ISP Proxies for Running Multiple Accounts: What to Buy

Buy one dedicated ISP address per account you intend to keep. Not per team, not per campaign, per account. The address is the only identifier you cannot fake, and sharing it links the accounts behind it whatever else you do.

Do not buy the shared tier for account work. Shared static addresses are sold at a fraction of the dedicated price, and the discount is exactly the risk: you inherit whatever the last tenant did with that address, and you have no way to find out what that was.

Budget roughly $1.50 to $2.70 per account per month. That is the range of published dedicated entry rates across three vendors read on 23 September 2026, and it is a small enough number that the argument for economising on it is usually a false one.

Match the address's country and city to the account's history, not to the platform's headquarters. An account with two years of activity in Lyon should not begin appearing from Amsterdam because Amsterdam was in stock.

Skip ISP and go to mobile only when a specific target has demonstrably rejected a clean, dedicated ISP address. That is a real category, it is just a much smaller one than mobile proxy vendors imply.

An ISP proxy is an IP address that belongs to a consumer internet service provider's allocation, hosted on a server in a datacenter rather than on a household router. Vendors also sell it as a "static residential proxy", which is the same product under a different name.

The important consequence is that it carries two properties that normally come apart:

The ownership record looks residential. When a platform looks up who announces the address, it sees a consumer ISP's autonomous system rather than a hosting company's. That single lookup is one of the cheapest and most widely used signals in abuse detection, and an ISP range passes it where a datacenter range does not.

The performance is a datacenter's. The machine is racked, on a wired uplink, with no household bandwidth to share and no laptop that closes at midnight. Latency is stable and throughput is high, which is the opposite of what you get from a peer-to-peer residential pool where the exit is somebody's actual computer.

That combination is why the product exists. Our explainer on what ISP proxies are covers the mechanics, and the comparisons against residential and datacenter proxies set out where each line sits.

One word of caution on naming. Some vendors sell "rotating ISP proxies", which rotate through a pool of ISP addresses. That is a scraping product, not an account product, and buying it for logins gets you the worst of both: ISP pricing with rotation's instability. Our note on rotating ISP proxies explains what they are for.

The instinct that rotation equals stealth is the single most expensive misconception in this space, and it is backwards for account work.

Think about what a platform's risk engine is actually built to notice. It is not looking for a datacenter IP as such. It is looking for changes in the pattern that a genuine user would not produce. A real person logs in from one or two addresses for months. Their ISP occasionally reassigns them a new address in the same city, and the platform sees that as normal because it happens to everyone on that network.

Rotation produces none of that. An account that appears from a different address every session, in a different city, on a different subnet, is describing a travel pattern no human has. The platform does not need to know the addresses are proxies; the inconsistency is the signal on its own.

There is a second, subtler cost. Risk engines build a profile of the addresses an account normally uses and treat a new one as a small risk increment, usually satisfied by a verification step. On a static address that step happens once, early, and then never again. On a rotating pool it happens constantly, which both annoys the operator and teaches the platform that this account is unusually mobile.

The exception proves the rule. Rotation is correct when you are reading, not logging in: collecting public pages, checking search results, sampling prices. There the goal is spreading load across many addresses so no single one trips a rate limit, and persistence is worthless. Our comparison of rotating versus static residential proxies covers the split.

Every ISP vendor sells two tiers and they are described almost identically. The gap between them is the most consequential decision in this post.

Shared. The address is used by several customers at once. Published entry rates read on 23 September 2026 were $0.23 per IP on Webshare's static residential tier and from $0.27 per IP on Decodo's shared ISP tier.

Dedicated. The address is yours alone for the term. Published entry rates the same day: $1.47 per IP for Webshare's dedicated static residential, from $2.00 per IP for Decodo's dedicated ISP, and from $2.70 per proxy on IPRoyal's 30-day ISP term.

The multiple between the two tiers is six to seven times, and that multiple is the price of not inheriting a stranger's history. Consider what a shared address means in practice. You do not know how many other tenants are on it. You do not know which platforms they use it for. You cannot find out whether it was used to register forty accounts on your platform last Tuesday. And critically, when the platform associates that address with a pattern of abuse, your account is inside the association whether or not you caused it.

For reading public pages, shared is perfectly reasonable and the saving is real. For anything holding a login, the dedicated tier is the product and the shared tier is a different product that happens to share a name. Our explainer on dedicated residential proxies covers the same distinction from the residential side.

Twenty accounts, one dedicated address each, priced on published entry rates read 23 September 2026. Confirm current rates on each vendor's page before you buy.

Two comparisons make those numbers meaningful.

Against rotating residential, per gigabyte. Webshare's published rotating residential entry rate was $1.40/GB on the same day. Divide a dedicated address's monthly price by that rate and you get the break-even traffic: $1.47 divided by $1.40 is 1.05 GB, and $2.00 divided by $1.40 is 1.43 GB. In other words, an account that moves more than about one gigabyte a month costs more on rotating residential than a dedicated ISP address costs outright, and the rotating option does not even give you the persistence you were buying. Only accounts that are genuinely idle come in under that line.

Against the browser licence. Twenty accounts fit comfortably inside the entry tier of most antidetect browsers. At $29.40 to $54.00 a month for the addresses, the proxy layer is once again the larger line item, which is worth carrying into any budget conversation.

What that table does not include, deliberately, is a mobile proxy price. Mobile is generally sold per port per month rather than per address, with the port shared across the carrier's own NAT, and we did not read a current published rate today. Rather than reconstruct one from memory, the honest answer is that mobile sits materially above every row in that table and you should price it on the vendor's own page.

A static address is an asset and it can be destroyed. There are exactly three ways, and each has a different mitigation.

Burned by a neighbour. Somebody else on the address did something the platform reacted to. This is the shared-tier failure mode and it is unfixable after the fact, because you had no visibility and no control. Mitigation: buy dedicated. That is the whole mitigation.

Burned by you. Your own behaviour on the address triggered a response: too many registrations, too fast a cadence, an action pattern that does not look human. This is the most common failure among people who bought correctly, and it is a behaviour problem wearing an infrastructure costume. No address survives being used the way a script uses one. Our write-up on why proxy accounts get suspended covers the patterns.

Burned by the range. The platform stopped trusting the whole subnet, not your address. ISP proxy ranges are known to the commercial IP intelligence databases that sell risk scores, and being in a consumer ISP's allocation is not a guarantee of a clean score. When a vendor sells thousands of addresses out of a handful of prefixes, a platform that decides to act against those prefixes takes all of them at once. Mitigation: spread important accounts across vendors and countries rather than buying one block from one supplier, and check the address's reputation before you commit an account to it. Our explainer on IP reputation covers what those databases actually score.

That third mode is the one buyers underestimate, because it is invisible until it is total. It is also the strongest practical argument against putting every account you own behind addresses from a single vendor, however good the price.

Vendors sell ISP addresses on different commitment shapes, and the cheapest per-day rate is not automatically the best buy.

IPRoyal's published ISP pricing on 23 September 2026 ran from $1.80 per proxy for 24 hours, $2.70 for 30 days, $2.55 for 60 days and $2.40 for 90 days. The per-day rate falls as the term lengthens, which is the usual shape. Decodo's ISP page stated no minimum purchase on either its per-IP or per-GB plans, with a 3-day free trial and a 14-day money-back window.

The question longer terms raise is not discount, it is what you are holding. If an address burns in week three of a 90-day prepayment, the discount you captured is gone and you are carrying a dead asset. Before committing to a long term, get three answers in writing:

• Is there a replacement policy for a flagged address, and does it cost anything? Some vendors replace on request, some charge, some decline.
• Is there a monthly replacement allowance? A stated number is much better than goodwill.
• What happens to the address when I cancel? If it goes straight back into the pool for someone else, that is worth knowing, both for your own exit and for what it tells you about the addresses you were sold.

A short first term while you learn whether a vendor's ranges work on your specific platforms is usually worth more than the discount on a long one. Our roundup of ISP proxy providers covers who sells what.

The operational details are few, and getting them wrong undoes the purchase.

One address per account, recorded. Keep a mapping of account to address, with the date it was assigned and the vendor it came from. When something goes wrong, this table is the only thing that tells you whether the problem is one account, one address or one range.

Assign inside the browser profile, never at the OS level. A system proxy routes everything through one exit, which is the failure the whole exercise exists to prevent.

Use credentials rather than IP whitelisting where you can. Whitelisting breaks whenever your own connection changes address, which on home broadband or a laptop that moves is often. Where a vendor offers both, credentials travel with the profile and whitelisting does not.

Verify the address before you trust it. Three checks, once, per new address:

# 1. What does the exit actually look like?
curl -x http://USERNAME:PASSWORD@ISP_PROXY_HOST:PORT -s https://ipinfo.io/json

Read back the country, region, city and the organisation or ASN field. Confirm the ASN belongs to a consumer ISP rather than a hosting company, confirm the city matches what you were sold, and confirm the timezone you set on the account's browser profile agrees with the location. Then check the address against an IP reputation lookup before you attach anything valuable to it. An address that arrives already carrying a poor score is a support ticket, not a purchase you have to live with.

Do not share one address across a "family" of accounts you consider related. If they were genuinely one person's accounts on one connection, fine. If they are twenty seller accounts you would prefer the platform not to associate, the address does the associating for it.

SparkProxy's place in this stack is the read-only half. Monitoring competitors, checking how a listing renders in another country, pulling public data at volume: that is rotating datacenter work, priced flat per thread from $75/mo for 100 threads with unlimited bandwidth, and it should never touch an account profile.

Mobile proxies exit through a cellular carrier, where carrier-grade NAT puts a large number of real subscribers behind each public address. That shared-by-design property is exactly why platforms treat those addresses gently: acting against one punishes hundreds of genuine users.

It is a real advantage and it is expensive, and the honest guidance is to treat it as an escalation rather than a starting point. Buy mobile when:

• A specific platform has rejected, challenged or restricted an account on a clean, dedicated ISP address, and you have ruled out behaviour as the cause.
• The account type is one the platform associates strongly with phone use, and desktop-shaped signals are drawing attention.
• You are operating in a market where fixed-line broadband is uncommon and a residential ISP address would itself be the anomaly.

Do not buy mobile because a forum recommended it in general terms. The property you are paying for, shared carrier NAT, also means you have no control over who else is behind your address today. Our comparison of ISP proxies against mobile proxies covers the trade in detail, and our post on mobile or residential proxies for account work weighs those two against each other where ISP is not an option.

1. Dedicated, in writing. Ask the vendor to confirm the address is not shared for the duration of the term. "Static" alone does not mean it.
2. Which ASN, and can you see an example? Ask for a sample address to look up before you buy. A vendor who will not provide one has told you something.
3. City or country only? Check the granularity you need is actually available for the market you need it in.
4. Replacement policy and allowance. Get a number, not a reassurance.
5. What happens at term end? Both to your access and to the address.
6. Trial terms on the static product specifically. Trials often exclude static addresses, which are the ones you need to test.
7. Are the addresses concentrated in a few prefixes? If your twenty accounts sit in two subnets, they are one decision away from going together.