Running a full go-to-market motion with a team of two
The work did not get smaller. The number of people who have to touch it did. Here is which parts actually move off a human's desk, and which never should.
The claim that two people can run a full go-to-market motion sounds like a headcount argument. It is not. The work is the same size it always was. The difference is how much of it has to pass through a person to get done.
We have watched enough teams try this to be specific about where the line falls. Some of it moves cleanly. Some of it moves and then quietly costs you a quarter.
What moves off a desk cleanly
The tasks that automate well share a shape: the inputs are already in a system, the output is checkable, and being wrong is cheap and reversible.
- Research before a first call: who they are, what changed recently, who else is in the account.
- Logging. Every touch, every outcome, without anyone remembering to.
- Follow-up drafting where a human still presses send.
- Noticing that a deal has gone quiet, and saying so before the forecast does.
- Reconciling what the CRM says against what actually happened in the inbox.
What does not, and probably never will
The tasks that resist automation share a different shape: the inputs are not written down anywhere, and being wrong is expensive in a way you find out about late.
- Deciding which segment to walk away from. The data supports several answers and the choice is a judgment about what kind of company you want.
- The first ten minutes of a hard renewal call, where the useful information is in what the customer is not saying.
- Pricing exceptions. Every one is a precedent, and precedents compound in ways a model cannot see.
Automate the work where being wrong is cheap. Keep the work where being wrong is a precedent.
The failure mode nobody warns you about
A two-person team that automates the checkable work does not run out of hours. It runs out of attention. Every automated action still generates something to review, and a queue of two hundred approvals is not leverage. It is the same work with worse ergonomics.
The fix is unglamorous. Decide up front which actions need a human at all, and let the rest run and report. If you find yourself approving the same class of action forty times without ever rejecting one, that class does not need approval. It needs a log.
Where to start if you are two people today
Start with logging and research, because they are pure recovery. Nothing you were doing well gets worse, and you get hours back in the first week. Add drafting once you trust the research. Leave anything that touches price, terms, or a relationship in human hands until you have a quarter of evidence saying otherwise.
The teams that get this wrong start at the end, with an agent that sends things. The teams that get it right spend the first month letting the system prove it understands the motion before it is allowed to participate in it.
