Proxies are sold under two different models: a dedicated address billed monthly at a fixed price, or a residential or mobile channel billed per gigabyte with a new address per session. Mixing these models in one calculation makes no sense — the question isn't "which channel is cheaper" but "what will my specific monthly load cost under each model." Here is the formula for both schemes and three ready-made scenarios to calculate for your own task instead of guessing.
The formula for a dedicated address
The monthly cost of a dedicated datacenter or ISP address is a fixed fee per IP multiplied by how many addresses you need at once. Traffic within the plan is usually not tightly capped, or the cap is large enough that it doesn't matter for most automation. Formula: address_price × number_of_addresses × number_of_months. The only variable that actually moves the budget is how many parallel addresses you need, not the traffic volume.
The formula for a gigabyte package
Residential and mobile traffic is calculated differently: page_size × number_of_requests × retry_coefficient. The retry coefficient is mandatory, not optional: some sessions drop, some requests get blocked and need a retry with a new address. Without a 15–25% buffer over the "ideal" volume, a traffic budget almost always ends up underestimated. Reference figures: a text page is 0.3–1 MB, a product card with images is 2–5 MB, an autoplay video feed eats tens of MB per minute.
Calculation 1: light monitoring
A hundred text pages a day, no images, checked once daily. Base volume: 100 pages × 0.7 MB ≈ 70 MB a day, about 2.1 GB a month. With a buffer for retries and overhead — 2.5–3 GB a month. For this load a dedicated datacenter address is almost always cheaper than a traffic package, unless the task specifically requires a residential origin: the gap between per-gigabyte pricing and a flat fee favors the dedicated channel.
Calculation 2: medium automation
A thousand product cards a day with images, crawled once daily. Base volume: 1,000 × 3.5 MB ≈ 3.5 GB a day, about 100–105 GB a month at continuous operation — but such a volume is more often checked a few passes a week rather than daily, giving 20–40 GB a month depending on frequency. Add a 20% buffer for retries. Here a residential traffic channel usually fits the budget more precisely than several dedicated addresses, because the load is uneven and grows with the catalog.
Calculation 3: heavy catalog parsing
Ten thousand cards with a full set of images, one full pass a week plus targeted rechecks. Base volume of one full pass: 10,000 × 4 MB ≈ 40 GB. At four passes a month — about 160 GB, plus a 20–25% buffer for retries and heavier-media pages, roughly 190–200 GB a month. At this volume it matters not just how many gigabytes but how many parallel threads you run: without enough parallelism a full pass stretches over days and the data goes stale faster than it's collected.
Where a dedicated address beats a package
The break-even point depends on price per gigabyte and how geo-sensitive the task is. If geo-sensitivity isn't critical — calls to an open API, uptime monitoring, downloading documents — the comparison is straightforward: divide the monthly fee for a dedicated address by the package's price per gigabyte to get the threshold above which the dedicated channel is cheaper. For loads below a few gigabytes a month per address, a package is often cheaper; for constant round-the-clock load without rotation, a dedicated channel is almost always more economical. On where saving on channel type turns out illusory, see cheap versus expensive proxies.
Where you cannot cut corners
Don't budget traffic right at the edge with no buffer for retries: a session broken by a timeout or a blocked address costs more than the gigabytes "saved" on a minimal package. Don't cut costs on channel type where a platform cross-checks ASN and reputation — a cheap datacenter IP blocked on the fifth request costs more than a solid residential one. Savings come from disabling images, fonts and trackers in automation, not the cheapest traffic seller. On verifying a channel delivers promised quality, see how to check proxy quality.
Frequently Asked Questions
How do I budget for retries if the failure rate isn't known in advance?
Start with a 20% buffer over the calculated volume and adjust after the first month: measure the actual ratio of successful requests to total attempts and recalculate the buffer for your task.
Can dedicated addresses and a traffic package be mixed in one project?
Yes, that's common practice: tasks without geo-sensitivity move to dedicated addresses at a fixed price, while the geo-sensitive part stays on a traffic package. Such a split usually cuts the overall bill more than one channel for everything.
What if actual traffic use is consistently higher than the calculation?
Check whether extra resources load — images, fonts, analytics scripts — that the task doesn't need. Then check the retry share: above 25–30%, the problem is more often address quality or retry logic than page volume.
Current rates for every channel and a volume calculator live in the proxy section.