Short answers, with the number behind each one

Safer than most alternatives on the credential side: your session cookie is never stored in our database. It is consumed directly into an encrypted browser profile on our server, actions run on a dedicated residential IP with per-account daily budgets and warm-up, and disconnecting destroys the profile. On the policy side we are honest: all LinkedIn automation, ours included, is against LinkedIn's User Agreement, and our safety limits reduce that risk rather than eliminate it.

No tool can promise it will not. LinkedIn does not publish its limits and its help center says automated activity can lead to a restriction. What we can show is the mechanism: connection requests capped at 3% of the account's connections per day under a weekly cap of 100 to 200, a warm-up that starts at 10% of the limit, business-hours pacing with a per-account start time that moves daily, one static residential proxy and one encrypted browser profile per account, and a pause of the challenged action family on the first security check.

It depends on the account, not on a setting. The connection budget is 3% of the account's connections per day, under a weekly cap of 100 to 200 by LinkedIn plan, and the weekly allowance is spread across the week rather than spent by Wednesday. Each action type has its own daily budget, reads and writes are separate, and a new account starts at 10% of its limits and earns the rest through use.

The family of that action pauses on the account for 24 hours, and only that family: a challenged connection request does not stop replies to people who wrote in. The ramp for that family is lowered so it comes back slower. You are notified once per account per day, in-app and by email. Three rate limits in the same family within 24 hours trigger the same pause.

The session cookie: never, in any form, not even temporarily. Your LinkedIn email and password are stored only if you explicitly opt in to automatic session recovery, and then only encrypted with a dedicated rotatable key that is never exposed to the browser.

It is used once, to open an encrypted browser profile on our server bound to the account's proxy, and it is not stored in the database. Reconnecting means pasting a fresh one. Disconnecting ends the session from LinkedIn's side as well, if you revoke it there.

Yes. An account is linked to one or more workspaces and any of them can put it in a pool. Its daily budget does not multiply: every workspace draws from the same per-account limits, so two clients rotating the same account share its day.

Leads bound to that account wait; nothing is failed and nothing moves to another account. New leads are routed to the accounts still usable. When the session is reconnected, the waiting leads resume where they stopped. Signal agents bound to that account pause and resume on their own.

Yes, by default, with a per-account off switch. A daily sweep reads the account's own sent-invitations list on LinkedIn and withdraws the oldest ones to stay under the roughly 400 pending LinkedIn allows. It never withdraws an invitation younger than 7 days, and withdrawals do not spend the connection request budget.

Only if you choose that. During onboarding you answer whether the AI may send for you; the default is manual review, where every LinkedIn message waits in the inbox as a draft you approve, edit, retry or reject. Email sends are never review-gated, because the copy is fixed per campaign and written once. Approving a draft is what resumes the send.

A second model, the critic, reads the full brief the drafter saw and returns pass or fail with one concrete objection. A fail is redrafted with the objection injected, up to three drafts. A draft that keeps failing is not dropped: it ships to your review queue flagged, pinned on top, with the reason on the card. Before each paid critic pass, two free checks run: a voice check and a grounding check that fails a draft copying six or more consecutive words from your playbook or offer text.

What you told it, in order of precedence: the company context (what you sell, for whom, and why you built it), what a message is allowed to want from the reader, then the playbook in five sections (identity, voice, boundaries, angles for openers, objections for replies). Facts win over tone: when the playbook asks for something the facts make untrue, the draft says less rather than inventing. The workspace playbook is generated once from your website and stays hand-editable.

Per LinkedIn account, yes. A voice profile is extracted from that account's real message history, excluding anything CampaignStack itself sent, and injected into LinkedIn drafts once you approve it. Email and content drafts never use it. A fidelity check before each critic pass fails a draft that drifts from the profile's measured habits.

The reply cancels every queued touch to that person in the workspace, so a follow-up never lands on top of an answer. One conversation step then owns the thread: it drafts a reply, waits for your approval if the workflow is review-gated, sends, and listens again, until a week of silence closes it. If the reply contains a scheduling link, a booking agent takes a slot first and the reply confirms the date and time.

A score. Every lead is scored against every campaign's ideal customer profile, and the score decides campaign membership: above the threshold the lead is in, below it the lead is invisible to that campaign. Signals add to the score, so a job change or a post engagement can move someone in. A workspace exclusion list is checked again at every send, not only at import.

On people: job changes, title changes, headline changes, location changes, and engagement on posts you watch. On companies: funding events, headcount and follower changes, hiring, and news, from public sources with no LinkedIn account involved. Signals expire after 60 to 90 days and stop counting. Signals are shared across customers; who is watching what, and anything derived from your outreach, is never shared.

No. Public professional data (people, companies, public posts) is shared across customers by design, like Apollo. Everything derived from your activity, including campaigns, lead lists, conversations, sends, and replies, is scoped to your workspace and invisible to every other tenant.

Exclusions. A workspace exclusion list is honoured by every consumer: campaign sourcing, every signal intake, and a second check at dispatch. Onboarding pre-fills a rule for your own company name and the current employer on each connected account, since the two differ when someone builds on the side. Rules match on company, title, LinkedIn member id, email or URL, and a suppression CSV from a previous tool imports straight into it.

Yes. CSV, a list of LinkedIn URLs, Google Sheets and Apollo lists all import, and the result lands either in a campaign workflow or as a standalone lead list. Imported leads stay private to your workspace. Email and phone enrichment runs on the built-in contact finder, charged only when something is found.

$299 a month per client workspace, or $249 a month billed annually, with 25,000 credits included every month. LinkedIn accounts and team members are not billed per seat. Usage past the allowance is metered at $35 per 10,000 credits, or bought as packs. There is no free tier.

14 days with the full product and a 20,000-credit pack that expires with the trial. When the trial ends without a subscription the workspace locks: no new campaigns, no sends, a read-only inbox, and nothing is deleted. Subscribing unlocks it and everything resumes on the next tick. Access is by invitation; ask for one on the signup page.

The unit every action is priced in. A LinkedIn send or connection request is a few credits, an AI-drafted message counts the drafting, an enrichment counts the lookup, and a found email or phone is charged only when found. Every LLM call is metered and the price of an action never depends on which model happened to answer it.

The annual price is the only standing discount, $249 instead of $299 a month per workspace. There is no per-seat multiplier to negotiate away: an agency with three clients pays for three workspaces however many accounts or teammates sit inside them.

Claude or ChatGPT will write you a good message today. What neither can do on its own is hold a LinkedIn session, keep an account under a daily budget, notice that the person replied four days ago, or stop when a meeting gets booked. CampaignStack is those parts. The model writes, the system decides who, when and how many, and a review queue sits in front of every send. If you already run an agent, connect it to the MCP server and it gets the same guardrails.

Operate the platform: read and edit campaigns, ICPs and workflows, source and score leads, craft messages under the same critic, work the review queue, read the inbox and reply, watch signals, book calendar slots, and run the admin surface. Over 300 tools, each behind a scoped API key, and every write held to the same per-account budgets as the app.

HubSpot as the system of record, Apollo and Instantly for lead lists and email sequences, the built-in contact finder for enrichment, Google for email sending and calendar, and n8n, Make and Zapier over webhooks in both directions. Each has an integration page describing the setup.

Yes, from a connected Google account, inside the same workflow as the LinkedIn steps. Email has its own per-account daily budget and warm-up: 10% of the limit on day one, up a few points per active day, decaying on idle days. The default template pairs LinkedIn and email so a follow-up after silence can change channel instead of tone.

The encrypted browser profile on our server is destroyed, the account record is deleted, stored recovery credentials are wiped, and the account's proxy is returned to the pool with a fresh IP. Deletion is retried automatically until the server confirms it.

No. We are a young company and have not been through a certification audit or an external penetration test yet. This page describes the real architecture instead; if your procurement requires a certificate, we are not ready for you yet.

Yes. Leads and lists export as CSV, and the guide on leaving any outreach tool lists what to take with you and in which order: the suppression list first, because it is the one with legal teeth. Cancelling from the billing page stops the next charge; the workspace locks at the end of the paid period and nothing is deleted on that day.

Yes, and on LinkedIn itself, not in the tool. The pending count belongs to the account. A new tool starts against whatever backlog is already there, and an account near the cap sends nothing until it drains.