Using it properly

The software handles behaving like a person. The rest — how much you run, what you switch on, what infrastructure you use — is on you, and that is what decides whether accounts survive. This page covers the places users most often hurt themselves.

1. Enabling more tasks does not mean getting more done

Each session, every account picks 3–8 tasks at random from the ones you enabled (default) — not all of them. Session time is one budget split across the tasks that were picked, so the more you enable, the thinner each slice gets.

Each task also has its own time ceiling to do something meaningful (Gmail 210s, Docs 195s, Finance 190s). Forcing Min = Max = 8 gives every task a sliver: enter, scroll a little, leave. The log looks busy while no real usage trace is left behind.

2. Five actions leave public traces — off by default, and best left off

These five write to the outside world and stay attached to the account more or less permanently: YouTube comments, Shorts comments, seeding comments and replies, and newsletter sign-ups on third-party sites. The software keeps them off by default and will never auto-enable them, even if the configuration cannot be read.

Reason: a weak or repetitive comment is something real people see and report — very different from a view.

3. Accounts are deliberately different — do not force them to match

The number you type is the average across the herd, not an order for each account. Every account gets a fixed "personality": some watch longer, some barely, some like often, some never. Set 40% and the herd averages ~40%, but a share of accounts will never do it — that is intentional.

4. Seeding: use percentages, not seconds

The default watches 25–70% of the video, with 12% of views abandoned early. Percentages matter because one setting in seconds behaves very differently across videos: 60 seconds is 100% of a one-minute video but 20% of a five-minute one. Retention rate is what gets compared.

The early-abandon share looks wasteful but is necessary: every real video has people who click and leave. Without them your retention curve is suspiciously perfect.

5. Mix the entry paths

Declare the same video in several groups with different modes (YouTube search, channel, suggested video, home feed, subscriptions feed, Google search). If every account reaches the video by one identical route, the traffic-source mix itself is a signal, no matter how natural each account looks.

6. Proxies — the most common failure

• The app can only check that the proxy port is open, not that the proxy can actually reach the internet. An out-of-bandwidth proxy still "answers" and passes the check, yet the browser loads nothing.
• If your proxies egress over IPv6, test them against an IPv6-capable endpoint (e.g. api6.ipify.org). Testing against an IPv4-only site reports a false "dead" — and some non-Google sites genuinely will not load.
• If an account shows Proxy Dead or a profile fails to start, check the proxy first — not the profile.

7. New accounts: go slow, and keep sensitive work separate

Freshly created or freshly bought accounts should run nurture tasks for a few days before you touch the security group (password change, recovery email, two-step verification, app passwords). Those actions already attract closer inspection; stacking them into one session on an account with no history is the worst case.

8. Read the results correctly

• Verify means Google is asking for verification; these usually need manual handling — retrying does not help.
• A batch of accounts failing at once is almost always infrastructure (dead proxies, lost connectivity), not the accounts. Stop and check proxies before continuing.
• Continuing through a mass failure only makes it worse — each attempt is another failed sign-in on the account's record.