimates annual visitors, then multiplies that volume by the number of tests and the traffic share used in each test to estimate experiment participant records. It multiplies participant records by events per participant and average event size to estimate event payload storage. Assignment records and a small amount of per-variant metadata are added, then the selected overhead percentage is applied for indexes and other database costs.
How to use this calculator
- 1Enter your average number of unique visitors per month.
- 2Add the number of experiments expected to run during the year.
- 3Set the typical share of traffic included in each experiment.
- 4Enter the average number and size of events saved per participant.
- 5Add an allowance for indexes, row overhead, and database headroom.
- 6Review the estimated annual storage requirement in GB and TB.
Example Calculation
Monthly visitors
100000
A/B tests per year
12
Traffic included per test
50%
Average variants per test
3
Events per participant
20
Average event size
1
Assignment record size
200
Database storage overhead
40%
Estimated annual database storage
208.5 GB
With 100,000 monthly visitors, 12 tests per year, 50% test traffic, and 20 one-KB events per participant, the estimate is about 204 GB of annual database storage after a 40% overhead allowance.
Frequently asked questions
What does this A/B testing storage calculator estimate?
It estimates annual database storage for experiment participant assignments, tracked events, basic test metadata, and a configurable database overhead allowance.
Why does the number of A/B tests affect storage?
A visitor can be included in more than one experiment over a year. Each test can create additional assignment records and event records tied to that experiment.
Should I count the control as a variant?
Yes. For example, a control plus two treatments should be entered as three variants. Variant count mainly affects the small metadata estimate in this calculator.
What should I use for average event size?
Use the typical stored size of an event payload, including its properties. A small event may be around 1 KB, while events with many attributes can be considerably larger.
Does the estimate include indexes and database overhead?
Yes, the final result applies your chosen overhead percentage. Adjust it to reflect your database design, indexing strategy, and desired headroom.
Does this include backups and replicas?
No. The estimate is for primary database capacity. Plan additional storage separately for backups, replicas, write-ahead logs, exports, and disaster recovery.
Explore Related Calculators
Assumptions and warnings
Assumptions
- Monthly visitor volume remains broadly consistent throughout the year.
- Each included visitor generates one assignment or exposure record for each test they enter.
- The events-per-participant figure represents only events retained in the database for a given test.
- The storage overhead percentage covers indexing, row overhead, and capacity headroom, but not separate backup or disaster-recovery copies.
- Storage is shown in decimal GB and TB, where 1 TB equals 1,000 GB.
Warnings
- This is a capacity-planning estimate only; actual storage can differ because of event schemas, compression, retention rules, partitioning, replication, and database engine behavior.
- Add separate capacity for backups, replicas, logs, and non-experiment analytics data if they are stored outside the estimated overhead.