How do you stop two client seats messaging the same company?

Two versions of the same collision

The first happens inside one client. Two seats work the same target account, four contacts get invitations in the same week, and what looks coordinated on your plan looks like a swarm from the recipient's side of it. Worse, the four messages usually say nearly the same thing, because they came from the same brief.

The second happens across clients, and agencies miss it because the campaigns live in different plans and different documents. Two of your clients sell to the same buyer, both lists were built well, and neither operator has any reason to open the other's roster. Nothing in either campaign looks wrong when reviewed on its own, which is exactly why it survives review.

The cost is the same in both cases. Invitations marked as unwanted, acceptance falling while volume holds, and the account absorbing the damage belongs to a real person you then have to explain it to.

The rule

One company, one seat, one calendar week. When two seats have a claim, the seat with the strongest one goes first: an existing relationship, a mutual connection, a prior thread, or simply the client whose offer fits that account better on the merits.

Prefer depth to breadth inside an account. One well-aimed contact outperforms four simultaneous invitations and leaves you somewhere to go next week. If the first contact does not accept within the week, the next attempt is a different person, a different angle, or a different seat's turn, and the account stays workable either way.

When claims genuinely tie, break it the boring way: whoever has the stronger reason to be in that person's inbox this month, written down in one sentence. If neither operator can write that sentence, the company is not ready for either seat, and the rule has just saved you an invitation you would have regretted.

How it is actually enforced

There is no switch labelled one company per week, and we are not going to describe one. What exists is a check the agent runs before enrolment: company interaction history for prior touches on any seat, and a search of active sequences for anyone already in flight at that company. A hit means hold or reroute, and the decision is a person's.

Suppression carries the rest. Importing a CSV makes that contacted universe suppression from the moment of import, so nobody receives a duplicate first touch, and a client's pre-existing outreach is visible to the checks from day one rather than from month three when someone finally digs out the export.

Marking makes a decision durable. A company can be marked so the agent treats it as held, and that mark lives in the workspace rather than in a chat message you lose two conversations later. The distinction matters: the check is mechanical, the judgment is not, and the mark is how a judgment survives the week.

Running the rule week to week

Run the enrolment check at list level rather than per person. Before a lead group goes live, look at the companies in it against what is already in flight across the workspace, and pull the overlaps out before anything is queued. Cleaning a list is free. Unsending an invitation is not.

Reconcile weekly, not monthly. The whole point of a one-week window is that held companies come back into play quickly, so a standing weekly pass over what was held, what accepted and what went quiet is the maintenance the rule needs. Held companies that never get revisited are just a list you are losing.

Give the rule an escape hatch you write down. A genuine warm introduction, an inbound request or an event follow-up should override the queue, and those exceptions should be recorded rather than argued each time. Otherwise the rule gets quietly ignored the first time it costs someone a meeting, and then it is not a rule.

What suppression can and cannot see

It sees everything sent through the workspace on any seat, plus whatever history you imported. That combination covers the overwhelming majority of agency collisions, because most of them are between two campaigns you are running yourself.

It cannot see a message a client sent by hand from their phone last quarter unless that history was imported, and it cannot see anything sitting in a CRM, because there are no CRM, Zapier or webhook tools in this build. Ask for the export during onboarding, not after the first complaint, and ask again when a client changes the person who owns the relationship.

One timing detail people get wrong: LinkedIn asks you to wait up to three weeks before re-inviting someone whose invitation you withdrew. Withdrawing an invitation to free a slot does not reset the clock in your favour, so it is not a workaround for a collision you caused.

Why this outranks volume as a discipline

Acceptance rate is both the outcome you want and the signal that protects the account. Volume holding steady while acceptance falls is one of the behavioural patterns that gets scored, and collisions push acceptance down from a direction campaign stats do not show you, because the damage lands on the recipient's side of the transaction.

The spread is large enough to be worth protecting. Two of our own audiences accepted at 32% and 2.6% with comparable copy, and across 861 conversations on five seats over three years the average was 41%. Those numbers came from working an account once at a time, not from sending more into it.