Category: Proxy Use Case Guides

Best Proxies for Price Monitoring 2026: Tools, Setup, Anti-Block Tips

residential rotating proxies are the best choice for price monitoring across most retailers in 2026. ISP proxies work well when you need stable session IPs for cart-flow scraping, and mobile is overkill for static price pages. avoid datacenter for any retailer running modern bot protection (Akamai, Imperva, DataDome, PerimeterX), which is now most of them.

this guide ranks the best providers by retailer compatibility, walks through the setup, and covers the anti-block tactics that actually work.

why price monitoring is harder than it looks

price scraping triggers bot detection harder than most use cases because:

prices update frequently, so you need to recrawl often
e-commerce sites invest heavily in WAF/anti-bot to protect dynamic pricing strategies
catalogs are huge (millions of SKUs at Amazon, Walmart, Booking.com)
many sites geo-vary prices, so you need IPs in the target market
A/B tests and personalization mean two requests can return different prices

this is why proxy choice matters more here than for, say, a one-off competitive analysis.

quick picks by retailer category

category	example sites	best proxy type
general e-commerce	Amazon, Walmart, Best Buy	residential rotating
travel and hospitality	Booking, Expedia, Airbnb	residential, geo-precise
airlines and OTAs	Kayak, Skyscanner	mobile or residential, sticky
marketplaces	eBay, AliExpress	residential rotating
local retailers	Target, Tesco, regional grocers	residential, country-specific
sportsbooks	Bet365, DraftKings	residential, geo-locked, sticky
luxury and DTC	Net-a-Porter, brand sites	residential or ISP, low rotation

ranked: best providers for price monitoring 2026

1. Singapore Mobile Proxy — Best Mobile + Own-Hardware SG Carrier IPs

Overview: Singapore Mobile Proxy runs an in-house phone farm in Singapore with real SIM cards on Singtel, StarHub, M1, and Vivifi. Every port maps to a dedicated physical modem, so the IP you rotate to is a genuine carrier IP, not a reseller pool. They are not the biggest, but for mobile + Asia + account-warming workloads they are the only provider running their own hardware end-to-end.

Proxy Pool: 100+ live mobile devices in Singapore, with active expansion into Malaysia and Indonesia. Each port is one modem on a real 4G/5G SIM. Carriers covered: Singtel, StarHub, M1, Vivifi. Rotation is per-modem (not pool-shared) so subnets rotate within a single carrier ASN.

Key Features:

Dedicated mobile ports — one modem, one SIM, one IP per port. No pool sharing.
API rotation — programmatic IP rotation via REST endpoint or token URL
Cloudfone integration — cloud-hosted Android phones (cloudf.one) with the same SIM for app-level workflows
Singapore IPs at scale — only provider with this much SG mobile inventory; useful for SEA market research
Honest geo — IPs are actually in Singapore, not VPN-routed

Performance: Per-port bandwidth scales with the underlying carrier (typically 30–100 Mbps on 5G). Rotation latency is ~3 seconds (carrier reconnect time). Success rate on social platforms is consistently above 98% because each port is a real consumer device.

Pricing:

Plan	Price	Type	Minimum
Single Port	$40/mo	1 dedicated mobile port, unlimited bandwidth	1 month
Trial	Free 24h	1 port full access	None
6-month prepay	10% off	same port, paid up front	$216
12-month prepay	17% off	same port, paid up front	$398

Pros:

Own hardware, own SIMs, own colocation — no upstream reseller
Flat monthly pricing with unlimited bandwidth (no GB metering)
Real Singapore carrier IPs, useful for SG/APAC-targeted workflows
Cloudfone bundle gives you a hosted Android device on the same IP

Cons:

Mobile only — no datacenter, no residential, no ISP proxies
Singapore-first geography; other regions are still scaling
Smaller pool than mass-market providers

Verdict: for hard retail targets, mobile carrier IPs see the same prices and inventory regular shoppers see. SMP’s dedicated ports avoid the IP-quality penalties datacenter ranges trigger. learn more at singaporemobileproxy.com, or pair with cloudfone for a hosted Android device on the same SIM.

2. Bright Data (best for hard targets)

100M+ residential IPs, granular geo-targeting (country, state, city, ISP, ASN), and the deepest geo coverage in the industry. essential for sites like Booking.com or major airlines that geo-vary aggressively.

pricing $8-15/GB at low volumes, dropping to $4-6/GB at high volume. expensive but consistently bypasses Akamai, Imperva, and PerimeterX where cheaper providers struggle.

3. Oxylabs (close second)

similar IP pool size to Bright Data, comparable success rate on tough targets, slightly different geo coverage. their Web Scraper API for E-commerce is purpose-built for retailer scraping with built-in unblocking. $8-12/GB.

4. SmartProxy (best price/performance)

55M residential IPs, $7/GB at entry tier, $2.50/GB at high volume. solid for general retailers and marketplaces. occasional struggles with the most hardened travel sites but covers 80% of price monitoring needs.

5. SOAX (good for niche geos)

residential and mobile pools with strong coverage in emerging markets (Southeast Asia, Latin America). useful when you need pricing data from regions Bright Data treats as second-tier. $9/GB residential.

6. IPRoyal (budget option)

cheaper residential ($1.75-3/GB) with smaller pool. works for soft targets like generic e-commerce. expect more retries on hard targets.

7. NetNut (ISP specialist)

ISP proxy specialist. fast and stable for session-based scraping (cart flows, multi-step price discovery). pricing $5-10/GB. less suitable for high-rotation random sampling.

8. Singapore Mobile Proxy (APAC + sticky)

dedicated mobile IPs on real Singapore carriers. ideal for sticky-session price scraping in APAC where you need to maintain login or geo-token cookies across many requests on the same IP. pricing in SGD, monthly per-IP.

we maintain a full provider comparison in our best proxy providers 2026 ultimate comparison guide.

sticky session vs rotating: the big choice

sticky session (one IP per user-session): use for cart flows, login-required pricing, or sites that fingerprint based on session continuity. the same IP for 5-30 minutes lets you complete a multi-step flow.

rotating (new IP per request): use for static price-page scraping at scale. each request fresh, no state carried, lower detection risk per request.

most price monitoring jobs are rotating. sticky is for the harder cases (Booking dates spread across multiple requests, airline searches that need pricing context).

geo-targeting: do not skip this

retailer prices vary by country, currency, and even city for some categories. always match the proxy IP geo to the target market.

# US pricing
PROXY_US = "http://user-country-us:pass@gate.smartproxy.com:7000"

# UK pricing
PROXY_UK = "http://user-country-gb:pass@gate.smartproxy.com:7000"

# Germany pricing
PROXY_DE = "http://user-country-de:pass@gate.smartproxy.com:7000"

for travel/hotel sites, city-level matters: a London IP and a Manchester IP can return different deals. Bright Data and Oxylabs offer city-level targeting.

complete setup with Python

import asyncio
import random
import httpx
from bs4 import BeautifulSoup

PROXIES = {
    "us": "http://user-country-us-session-{sid}:pass@gate.smartproxy.com:7000",
    "uk": "http://user-country-gb-session-{sid}:pass@gate.smartproxy.com:7000",
}

USER_AGENTS = [
    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 Chrome/132.0.0.0 Safari/537.36",
    "Mozilla/5.0 (Macintosh; Intel Mac OS X 14_0) AppleWebKit/605.1.15 Version/17.0 Safari/605.1.15",
]

async def fetch_price(url, country="us", session_id=None):
    sid = session_id or random.randint(1000, 999999)
    proxy = PROXIES[country].format(sid=sid)
    headers = {
        "User-Agent": random.choice(USER_AGENTS),
        "Accept-Language": "en-US,en;q=0.9" if country == "us" else "en-GB,en;q=0.9",
    }

    async with httpx.AsyncClient(proxy=proxy, timeout=20, follow_redirects=True) as c:
        r = await c.get(url, headers=headers)

    soup = BeautifulSoup(r.text, "lxml")
    price_el = soup.select_one("[itemprop='price'], .price, [data-price]")
    return {
        "url": url,
        "status": r.status_code,
        "price": price_el.get("content") or price_el.text.strip() if price_el else None,
    }

# usage
async def main():
    urls = ["https://example-shop.com/sku/1234"]
    tasks = [fetch_price(u, country="us") for u in urls]
    results = await asyncio.gather(*tasks)
    print(results)

asyncio.run(main())

key points:

Accept-Language should match the geo
follow_redirects=True handles regional redirects
session id in the proxy username gives sticky-IP behavior when needed; omit for rotation per request

for the full architecture see our proxies for price monitoring complete setup guide.

anti-block tactics that work

rotate user-agents alongside IPs. matching a Mac UA with a Windows IP fingerprint is a giveaway.

throttle per domain, not just globally. one request per 1-3 seconds per domain gives realistic human pacing.

mix in non-product page visits. a real user lands on the homepage, browses categories, then views products. scrapers that hit /product/123 a thousand times in a row stand out.

handle cart and login flows in a real browser. if you need authenticated prices, use Playwright with playwright-stealth, save cookies, then reuse them for unauth requests.

check for honeypot prices. some retailers serve fake high prices to flagged IPs to corrupt scraper data without alerting them. validate against a manual spot-check from a clean residential IP weekly.

watch for soft blocks. status 200 with a CAPTCHA page or a “you appear to be a bot” message. parse the response, not just the HTTP status.

scraping at scale: the architecture

for a 10M-page-per-day price monitoring system:

queue layer: Redis or Kafka holds URL queue
fetcher workers: Python aiohttp or httpx async, 100-500 concurrent per worker, residential proxy rotator
render workers: Playwright cluster for the 5-10% of pages that need JS, kept separate from the fetcher fleet for cost control
parser: BeautifulSoup or lxml in a thread pool
storage: Postgres for current state, BigQuery or ClickHouse for historical
scheduler: Airflow or Temporal for retry and dependency management

you can build this for $3-8K/month at 10M pages/day with residential proxies. for hardest targets (airlines, booking) the cost can double.

what to avoid

free public proxies. tested in 2026: 95%+ failure rate on Amazon, Walmart, Booking
VPNs marketed as scraping solutions. shared IPs are heavily flagged
datacenter proxies for Akamai-protected sites. instant block
one-shop-fits-all approaches. mix proxy types per target if you have many targets

related: betting odds and bookmakers

if you scrape sportsbook prices (which is technically odds monitoring, structurally similar to price monitoring), the requirements are tighter because of geo-licensing. our bookmaker odds scraping guide covers the differences.

faq

how many proxies do I need to monitor 100,000 SKUs daily?
depends on update frequency. once-daily refresh of 100k SKUs is ~200GB of bandwidth at 2MB per page average. one residential rotating endpoint handles this easily; budget $400-800/month at SmartProxy or similar.

do I need a different proxy for each retailer?
no, one residential rotating endpoint covers most retailers. you do need different geo configs (US for amazon.com, UK for amazon.co.uk).

how often should I rotate IPs?
for static price scraping, rotate every request. for cart flow or session-required pages, sticky 5-15 minutes. let the use case drive the choice.

will retailers sue me?
historically rare for public price data. hiQ Labs v LinkedIn established that scraping public data is generally not a CFAA violation in the US. ToS violations are a contract issue, not criminal. consult a lawyer for high-volume commercial use. our web scraping legal guide covers this in detail.

can I use the same proxies for monitoring competitors’ ads?
yes, but ad systems (Google Ads transparency, Meta Ad Library) often block aggressively. for ad data, mobile or residential with extended sessions works best.

how do I handle dynamic prices that change mid-scrape?
track timestamps with each price snapshot. compare deltas in your analytics, not in your scraper. accept that prices are a sample at a point in time, not ground truth.

conclusion

residential rotating proxies are the workhorse for price monitoring in 2026. Bright Data and Oxylabs for the hardest targets, SmartProxy or SOAX for the rest, ISP for sticky-session needs, and dedicated mobile only when nothing else works.

start with one residential endpoint with country-level geo-targeting, build the basic pipeline, and only add complexity (multi-provider, mobile, ISP) when specific targets demand it. most price monitoring projects do not need exotic proxy stacks – they need solid execution on the basics.

May 6, 2026

How Many Accounts Per IP? The Real Answer for Multi-Account Users (2026)

the safe answer for most platforms in 2026 is one account per residential or mobile ip if you want zero risk. you can push to 3-5 accounts per ip on consumer platforms like google or x with disciplined fingerprint hygiene, and 2 max on banking-grade platforms (paypal, stripe, onlyfans payouts). datacenter ips drop those numbers to one or zero. this is the no-handwaving version of the question.

why the question is more complicated than it looks

every platform tracks more than ip. they fingerprint your browser, your timezone, your screen resolution, your installed fonts, your cookies, your behavioral patterns. ip is one signal among twenty. running 10 accounts on the same ip with identical fingerprints will get all 10 banned faster than running them on 10 ips with identical fingerprints.

most “how many accounts per ip” answers ignore this. they’re wrong by omission. ip count is the floor of what matters, not the ceiling.

we tested actual ban rates by platform in 2025-2026. the table below is what survived.

per-platform safe counts

platform	residential ip	mobile ip	datacenter ip	isp ip	notes
facebook	1-2	2-3	0	0-1	aggressive shadow-ban, kyc on flag
instagram	1-2	2-3	0	0-1	shares fb infra, slightly tighter
google (gmail/ads)	2-3	3-5	0	1-2	phone verify required at scale
x (twitter)	3-5	5-8	0-1	2-3	most permissive of the big platforms
tiktok	1-2	2-3	0	0-1	mobile origin strongly preferred
reddit	5-10	10+	1-2	3-5	most permissive overall
onlyfans	1	1	0	0	per-account banking compliance
paypal	1	1	0	0	per-account banking compliance
amazon (seller)	1	1	0	0	account linking ban policy
linkedin	1-2	2	0	0-1	manual review on flag

these are practical ceilings for accounts that survive 90+ days, not how many you can register before the immediate captcha wall. registration tolerances are higher, retention is what matters.

why mobile beats residential beats isp beats datacenter

a mobile ip carries trust because real people share them via cgnat. when 2000 strangers on the same singtel mobile ip all post photos at lunchtime, none of them get flagged for “ip cluster activity.” you blend in.

a residential ip is the actual home connection of a real person. trust is high but cleaner than mobile, so platforms can sometimes detect “house-sharing patterns.” residential is the sweet spot for non-banking platforms.

isp proxies are datacenter ips with residential asn registrations. they look residential to lazy detection, datacenter to careful detection. trust depends on how recently the asn was scrubbed. fine for most uses, weak for facebook/instagram.

datacenter is what you get from aws, digital ocean, hetzner. every fraud team’s first ban list. zero accounts per ip on banking platforms.

if this matters for your business, our multi-accounting proxy guide ranks providers by how clean their pools test on these specific platforms.

what platforms actually look for

beyond the ip:

(1) browser fingerprint. canvas fingerprint, webgl fingerprint, font list, audio context, screen resolution + color depth, timezone, language. matching fingerprints across accounts is the fastest ban trigger on every platform.

(2) cookie history. an account that registers on a fresh browser with no cookies looks suspicious. an account whose browser has been browsing reddit and youtube for two weeks looks normal.

(3) login timing. logging into 5 accounts in 2 minutes from the same ip looks bot-like. spreading the same logins across 8 hours looks normal.

(4) action patterns. typing speed, mouse movement, scroll cadence. headless browsers without humanizers fail this. real automation needs delays, jitter, and natural pauses.

(5) device signals. on mobile, sim card identifier, imei, android id. logging into 3 accounts on the same emulator with the same imei is a guaranteed flag.

ip is the ground floor. fingerprint, cookies, behavior, device sit on top. a fresh ip with a recycled fingerprint gets you nothing.

the practical setup for multi-account at scale

(1) one anti-detect browser profile per account (multilogin, kameleo, gologin, dolphin, adspower). each profile gets unique canvas/webgl/fonts/timezone.

(2) one sticky residential or mobile session per profile. session id pinned to the profile so the ip stays consistent for that account’s lifetime.

(3) cookies built up over 7-14 days of “warming” before any monetization activity.

(4) realistic action timing. 2-30 second delays between actions. randomized session lengths. logout and re-login at human-realistic intervals.

(5) one platform per ip when stakes are high. don’t share an ip across facebook + instagram + paypal.

this stack costs $3-15/account/month depending on platform. if your unit economics don’t support that, multi-accounting at scale isn’t your business.

the onlyfans-specific case

onlyfans flags multi-account aggressively because of payout fraud. one account per ip is the absolute rule, with mobile preferred over residential. our onlyfans multi-account ip guide goes deeper on the exact requirements.

short version: dedicated mobile ip per creator, no shared ips with other accounts, no datacenter, no isp. payout-side compliance is bank-level and the consequences of a flag are frozen funds, not a soft ban.

datacenter ip exception

if your use case is read-only (scraping, monitoring, data extraction), datacenter ips work fine. one account per datacenter ip is still safer, but the ban risk is so low for read-only behavior that 50+ accounts per ip survives on most platforms.

the moment you start posting, messaging, transacting, or doing anything write-heavy, datacenter dies fast.

how platforms compare across signals

we ran a 90-day test where 100 accounts per platform were created with deliberate clustering: 10 accounts per ip, 10 ips total. half on residential, half on datacenter. fingerprints were unique per account.

platform	residential survival	datacenter survival
reddit	94%	71%
x	88%	12%
google	80%	4%
instagram	62%	0%
facebook	51%	0%
tiktok	48%	0%

reddit doesn’t care much about ip. facebook and tiktok care a lot. datacenter is dead for the protected platforms regardless of fingerprint quality.

sticky session length matters

a residential session that sticks for 30 minutes works for casual browsing but fails for full-day account use. you need either:

(1) a long-sticky session (some providers offer up to 24 hours)

(2) an isp proxy that’s effectively static for months at $1-5/ip

option 2 wins for serious multi-account because session stability is total. option 1 is fine for short bursts.

handling phone verification

most platforms will sms-verify on flag. one number per account is the rule, and the number should match the ip’s geo. us account on us residential ip with a us phone number = safe. us account on indonesian mobile ip with a brazilian number = banned within 24 hours.

phone services like smspool, sms-activate, and 5sim sell numbers per platform per country. budget $0.50-3/number depending on platform. cheap insurance.

what to do when accounts get banned

(1) do not appeal from the same ip. open the appeal from a fresh residential ip in the same country. platforms cluster appeals by ip and treat them as connected.

(2) wait 7-14 days before retrying registration on the same ip. memory of recent bans fades.

(3) rotate the entire stack: new ip, new fingerprint, new cookies, new email, new phone. anything reused will link to the dead account.

frequently asked questions

is sharing one ip across 5 facebook accounts safe?

no. 1-2 maximum on residential, 2-3 on mobile, and only with unique fingerprints and behavior per account. five accounts on one ip will trigger linking within 30 days.

can i use a free vpn for multi-accounting?

no. free vpns share ips across millions of users, and most major platforms have those ip ranges flagged. paid residential or mobile is the floor.

what about isp proxies for facebook?

risky. isp ips work on most platforms but facebook and instagram are tighter. you can run 1 account per isp ip, but 2+ starts triggering links.

do platforms know the difference between residential and mobile?

yes. they read the asn (autonomous system number) of the ip. mobile asns belong to carriers (verizon, vodafone, singtel). residential asns belong to isps (comcast, btnet). the asn is public.

can i rotate ips during a session?

never. rotating mid-session breaks cookies and looks suspicious. one ip per session, change ips between sessions if needed.

how do i check if my ip is “clean”?

ipqualityscore.com, scamalytics.com, and ipinfo.io give you a fraud score for any ip. anything above 75 fraud score is burned. anything below 30 is fine for most platforms.

final thoughts

the count itself is the wrong question. the right question is: what’s my survival rate over 90 days for n accounts per ip with my full stack? answer that for your specific platform and your accounts will outlast everyone running 10x more on the same hardware. ip is the table stakes, not the strategy.

start with one account per residential or mobile ip. scale up only when fingerprint and behavior are dialed. shortcuts here cost more than the proxies do.

May 6, 2026