Proxy Bandwidth Calculator: GB & Cost per Result

Enter the total traffic for an isolated test window, the number of tasks started and the number that produced usable results. A task can include several requests and retries. Use the same task definition for your next batch.

Traffic per valid result = total pilot traffic ÷ valid results. Multiply that amount by the number of valid results you want next. The calculator assumes the same task mix and failure pattern; changing the destination, concurrency or browser behavior calls for another pilot.

The price field is optional. Enter the marginal traffic rate for the billing unit you select. Estimated usage cost excludes a minimum purchase, unused prepaid traffic, taxes and separate platform fees. Check the current mobile pool pricing and your account quote before buying.

A decimal GB contains 1,000,000,000 bytes. A GiB contains 1,073,741,824 bytes. Mixing these units changes the result. Match both selectors to the meter and rate you are using; a dashboard label alone may be ambiguous. NIST defines the binary prefixes.

Suppose a pilot starts 100 tasks, produces 80 valid results and consumes 0.5 GiB. Those numbers are illustrative. Dividing 0.5 by 80 gives 0.00625 GiB per valid result. At the same measured behavior, 1,000 valid results would use 6.25 GiB, or about 6.71 decimal GB.

If you already measured the complete pilot, its total includes the traffic spent on retries and failures. Adding a second retry multiplier would count that traffic twice. If no task produced a valid result, fix the pilot before projecting a larger run; the calculator leaves the projection undefined.

1. Run one workload in a separate usage window. Record the meter before and after, and wait for usage reporting to settle.
2. Define a valid result before starting: for example, the required fields are present and the requested region is correct. HTTP 200 alone does not establish that.
3. Keep failed tasks in the attempted count and their traffic in the sample. Keep retries within the original task count. A task that succeeds on its third attempt still produces one valid result.
4. Repeat with another small batch. If the results differ, keep separate scenarios instead of averaging away a broken route.

Our pilot measurement guide defines throughput and concurrency. For browser workloads, the Playwright traffic guide helps identify resources to investigate. Browser byte totals and the provider usage meter have different scopes; use the meter for a billing estimate.

Compare the projected usage with the smallest sensible next batch and the available stock in your required country. A larger prepaid bundle may reduce the unit rate while increasing the cash you commit. Keep that purchase decision separate from estimated traffic consumption.

The chart scales your entered sample across three workload sizes. It does not predict improvements, show a confidence interval or claim a typical page size. Download the assumptions with the result so another person can reproduce your estimate.