PhantomHow the bandwidth estimate works
The calculator converts each response size to megabytes, multiplies it by request count, and adds the percentage you enter for retries and overhead. It uses decimal units: 1,000KB equals 1MB and 1,000MB equals 1GB.
The recommendation checks the exact unrounded estimate against the package data used by the store. If the estimate exceeds the largest package, the calculator says so instead of inventing another tier.
Proxy bandwidth examples: requests to GB
The examples below include a 20% allowance. Use the transferred size of the whole task: a browser page may include many requests for images, scripts, and fonts, while an API call may return only a small JSON response.
| Requests or page loads | Average transferred size | Base traffic | With 20% allowance |
|---|---|---|---|
| 10,000 API requests | 100 KB | 1 GB | 1.2 GB |
| 5,000 HTML requests | 400 KB | 2 GB | 2.4 GB |
| 1,000 browser page loads | 3 MB | 3 GB | 3.6 GB |
| 10,000 large responses | 1 MB | 10 GB | 12 GB |
These are assumed workloads, not measured averages for every website. For example, 10,000 requests at 100KB each transfer 1GB before overhead; entering 20% raises the planning estimate to 1.2GB.
How to measure average response size
- Run a small sample with the same HTTP client or browser planned for production.
- Record transferred bytes, not only the uncompressed document or response body.
- Include representative successes, redirects, error pages, and downloads.
- Divide total transferred data by completed sample requests.
- Enter a retry allowance based on observed failures and normal request overhead.
Browser automation often transfers far more than a direct API request because supporting assets load through the proxy. If you block images, fonts, or media in the sample, only do so when the real workflow will use the same rules.
How many requests can I make with 1GB of proxy data?
Divide 1,000,000KB by your average transferred KB per request. At 100KB per request, 1GB covers approximately 10,000 requests before retries and overhead. With a 20% allowance, budget for roughly 8,333 requests. A 1MB response uses ten times as much data as a 100KB response.
Full browser page loads can download multiple assets, so the number of page visits can be much lower than the number of individual HTTP requests. Measure the complete page load if page visits are what you want to estimate.
Does the calculator estimate proxy cost?
It shows the price of the smallest listed PhantomProxies package large enough for your estimated traffic. This is a one-time package price, not a quote for exactly the GB you enter. Availability is checked at checkout, and actual traffic can differ from the estimate.
See current residential proxy pricing to compare package sizes before buying.
Do rotating or sticky sessions change the calculation?
Both modes transfer data through the proxy. Estimate the traffic your workflow actually sends, including any repeated navigation or retries. Changing the session mode alone does not specify a fixed bandwidth saving.
Compare sticky and rotating proxy sessions to choose a mode before measuring a representative sample.
What the estimate does not guarantee
Actual metering can vary with headers, uploads, redirects, compression, retries, and upstream accounting. The estimate does not promise a number of successful results or uninterrupted routes. Validate it by comparing a small run with the balance shown in your dashboard.
Continue with the detailed bandwidth planning guide, review workflow-specific use cases, or use the setup documentation to configure a test connection.