Multi-threading target accounts in client outbound: contact count, order, and account-level stop rules
Most agency sequences are built per contact. The prospect enters, gets four or five steps, and exits on reply or on the last step. That works until the client's ICP has buying committees, and someone decides the fix is to load three people from every account into the same sequence on the same Monday.
Then three near-identical emails land in one building. One person replies "not interested," and the other two get step three that afternoon. The account is burned, and the client's domain carries the complaints.
Multi-threading is worth doing for the right accounts. It just has to be run as an account motion, with an account-level stop rule, not as three unrelated contact sequences that happen to share a company name.
Why bother with more than one contact
The case for it is real. Artem Smirnov's breakdown, citing Gong's research on 1.8 million new business deals closed in 2024, notes that 77% of those deals were multithreaded and that won deals had twice as many buyer contacts as lost ones. Inbound Labs makes the same practical point for SaaS: deals have committees, so sequencing one contact per account leaves the budget owner or the person who has to integrate it out of the conversation.
Smirnov also adds the caveat most people skip. Those figures describe deals that already exist. A cold list has no deal yet. Nobody has agreed to talk. The research argues against one contact per account. It does not argue for emailing ten strangers at the same company on day one.
How many contacts, by account size
Smirnov's planning guidance for a first wave, before anyone has replied:
- Small, founder-led company, under about 50 people: 1 or 2 contacts, with 1 in reserve
- Mid-size company, about 50 to 500 people: 2 or 3, with 1 or 2 in reserve
- Large account, or one deal that could carry a quarter: 3 or 4 across two teams or locations, with 2 or more in reserve
FirstSales' cadence comparison lands nearby. It describes mid-market cadences as usually one or two contacts, and enterprise cadences as spread across three to five contacts, each getting a shorter individual sequence.
For an agency, put this in the client's campaign entry rules by segment. A client selling to 30-person agencies and a client selling to 3,000-person manufacturers should not share a default.
Stagger the starts
Do not start everyone on the same day. Smirnov's example calendar for one account:
- Day 1: person 1 gets step 1.
- Around day 4 or 5: person 1 gets step 2, person 2 gets step 1 with a different angle.
- The following week: both sequences continue on their own spacing, and nobody at the account gets two emails on the same day.
- After person 1's last step: if the account is still quiet, person 3 comes out of reserve.
FirstSales recommends spacing contact starts by three to five days per stakeholder, because five contacts hit on one day looks like a blast, raises complaint risk, and raises the odds of an internal forward that exposes generic copy.
There is a sending reason too. Google's sender guidelines ask senders to keep spam rates below 0.1% and never reach 0.3%, and to avoid sudden volume spikes. Tripling contacts per account overnight triples sends from the same client mailboxes and adds two more inboxes per account that can press "report spam."
Write a different email for each role
The offer stays the same. The angle changes. Smirnov's example uses three people at the same firm: the person doing the work gets a line about the manual task, the budget owner gets a line about cash or cost, and the person who would run the change gets a line about the rollout. Inbound Labs frames the typical split the same way: budget owner, daily user, and the technical or operations stakeholder who integrates it.
Three copies of one email in one building read as a mail merge. In an AI write step, that means the role has to be an input, not just a token. If the draft for the CFO and the draft for the ops manager differ only in the first name, reject it.
The account-level stop rule
This is the part that breaks in most tools, and it is the part that matters most.
Any reply from anyone at the account stops every sequence at that account. Positive, negative, out of office with a named colleague, "wrong person," all of it. Smirnov's rule is the same: the moment anyone replies, every other sequence at that company stops, and new names come from the conversation.
Then route by reply class:
- Positive: the replier becomes the thread. Other contacts get reached through them, not with more cold sends.
- Not interested: suppress the replier. Decide per client whether the rest of the account is paused for a set period or suppressed entirely. Write that decision into the client policy before launch, not after the first angry reply.
- Wrong person or referral: open a new contact record for the named person, tagged with where the name came from, and reference the referral in the first line.
- Opt-out: suppress immediately. For opt-outs written as a sentence, consider suppressing the whole domain for that client.
A per-contact "stop on reply" is not enough here. Sequences must stop on reply, and in a multi-threaded account the unit that replied is the company.
Keep your own senders off the same account
Smirnov's other warning applies to agencies more than anyone: once you contact several people per company, the chance that two of your own senders land in the same account goes up. A LinkedIn seat on one side, an email mailbox on the other, maybe the client's founder sending "a quick note" from their own inbox.
Assign whole accounts, not contacts, to a sender or seat inside the client workspace. Plan email and LinkedIn to the same person as one sequence from one owner. And if the client has its own reps prospecting, get their account list into the same suppression check before your first wave goes out.
What depth costs
Sending capacity per client is fixed in any given month. Smirnov's arithmetic is simple: at one person per company, 5,000 contacts covers 5,000 companies; at three per company, the same capacity covers about 1,670. For a narrow, high-value ICP that trade is worth it. For a broad market with a long list, one or two contacts per account opens more doors. Decide per client and per segment, and show the client the trade in plain numbers.
Where GTM fits
GTM runs find, qualify, write, send, reply, and book inside isolated client workspaces, on each client's email domains and LinkedIn, with send windows, limits, and sequences that stop on reply. Treating the account as the unit for reply handling and suppression is part of running multi-threaded outbound without burning a client's best accounts.
If you run outbound for several clients and want account-level rules built in, join the waitlist.
Sources
- Artem Smirnov, "How many people to contact at one company in outbound": https://smirnovartem.com/blog/contacts-per-company-outbound
- FirstSales, "Outbound cadence by deal size: mid-market vs enterprise touch counts": https://firstsales.io/blog/outbound-cadence-by-deal-size/
- Inbound Labs, "Outbound sequence for B2B SaaS that books demos": https://inboundlabs.app/blog/the-outbound-sequence-for-b2b-saas-that-books-demos
- Google Workspace Admin Help, "Email sender guidelines": https://support.google.com/a/answer/81126