Do proxy prices drop at volume, and by how much?

Yes, published unit prices often fall at larger tiers, but the measured discount is ladder-specific. GeoNode residential falls from $0.792 per GB at 10 GB to $0.50 per GB at 1,000 GB, a 36.9% unit-price reduction. Bright Data shared datacenter falls from $1.40 per IP at 10 IPs to $0.90 per IP at 1,000 IPs, a 35.7% reduction. Both ladders were observed on July 25, 2026. GeoNode residential pricing, retrieved 2026-07-25T15:42:45Z · Bright Data datacenter pricing, retrieved 2026-07-25T15:31:32Z

Volume discounts are real in the dataset, but “bulk is cheaper” is incomplete. The unit rate can fall while the checkout total rises sharply. The discount only becomes economic value when the buyer uses the purchased traffic or addresses during the applicable period.

The safest calculation holds provider, product family, billing unit, and commitment constant. It then compares published rungs on that one ladder. Crossing providers would mix a volume effect with network, sharing, geography, expiry, and commercial differences.

GeoNode residential traffic

Every rung comes from geonode-residential-monthly. The vendor records a monthly commitment, while purchased traffic is explicitly non-expiring. GeoNode residential pricing, retrieved 2026-07-25T15:42:45Z

The unit discount is substantial, but the cash requirement changes too. Moving from the 10 GB rung at $7.92 to the 1,000 GB rung at $500 multiplies the checkout by about 63.1 while reducing the unit rate by 36.9%. GeoNode residential pricing, retrieved 2026-07-25T15:42:45Z

Because the recorded traffic does not expire, unused volume is not lost at the end of the billing month. That makes the larger tier easier to justify for a buyer with irregular but eventually sufficient usage. It still ties up more cash, and the dataset does not assign a value to that timing cost.

Bright Data shared datacenter IPs

These are four separate monthly rows in the same shared datacenter family. The recorded commitment resets each month and does not roll over. Bright Data datacenter pricing, retrieved 2026-07-25T15:31:32Z

The larger rate cannot be detached from its quantity. Pricing a 10-IP requirement at the 1,000-IP rate would produce a $9 result, but the published 10-IP checkout is $14. The volume discount belongs to the full package. Bright Data datacenter pricing, retrieved 2026-07-25T15:31:32Z

This ladder also shows why per-item prices should be reported as a range. Across these four observed rungs, the unit rate is $0.90 to $1.40 per IP, attached to monthly checkouts from $14 to $900. Stating only the low endpoint would hide the quantity required to obtain it. Bright Data datacenter pricing, retrieved 2026-07-25T15:31:32Z

Oxylabs shows a different discount curve

Oxylabs shared datacenter rows publish $1 per IP at 100 IPs, $0.75 per IP at 1,000 IPs, and $0.70 per IP at 3,000 IPs. The unit reduction is 25% by 1,000 IPs and 30% by 3,000 IPs. The attached monthly checkouts are $100, $750, and $2,100. Oxylabs datacenter pricing, retrieved 2026-07-25T15:31:32Z

The last step has diminishing unit savings. Moving from 100 to 1,000 IPs removes one quarter of the unit rate. Moving from 1,000 to 3,000 removes another 6.7% relative to the 1,000-IP rate. A buyer should calculate the next rung against actual demand, not assume every quantity increase produces the same percentage.

A lower unit rate can still cost more per used unit

Suppose the buyer needs 600 IPs. Buying a 1,000-IP package and dividing by all included addresses gives the published unit rate. Dividing the same checkout by the 600 addresses actually needed gives a higher effective rate. The second denominator is the one that belongs in the buyer’s budget.

Traffic plans work the same way. Use:

effective cost per used GB = checkout total / GB expected to be consumed before expiry

For per-IP plans, use:

effective cost per used IP = checkout total / assigned IPs actually required

Do not assume the unused portion can be carried forward. If rollover on the provider’s page or expiry is null, the term is not published. If the row says the capacity resets, unused capacity has no value in the next period.

When a volume discount is worth taking

First, measure demand at the same unit the plan sells. Then compare the current rung with the next published rung. Record four outputs: percentage reduction in unit rate, increase in checkout cash, unused quantity, and the time available to consume it.

A larger tier is economically better only when the value of the used capacity exceeds the extra checkout and any expiry risk. The dataset can compute the price side. It cannot predict future demand or establish that more addresses improve target success.

The measured answer is therefore a range of real curves, not one market discount. The two clean examples here reduce unit price by 35.7% to 36.9% between their entry comparison and 1,000-unit rung. Oxylabs reaches a 30% reduction between 100 and 3,000 IPs. A different provider, proxy type, commitment, or quantity must be recomputed from its own published ladder.