ISP or residential? How to pick the right network for the job

If you have ever opened our pricing page and stopped at the first decision, this post is for you.

ISP and residential look similar on a feature list. They both route real traffic, they both support sticky sessions, they both work with the same credential pattern. The bill at the end of the month is what makes them feel like different products.

Here is the short framework we give customers when they ask.

What the two networks actually are

A residential route uses an address from a real consumer connection. The pool is large, the addresses change often, and the per-request reputation is whatever the underlying user brought to it. We source these addresses ethically, through partners who get informed consent from device owners, so the pool stays clean and the addresses behave like the real users behind them.

An ISP route uses an address that lives on a static block at the same kind of provider, but allocated to us for steady use. The pool is smaller, the addresses are stable, and the reputation is consistent across long sessions.

Both feel like real users to the destination. The difference is how predictably they behave under load and what they cost.

When residential is the right call

Reach for residential when the work is bursty and broad. Examples we see all the time:

• Marketplace sweeps that need to land on many countries quickly.
• Discovery jobs where the success of any single request matters less than the success of the run as a whole.
• Continuous monitoring where the breadth of the pool matters more than the precision of any one address.

The right product here is Premium Unlimited. It bills flat, the pool is wide, and the workload can run continuously without per-gigabyte anxiety.

When ISP is the right call

Reach for ISP when each request matters. Examples:

• Localized pricing or availability checks, where the answer is only useful if the route really is in the target city.
• SERP work, where the rank you collect must match what the local user sees.
• Anything requiring precise control over city or ASN.

The right product here is Platinum. It bills per gigabyte, the routes are stable, and the targeting goes deeper than country alone.

Two questions: does each request need to land in the right place, and is the bill more sensitive to bandwidth or to volume? The answers point at one network or the other.

When the answer is “both”

Plenty of teams run both networks side by side. Discovery runs on Premium Unlimited because the pool is wide and the bandwidth is unmetered. Detail collection runs on Platinum because the destination is sensitive to location.

The dashboard treats them as one account, so you do not need to plumb two billing relationships. You assign the route at the job level and forget about it. Workers can mix Premium Unlimited for the breadth pass and Platinum for the precision pass, with the only difference between them being one configuration value.

Common questions

The two we hear most often:

“Can I use ISP for everything if I do not mind paying per gigabyte?” Yes, and some teams do. It works because per-request quality is high. The reason most teams move some traffic to Premium Unlimited is bandwidth, not capability. Heavy continuous crawls cost more on metered networks than they save in precision.

“Can I use residential for everything if I want to lock the bill?” Sometimes. It works when each request can tolerate variance in the route. It does not work when the destination is location-sensitive, or when the workload depends on holding the same city for hours.

A starting recommendation

If you are unsure where to start, default to the one that matches your bill shape. A team that hates surprise invoices should start on Premium Unlimited and add Platinum when a precision use case shows up. A team that runs short, accurate jobs should start on Platinum and only move to Premium Unlimited when bandwidth becomes the constraint.

Either way, you can change the answer later. The credentials are the same, the dashboard is the same, and switching costs you ten minutes of configuration, not a migration.