Free Tool · Decision engine
Proxy Selector Wizard
Tell us whether you are running a crawler or a parser, how hard the target fights back and what rotation you need. The wizard scores residential, mobile, ISP and datacenter proxies against your answers and returns a full stack recommendation with cost estimate and matching providers.
Works with every proxy type
Residential, mobile 4G/5G, static ISP, datacenter and IPv6 endpoints are all supported. Every input is saved to the URL — copy it to share the exact result.
Step 1 of 7
What is the job?
Crawlers discover and fetch many URLs; parsers hit a known list and extract fields. They stress a proxy pool in completely different ways.
01 · Overview
What is the Proxy Selector Wizard?
The Proxy Selector Wizard turns the most common question in web scraping — 'which proxies do I actually need?' — into a seven-question decision tree with a scored answer. It asks what you are building (a crawler that discovers URLs, a parser that extracts fields from a known list, session work behind a login, or private API calls), how aggressively the target defends itself, what rotation policy your sessions require, your monthly volume, how price-sensitive you are, how granular your geo targeting must be, and whether you need SOCKS5. Those answers are scored against rotating residential, mobile 4G/5G, static ISP and datacenter proxies. You get a primary tier, a documented fallback tier, the reasoning behind each, a concrete rotation plan with session TTL and concurrency, a monthly bandwidth and cost estimate, a recommended client stack, warnings for combinations that contradict each other, and the providers from our benchmark set that actually sell the tier you were recommended.
02 · How it works
How the tool works
- 01Pick your workload: crawler, parser, logged-in session work, or private/mobile API calls. This sets the baseline response size and reputation profile.
- 02Rate the target's anti-bot posture from open to extreme. Hardened and extreme targets penalise datacenter ASNs heavily and reward residential and mobile exits.
- 03Choose your rotation needs: a new IP per request, sticky 1–30 minute sessions, multi-hour sessions, or fixed IPs that never rotate.
- 04Set monthly volume and budget sensitivity, so the engine can weigh price per GB against measured success rate.
- 05Choose geo granularity and protocol. City, state and ASN targeting pushes the recommendation toward residential; a SOCKS5 requirement filters the provider list.
- 06Read the result: fit score per tier, rotation plan, cost estimate, recommended stack, warnings, and matched providers. Copy the URL to share the exact recommendation with your team.
03 · Use cases
When to use this tool
Sizing a new crawler
Decide before you sign a contract whether a broad discovery crawl belongs on datacenter IPs with a residential escape hatch, or on residential from day one.
Fixing a rising block rate
Re-run the wizard with the target set to hardened and compare the recommended tier against what you are running today.
Choosing rotation for logged-in work
Session workloads on per-request rotation are flagged immediately, with the ISP alternative and a session TTL to use instead.
Budgeting a scraping project
The bandwidth and cost estimate turns a request-count target into a monthly invoice range you can defend in a planning meeting.
Shortlisting providers
Only providers that actually sell the recommended tier — and that speak SOCKS5 if you require it — appear in the match list, each with review and site links.
04 · Pro tips
Best practices
- →Answer for the hardest target in your crawl, not the average one — a single hardened domain dictates the tier you need for that domain, not for all of them.
- →If the wizard recommends residential but 90% of your targets are open, split the crawl: datacenter by default, residential only for the domains that block.
- →Treat the cost estimate as a floor. Retries on failed requests are billed as bandwidth on almost every metered plan.
- →When rotation is sticky, one session id per worker is the rule — sharing a session across workers is what makes providers look unreliable.
- →Re-run the wizard quarterly. Target defences change faster than proxy contracts do.
05 · FAQ
Frequently asked questions
How does the proxy selector wizard choose a proxy type?+
Seven answers are scored against four proxy tiers. Workload (crawler, parser, logged-in session or private API), target hardness, rotation policy, monthly volume, budget sensitivity, geo granularity and protocol each add or subtract points from residential, mobile, ISP and datacenter. The tier with the highest total becomes the primary recommendation and the second-highest becomes the documented fallback, so you can see exactly why one tier beat another instead of getting an unexplained answer.
Should a crawler use residential or datacenter proxies?+
Start a crawler on datacenter proxies when the targets are tolerant — the cost per delivered page is five to ten times lower and discovery crawls burn a lot of requests on pages you never use. Promote a domain to rotating residential the moment its block rate passes roughly 5%. Running the whole crawl on residential because a handful of domains need it is the most common way scraping budgets get wasted.
What proxies do parsers need?+
Parsers hit a known URL list repeatedly, so per-domain reputation matters more than pool breadth. Rotating residential with sticky sessions works for most consumer sites; ISP proxies are better when the parser authenticates, because a stable IP avoids repeated re-verification. Datacenter is fine for open data sources and internal APIs.
How do I decide my rotation needs?+
Match rotation to session lifetime. Stateless requests rotate every call. A pagination run or checkout flow needs sticky sessions of about 10 minutes. Account warm-up and long dashboards need hours, which pushes you off residential peers and onto ISP proxies. Allowlisted integrations need fixed IPs that never rotate.
Is the cost estimate accurate?+
It is an order-of-magnitude planning figure. The wizard multiplies your volume band by a typical response size for your workload, adds 15% protocol overhead and applies the market rate for the recommended tier. It does not model retry waste, failed requests you still pay for, or committed-volume discounts, so treat it as a floor and add 10 to 30% if your block rate is above 5%.
Can I share my recommendation with my team?+
Yes. Every answer is written into the page URL, so copying the address bar reproduces the exact same wizard state and recommendation for anyone who opens it. Nothing is stored on our servers.
07 · Related tools
Continue your audit
Free tool
Bandwidth Cost Calculator
Estimate monthly SOCKS5 bandwidth spend from request volume, average page size, overhead and per-GB pricing.
Free tool
Speed & Availability Checker
Measure latency, jitter, success rate and throughput of any proxy exit node against global endpoints, with configurable test settings and saved run history.
Free tool
cURL Command Generator
Generate copy-paste ready cURL, wget, Python requests and Node fetch commands that route through a proxy.
Also known as
