What the role actually does
A customer success manager owns the client relationship after the sale and outside the project. That last part is the distinguishing bit: a delivery lead owns whether the work is going well, and a CSM owns whether the relationship is, which are different questions with different evidence.
In practice that means a defined contact rhythm independent of project milestones, watching for signs an account is drifting, running the reviews that surface what the client actually needs next, owning renewals and expansion, and being the person a client raises something with before it becomes a complaint.
The role also carries a job most businesses never assign: knowing why clients leave. Not from exit conversations, which are polite fiction, but from patterns across accounts over time. That requires someone accountable for the question, and in most small businesses nobody is.
At smaller scale the title varies — account manager, client partner, relationship manager — and the distinctions matter less than the ownership. If nobody currently owns the client between projects, the gap is the same one regardless of what you call the hire.
Worth being clear about the boundary with delivery, because it is where these hires get muddled. If the same person owns both the project and the relationship, the project always wins — it has deadlines and the relationship does not. That is precisely the arbitration this role exists to remove.
When you genuinely should hire one
Hire when you have enough accounts that nobody can hold them all in mind. The threshold is lower than founders expect — somewhere past fifteen or twenty active relationships, attention starts being allocated by whoever is loudest, and past thirty it is entirely so.
Hire when the revenue is recurring or renewable. If most of your revenue depends on clients continuing rather than on new sales, the economics are unambiguous: a small improvement in retention outperforms a large improvement in acquisition, and nobody currently owns the variable.
Hire when the founder is the relationship. Clients who stay because of their connection to you are clients who leave when your attention moves, and building a relationship with somebody else in the business is one of the few durable protections against that. It has to start before you are stretched, not after.
As general market context, customer success roles in the US broadly run somewhere from the sixty to a hundred and ten thousand range depending on seniority, account complexity and whether expansion targets are attached. That is loose context rather than a benchmark and it varies a lot by sector.
Where the hire fails
It fails when they have no signal to work from. Twenty-five accounts and no data about which are drifting means they will work the accounts that contact them — which is the same allocation-by-urgency that caused the churn you hired them to prevent.
It fails on measurement, and it fails slowly. Churn is a lagging outcome with a long delay, so a CSM who is not preventing it looks fine for two or three quarters. Their reports will be impressions — this one feels good, that one has gone quiet — and impressions cannot be trended, checked or handed over.
It fails when the role has no authority. A customer success manager who notices a problem and can only escalate it is a very well-informed observer. They need to be able to commit something: time, a fix, a concession, a call with the right person.
And it fails through knowledge concentration, which is the version that hurts most later. Everything they learn about each account lives with them unless something records it. A successful CSM in an uninstrumented business quietly recreates founder dependency with a different name on it, and their departure costs you the entire accumulated model of every client.
And it fails when the metric is satisfaction. Satisfied clients leave routinely, because satisfaction assesses the deliverable and connection is a property of the relationship. A CSM measured on satisfaction scores will optimize the thing that does not predict staying.
What has to exist first
A health signal per account, assembled from things the business already produces: days since a human made contact, reply latency against that client's own baseline, work items past their promised date, whether any conversation this quarter was about something other than the deliverable. None of that is exotic and most of it is currently uncaptured.
A place for what they learn to live. What each client said in the last review, what they mentioned wanting next, who the real decision-maker is now. If that lives in one person's head, the role is generating an asset that walks out the door.
Renewal dates somewhere the business looks weekly, rather than in a contract. The renewal is the most engineerable moment in a client relationship, and a CSM who discovers dates eleven days out is negotiating rather than reviewing.
And a defined action per warning sign. A signal that produces a feeling of unease is not an instrument. Each one should name what happens and who does it, so the hire spends their judgment on the conversation rather than on whether to have it.
And a decision about what they can commit without asking — time, a fix, a concession, a call with the right person. Noticing a problem you cannot act on is a very well-informed form of powerlessness.
How to tell which you need
Name your three most at-risk accounts right now, and say what evidence you are using. If you can — formally or informally — you have a signal, and your constraint is capacity to act on it. Hire the CSM. They will arrive with something to work from and their first quarter will be useful rather than exploratory.
If you cannot, the hire inherits the blindness. Build the signal first: it is weeks rather than quarters, it costs a fraction of a salary, and it very often reveals that two accounts needed attention rather than twenty-five, which changes the size of the role you are hiring for.
Also check whether the work is relationship or coordination. If what you actually need is somebody to send updates, chase deliverables and keep clients informed, that is a delivery coordination gap and frequently a systems one — status that updates itself is cheaper and more reliable than a person producing it.
Where there are genuinely too many accounts for anyone to cover even with perfect instruments, you need both. The order still holds, because the signal is what makes the first ninety days of the hire productive instead of a survey.
One last check before posting the role: look at whether your last three departures had visible warning signs in hindsight. If they did and nobody saw them at the time, the gap is detection. If nobody could have seen them even with instruments, it is something else, and a hire will not find it either.
A CSM adds capacity to attend, not capacity to notice. If you cannot name your three most at-risk accounts and the evidence, the hire will work whoever emails loudest.