Video Bitrate A/B Test Formula
Learn how per-user video data and total delivery volume are calculated for two bitrate variants.
This calculation estimates video payload transferred during an A/B test. It combines expected viewing time, each variant's delivered bitrate, and the traffic split so you can compare per-user usage and total test volume.
- 100% Free
- No Sign-Up Required
- Private & Secure
- Mobile Friendly
Total A/B Test Video Data
Where:
Calculate data use for one viewer in each variant, multiply by the number of viewers assigned to that variant, then add the two totals.
Variables Explained
| Variable | What It Means | Unit |
|---|---|---|
| bitrateA - Variant A bitrate | Average delivered bitrate for Variant A. | kbps |
| bitrateB - Variant B bitrate | Average delivered bitrate for Variant B. | kbps |
| averageWatchMinutes - Average watch time per day | Expected daily viewing time for each participating user. | minutes |
| testDays - Test duration | Number of days the test runs. | days |
| totalUsers - Participating users | Total users included in the experiment. | number |
| trafficToB - Traffic assigned to Variant B | Percentage of participants assigned to Variant B. | percent |
Step-by-Step Calculation
Calculate total watch time per user
Convert daily minutes into seconds and multiply by the test duration.
watchSecondsPerUser = averageWatchMinutes * 60 * testDays
Calculate Variant A data per user
Convert kilobits transferred over the watch time into decimal GB.
dataPerUserA = bitrateA * watchSecondsPerUser / (8 * 1000000)
Calculate Variant B data per user
Apply the same conversion using Variant B's bitrate.
dataPerUserB = bitrateB * watchSecondsPerUser / (8 * 1000000)
Calculate users in each variant
Use the traffic allocation to split the participating audience.
usersInB = totalUsers * trafficToB / 100; usersInA = totalUsers - usersInB
Calculate total test delivery
Add expected video payload for both user groups.
totalTestData = dataPerUserA * usersInA + dataPerUserB * usersInB
Two-week 2.5 Mbps vs 4 Mbps test
Total watch time per user
20 * 60 * 14
16,800 seconds
Variant A data per user
2500 * 16800 / (8 * 1000000)
5.25 GB
Variant B data per user
4000 * 16800 / (8 * 1000000)
8.40 GB
Audience split
10000 * 50 / 100
5,000 users in each variant
Total test data
5.25 * 5000 + 8.40 * 5000
68,250 GB
Final Result
Variant B uses 3.15 GB more per assigned user, and the full A/B test is estimated to deliver 68,250 GB of video data.
Assumptions
- ✓Bitrates are sustained average delivered bitrates while users are watching.
- ✓Every participating user watches the entered average duration on every test day.
- ✓Data volumes are decimal GB, with 1 GB equal to 1,000,000,000 bytes.
- ✓The traffic split remains stable over the test period.
- ✓Only video payload is included.
Limitations
- !Adaptive bitrate streaming can make delivered bitrate differ from the entered average.
- !Actual viewing behavior can vary widely between users and days.
- !Ads, manifests, player assets, analytics, retries, and protocol overhead are excluded.
- !User assignment may not exactly match the intended split at every point in the experiment.
Common Mistakes to Avoid
Entering an encoding ladder maximum instead of the average bitrate actually delivered.
Using total watch minutes for the test as though they were daily watch minutes.
Treating kbps as kilobytes per second; the formula converts bits to bytes by dividing by 8.
Mixing decimal GB with GiB when comparing the estimate with reporting tools.
Assuming a 50/50 split lowers the data used by a Variant B user; it only changes aggregate volume.
Related Formulas
Frequently Asked Questions
How do you calculate video data use from bitrate?
Multiply bitrate in kbps by watch time in seconds, then divide by 8,000,000 to express the result in decimal GB.
Why does the formula divide by 8,000,000?
It divides by 8 to convert bits to bytes and by 1,000,000 to convert kilobytes to decimal GB within this bitrate-based calculation.
Does a higher bitrate always use more data?
For the same delivered viewing time, a higher average bitrate produces proportionally more video payload.
Does the traffic split affect data per user?
No. It affects the number of users receiving each variant and therefore the total test volume.
Ready to calculate your result?
Use the calculator to get instant results with your own inputs.