Proxies for Price Monitoring: Choosing a Proxy Type
Pick a proxy type for price monitoring by market and refresh rate: residential for regional prices, datacenter for volume, and catching wrong-market rows.
Choose the proxy type from two facts about your catalog: which markets the prices come from, and how often you refresh them. Prices that vary by country, region or ZIP code need exits in those places, which means residential proxies outside the US, Italy and Spain. A catalog in a single market that you refresh many times a day is cheaper on rotating datacenter proxies, billed either per gigabyte or by concurrent connections. Node4 sells each of these, so this is a guide to choosing among proxy types, not a ranking of providers.
A wrong-market price is worse than a missing one
A blocked request is easy to spot. A price from the wrong market looks normal and goes straight into the chart. For example, suppose a collector stores this record:
sku: KETTLE-2L-STEEL
retailer: example-retailer.co.uk
requested market: GB
exit reported: GB, Manchester
currency shown: USD
delivery region: "Deliver to United States" (cookie from an earlier run)
price parsed: 34.99
status: REJECT, currency and delivery region do not match GBThe proxy did its job here: the exit is in Britain. The retailer still showed US prices because the session carried a delivery cookie from a previous run. So before you store a price, check three fields against the market you asked for: the exit country (from an IP lookup through the same session), the currency symbol or code on the page, and the delivery region the site has selected. Clear cookies between markets, or keep one cookie jar per market, and log all three fields beside the price. The market research guide covers that acceptance rule in more depth.
Proxy types for price monitoring, compared
| Proxy type | Fits | Markets with a local exit | Billed by | Main limit |
|---|---|---|---|---|
| Residential | Regional and ZIP-level pricing, sites that refuse hosting traffic | 170+ countries; country, state and city targeting | Gigabyte | Partner-network devices; slower; exits can drop |
| Rotating Datacenter | Catalog sweeps a few times a day | US, Italy, Spain (Node4-owned blocks) | Gigabyte | Hosting-network addresses; three countries |
| Rotating Unmetered | Continuous refresh, many workers | US, Italy, Spain (Node4-owned blocks) | Concurrent connections | Same exits as Rotating Datacenter |
| Dedicated datacenter | A small list checked from fixed addresses | US, Italy, Spain (Node4-owned blocks) | Proxy per month, no data cap | Fixed addresses build a history on each site |
| Shared datacenter | Proving the scraper before you buy volume | US, Italy, Spain (Node4-owned blocks) | Proxy per month, pooled data | Up to five customers per address |
HTTP or SOCKS5 is a protocol choice, not a separate proxy type: every paid Node4 product accepts both, so pick whichever your HTTP client or browser driver supports. The trade-offs are in HTTP vs SOCKS5 proxies.
Working out the cost of one refresh cycle
Price monitoring has a fixed shape: catalog size times refresh rate times page weight. That arithmetic decides the billing model better than any feature list.
For example, take 5,000 products refreshed four times a day, with pages averaging 1.5 MB:
- 5,000 x 4 = 20,000 page loads a day
- 20,000 x 1.5 MB = 30 GB a day, roughly 900 GB a month
On a per-gigabyte product that volume sets the plan size. Two changes cut it hard. Blocking images, fonts and third-party scripts often brings a product page well under half its full weight. Calling the retailer's own JSON endpoint for price and stock, where the page uses one, can cut it further still.
The same job on Rotating Unmetered is sized by connections, not data. If each page takes 2 seconds and you want the 20,000 loads spread evenly across the day, you need only one or two connections; if you want each 5,000-product sweep finished in 10 minutes, you need about 5,000 x 2 / 600, or roughly 17 connections, plus headroom for retries. Connection pricing wins when sweeps are fast and frequent; per-gigabyte pricing wins when the catalog is small or pages are light. The proxy calculator runs this for your own numbers.
When each type earns its place
Residential is the type in this guide that covers markets outside the US, Italy and Spain, and the safer one for retailers that block hosting networks outright. Exits come from a partner network of consumer devices, not Node4's blocks, and target by country, state or city in the username, as described in country and city targeting. Sessions are best effort, with a 30-minute ceiling, so a multi-step flow (add to cart, apply a coupon, view the total) should check the exit at the end and restart if it changed. See residential plan prices.
Rotating Datacenter suits a US, Italian or Spanish catalog on retailers that do not block hosting traffic. Each new proxy connection gets a new exit from Node4-owned blocks, and a session ID holds one exit for up to 30 minutes counted from first assignment, enough for a cart-total check. See rotating plan prices.
Rotating Unmetered is the same exits billed by concurrent connections, for pipelines that refresh around the clock. See connection-based pricing.
Dedicated and shared datacenter proxies fit a short list of pages checked on a steady schedule, or testing a scraper before you buy volume. Shared addresses are used by up to five customers, so a site may already know them.
Which type to buy
If your prices differ by country or region outside the US, Italy and Spain, buy residential and log exit, currency and delivery region with every price. If your catalog sits in those three markets, start on Rotating Datacenter, and switch to Rotating Unmetered when the cycle arithmetic above shows connection pricing beating per-gigabyte pricing. For the mechanics of rotation and retries, see rotating proxies for web scraping; for Amazon specifically, see proxies for Amazon scraping.