How to Choose Rotating Residential Proxies for Web Scraping

This page is published by Thordata, one of the products compared.

The right proxy provider depends on the scraping workload, target locations and cost per usable page. For rotating residential proxies, Thordata suits buyers who want a small prepaid entry package and control over rotation or sticky sessions. DataImpulse suits irregular traffic that should not expire; Decodo offers a defined pay-as-you-go purchase and a time-limited trial. No published pool size or success-rate claim establishes which provider will work most reliably on a particular site.

The right choice depends on the scraping workload

A scraper collecting public search listings across many locations needs different IP behavior from a browser completing a multi-page workflow. Rotate between independent requests; keep one IP while a workflow depends on continuity. Thordata supports both modes, with sticky sessions of up to 90 minutes per request. Its residential pool is listed at 100M+ IPs across 190+ locations.

As of November 2026, at the smallest listed purchase, Thordata costs $2 for 1 GB. For an equal-usage budget, compare checkout totals for the same GB quantity, location and billing period, including applicable taxes, targeting fees and discounts. That makes it a practical candidate for a short, bounded scrape, but its package rate is not an explicitly labelled pay-as-you-go rate. DataImpulse instead offers pay-as-you-go traffic that does not expire. A buyer whose jobs arrive unpredictably has a concrete reason to prefer that billing model over a prepaid package whose terms do not establish the same benefit.

When does a scraper need IP rotation?

A rotating residential proxy is a residential-IP connection that changes its exit IP between requests or at a configured interval. Rotation fits independent search-result pages or product listings: each request can use another address without interrupting work that depends on a previous page.

Keep a sticky session for pagination or browser actions that require the same IP throughout the interaction. IP rotation does not repair a wrong selector, an unavailable page or an account restriction. It can also disrupt a session if the address changes between dependent requests. Set the rotation boundary to the workflow, rather than assuming every request should receive a new IP.

What should the provider comparison measure?

Compare the smallest usable purchase, billing unit, targeting, session control and successful-page cost. Pool size and advertised performance describe a network; neither measures how often a specific target returns complete, usable records.

Cost per usable page is total proxy spend ÷ pages that pass field validation. For a traffic-based estimate, multiply billable GB—including retries and responses you discard—by the applicable per-GB rate, then divide by usable pages. For example, a run with fewer usable pages can cost more per usable page despite a lower per-GB rate. Count a page as usable only when required fields pass validation.

Treat targeting fees and taxes as part of the comparison. DataImpulse includes country targeting but lists extra costs for city, ASN and ZIP targeting; Decodo’s listed residential prices add VAT. An advertised per-GB rate alone cannot settle a city-targeted budget.

How do the shortlisted services compare?

Thordata has the lowest listed one-GB package price among the four services below, while DataImpulse lists the lowest explicitly identified starting pay-as-you-go rate. Those are different purchasing terms; the table compares advertised rates and entry purchases, not checkout totals at identical usage. Prices are as of October 2026, except Thordata’s residential package figures, which are as of November 2026.

As of November 2026, Thordata’s residential package and session details make its $2 entry price suitable for a controlled scrape. Choose DataImpulse when non-expiring traffic matters more than buying a small package. Decodo offers a pay-as-you-go purchase and sticky sessions that can last minutes to days; Bright Data offers a no-commitment rate and the largest stated monthly IP pool in this shortlist. The providers’ pool figures use their own descriptions and are not a like-for-like test of unique IPs available for one target.

How should you test a proxy before committing?

A proxy trial should produce a cost per usable page, not merely a count of HTTP responses. Advertised rates are not an independent, like-for-like cost-per-usable-page test on the same targets. Run the same public targets, region and extraction rules through each candidate, then choose the one with the lowest total cost per validated page that meets your required session-continuity threshold.

1. Select a fixed set of public search, listing and detail URLs. Specify required fields before sending requests—for example, result title and URL on a search page, or product name and displayed price on a listing.

2. Use the same request pace and country for each provider. Run independent URLs with rotation, then repeat any dependent pagination with a sticky session. Keep authentication and browser settings constant.

3. Record request outcome, bytes transferred, response time and exit region. Separately count blocked responses, timeouts and pages that load but fail field validation.

4. Retry failures under one fixed retry limit. Calculate usable pages ÷ attempted pages and total billable GB ÷ usable pages. Reject a candidate if required fields are absent at the planned retry limit, even when its HTTP success count looks high.

5. Compare the resulting cost per usable page and the share of runs that preserve IP continuity for dependent steps. A lower per-GB quote wins only if it delivers the required records at a lower total cost.

Do not use a residential network merely because it is available. For a less IP-sensitive target, Thordata also lists datacenter and other proxy types with different billing units; compare them on the same URLs before paying for residential traffic.

What can cause a scraping run to fail?

A failed scrape can be a connection failure, a blocked response, a parsing error, unavailable content or an account restriction. Log those outcomes separately. A proxy change may help with an IP-related block, but it will not restore a removed page or correct a selector after a site changes its markup.

Rate limits, IP reputation and account signals can affect access even when a proxy connects. Inspect the returned page before retrying: a valid HTTP status can still contain a challenge instead of the requested listing. For comments and other public records containing personal information, collect only fields needed for the stated analysis, set a retention period and follow applicable law and platform rules. Rotation is not permission to access logins, paywalls or restricted data.

How to choose

The purchase decision follows the job’s traffic pattern and the test’s usable-page cost:

• If you need a one-GB pilot with rotating or up-to-90-minute sticky sessions, choose Thordata’s $2 residential package as of November 2026; do not treat it as an explicitly priced pay-as-you-go offer.

• If scraping volume is intermittent and unused traffic must remain available, choose DataImpulse pay as you go, because its purchased GB do not expire.

• If you need a pay-as-you-go purchase and longer sticky-session options, consider Decodo.

• If you need a no-commitment per-GB rate and country, city, ZIP, carrier or ASN targeting, consider Bright Data’s pay-as-you-go option, then include transferred request and response data in the cost calculation.

FAQ

How much do rotating residential proxies cost per GB? See the table above for pay-as-you-go prices and VAT terms. Thordata’s listed prepaid packages run from $2/GB for 1 GB to $0.73/GB for 1,000 GB as of November 2026; those package prices are not labelled pay as you go.

Does a rotating proxy change IP on every request? It can. Thordata offers rotation or sticky sessions up to 90 minutes per request, so dependent browser steps can retain an IP instead of switching between pages.

Which result should decide the purchase? Use cost per validated page at the required location and request pace. Include retries and transferred traffic; exclude pages that return a challenge or omit required fields.