Sticky Sessions Under a Stopwatch: How Long a Residential IP Really Stays Yours

You open a profile, sign in to an account, and start warming it up. At the fourteenth minute, the platform asks you to verify your identity. Nothing has changed on your side: the profile, fingerprint, and proxy settings are all the same. The exit IP has changed. The sticky session quietly moved you to another residential address, and the platform recorded a network change during an active session.

Most proxy dashboards let you choose a session duration of 10, 30, or 60 minutes, and sometimes longer. Depending on the provider, this may be a maximum or target duration rather than a guarantee that the IP will remain unchanged for the entire period. To see how much earlier real sessions end, we measured 20,913 sticky sessions across 17 residential proxy networks. Below, we explain the results, three IP-change patterns, and how to run the same half-hour proxy test yourself.

What is a sticky session and what does it not promise?

A residential proxy network routes your traffic through ordinary household devices such as home routers, computers, and phones. With rotation, each request may leave through a different device. With a sticky session, you add a session identifier to the proxy login, using the provider's own syntax, and the network tries to keep every request with that identifier on one exit device for a set period. The basic differences between these modes are covered in our guide to proxy types and their use cases.

The important word is “tries.” The exit device uses someone else's home connection. It can go offline, switch to another network, or receive a new address from the ISP. The proxy network may then assign the session to another device.

The IP can also change on the same device because the ISP reassigned its public address or the proxy network performed an internal rotation. Some networks also end sessions on their own timer, regardless of the duration you requested.

An IP change is not always critical for scraping. When you work with an account, however, the platform may treat it as one risk signal: requests with the same session cookie arrive from another address, sometimes from another city or network. This does not necessarily trigger an extra verification or restriction.

How did ProxyStats measure sticky session duration?

ProxyStats runs the same tests for every network it tracks. For this article, we used two weeks of long-session tests conducted from October 13 to 27, 2026:

Every network was tested under the same conditions: the same requested session duration, check interval, and regions. A session whose final check failed proves neither retention nor a break, so we excluded it. We also excluded hours when a network was under attack. Requests every 2 to 3 minutes kept the session active, preventing idle-time limits from taking effect. The full rules are described on the methodology page explaining how we measure each test.

For a repeatable test, save the proxy protocol and session identifier format with the results. Record the test node's time zone separately. Changing any of these parameters creates a different experiment, even when the provider and plan remain the same.

How often do sticky sessions lose their IP?

We requested 60 minutes from every network. The table shows how many sessions changed IP well before that point:

Three points stand out.

Even strong networks lose some sessions. On the best network, roughly one in eight sessions changed IP within 30 minutes. On the median network, it was about one in five. If your workflow assumes the session will last for the full requested period, account for these breaks.

The spread between networks is wide. Four of the 17 networks changed IP in more than half of their 30-minute sessions. One network changed IP in every 30-minute session, even though we requested 60 minutes.

A 10-minute test can hide a 30-minute session problem. The network that broke 100% of its 30-minute sessions lost only 11% during 10-minute tests. A quick check immediately after purchase would have shown a stable proxy. The problems appear later, in the middle of the workflow.

Store raw results for each session rather than only an aggregate percentage. This lets you recalculate the metric after excluding control requests that ended in an error.

Why does the connection region affect a sticky session?

The same network can behave differently by region. In 11 of the 17 networks, 30-minute sessions held better in the EU than in US-East. One network changed IP in 4% of EU sessions and 36% of US-East sessions, a ninefold difference.

If you manage accounts in a specific region, test the proxy from that same region. A result from somewhere else tells you very little.

How exactly does a sticky session break?

When a session lost its IP within 30 minutes, we recorded the first check that saw the new address. The timing reveals three common patterns. They suggest possible causes, but IP measurements alone cannot prove those causes.

For diagnostics, record the ASN and country before and after each change. Mark the address type separately. This makes it easier to distinguish rotation within one pool from a move to another network.

Early break in the first three minutes

Nineteen percent of all IP changes occurred within the first three minutes. The session identifier was assigned and the first requests succeeded, yet the network still moved the session. One possible explanation is that the network finalizes session binding after the first requests or the initial exit device quickly loses its connection. The provider's internal data would be required to determine the cause.

How to spot it: the IP on the first check differs from the address used at sign-in.

Timer at the same minute every time

Twenty-three percent of all changes landed in one window, between minutes 18 and 21. Roughly four out of five of these changes came from one network that ends sessions near minute 20, even though we requested 60 minutes. A timer is easy to recognize because the breaks cluster around one point.

How to spot it: several consecutive sessions change IP at almost the same minute. Repeated timing may indicate an internal timer or another systematic product limit, but additional checks are needed for confirmation.

Gradual attrition throughout the session

The remaining changes were spread across the entire 30-minute test, with 4% to 12% occurring in each three-minute window. This may reflect normal residential pool behavior: household devices disconnect, and the network reassigns their sessions. The exact cause cannot be determined from IP changes alone. Better networks show less attrition, but eliminating it completely is difficult.

How to spot it: breaks occur at random minutes, and more sessions survive a short interval than a long one.

How can you test your proxy in Afina?

You do not need specialized tools. In Afina, every browser profile stores its own proxy and isolated storage, so it can act as a separate test session.

1. Create 10 profiles and assign the same proxy to each one, using a different session identifier in every login. Set the longest session duration allowed by the provider
2. Open a page that displays the IP in every profile, such as api.ipify.org, then record the IP and time. This is minute zero
3. Refresh the page every 3 minutes for 30 minutes and record the IP. Afina's no-code automation builder can run the cycle: open the page, save the result, wait three minutes, and repeat
4. Run the test from the account region and during the hours when you normally work
5. Calculate the result: count how many of the 10 sessions changed IP after 10 minutes, after 30 minutes, and at which minute each change happened

Compare your numbers with the table above. One or two breaks out of 10 at the 30-minute mark occur on a typical network. Five or more appeared among the weakest networks we measured. If most breaks occur at the same minute, you have found a timer.

Check which changes your tools react to. If your version of Afina includes a profile killswitch, review its settings to see which network parameter changes it tracks and when it stops the profile. Do not equate this mechanism with disconnecting the entire device from the network. Direct checks of the actual IP address are still necessary.

Ten sessions provide an initial picture. We compare networks only after each has 150 included sessions for a given region and duration because individual runs vary widely.

What should you do with the result?

The test shows what to expect. It leads to several practical actions:

• keep work blocks shorter than the break point: if your network usually ends sessions around minute 20, complete sign-ins, account warming, and payments in shorter blocks, starting each block with a new session
• check the IP before important actions: a quick IP check before signing in or paying takes seconds and shows whether the session has moved
• do not trust one short check: a network that remains stable for 10 minutes may still fail every 30-minute session
• repeat the test after changes: providers update their infrastructure, and session behavior changes with it. A network that held well last month may behave differently this month, or vice versa

Where can you compare residential proxy stability?

ProxyStats compares networks with the same methodology and updates the results as new sessions arrive. This material is provided for informational and educational purposes only.

Before buying, review each network's measured 10- and 30-minute session retention, calculated with the same method used in this article.

Download