Customer Success Maturity Model: What Each Stage Hires

Every customer success maturity model asks you to grade your own behaviour. Are you reactive or proactive. Do you have playbooks. Is the team data-driven. The questions sound answerable until you try to answer them for your own team, and discover that your honest reply to most of them is "sometimes, depending who you ask."

That is not a flaw in how the models are written. It is a flaw in what they measure. Behaviour is invisible from outside a company and arguable inside it, which is why no two published models agree on the stages and why none of them can tell you what the distribution looks like across the market.

So we measured the one thing about post-sale maturity that leaves a public record. We read every live post-sale advert on our jobs board, 4,489 roles at 1,927 B2B software companies, with the full text of 4,090 of them, and grouped companies by how many distinct post-sale functions they are hiring into at once. Separating customer success from onboarding, or account management from technical account management, is a decision a company has to commit budget to. You cannot claim it in a survey.

What comes out is a four-stage ladder that is uncomfortable reading. Nearly three quarters of companies hiring post-sale roles are hiring into one function and nothing else. The dedicated operations role that published models put in the middle of the journey appears at one company in forty six. And the things those models actually grade you on, playbooks and health scores and dashboards, do not rise with maturity at all.

What a customer success maturity model is, and what it is for

A customer success maturity model is a staged description of how a post-sale organisation develops, usually from reactive firefighting to a measured, proactive operation. The stages carry names like ad hoc, managed, repeatable, proactive, predictive or optimised, and each one comes with a description of behaviours a team at that stage displays.

The purpose is sequencing. If you are deciding what to do next with a team of four, the model is supposed to tell you which of the fifteen things on your list are the ones that matter now, and which belong two years from now.

That is a real need, and it is why the format survives despite the problems below. A CS leader in planning season has to choose between hiring a second specialist, buying a platform, writing playbooks and building reporting, and the published answer to that question is almost always "do them in this order."

Where it goes wrong is the evidence. The sequence comes from the author's experience, not from a measurement, and the stage you place yourself in comes from self-assessment. Two leaders in identical organisations will place themselves two stages apart depending on temperament.

Why the published models disagree with each other

Go looking and you will find models with three stages, four stages and five stages, from consultancies, analysts and platform vendors. They are not variations on a standard. There is no standard.

The three-stage versions usually run reactive, proactive, predictive. The four-stage versions tend to add a distinction between having a process and having it work consistently. The five-stage versions borrow from software process maturity, which is where the whole format came from, and run something like ad hoc, managed, repeatable, measurable, optimised.

What they share is more interesting than how they differ. Almost all of them describe the top stage as predictive, data-driven and automated, and almost all of them describe the bottom stage as reactive and manual. The middle is where they diverge, and the middle is where every real team lives.

They also share an incentive problem worth naming. A large share of the models in circulation are published by companies that sell the software the upper stages require, and in those versions the upper stages describe a team that has bought a platform. That does not make the models wrong, but it does explain why the ladder so often has a product at the top of it.

The problem with grading yourself on behaviour

Try to use one of these models honestly and three problems show up immediately.

The first is that the questions are not falsifiable. "Do you take a proactive approach to risk" has no answer that can be checked. Every team does sometimes. The leader who says no is unusually honest rather than unusually immature, and the model rewards the opposite.

The second is that the evidence is self-reported, so the distribution is unknowable. None of the published models can tell you what share of companies sit at each stage, because nobody can observe behaviour across a market. Without a distribution you cannot tell whether being at stage two means you are behind or in the middle of the pack, which is the first thing anyone wants to know.

The third is that behaviour is the output, not the input. A team becomes able to work proactively because somebody was given the time and the remit to build the thing that spots risk early. The model describes the result and leaves out the structural change that produced it, which is the part a leader actually controls.

Structure is different. Whether a company has separated onboarding from ongoing management, or hired somebody whose whole job is the operation rather than the accounts, is a fact. It is budgeted, it is advertised, and it can be counted across a market.

The model you can audit: what a company separates

So here is the ladder, built from what companies are staffing rather than from what they say about themselves. A company's stage is the number of distinct post-sale functions it is hiring into: customer success, onboarding and implementation, account management, technical account management, and customer success operations.

Stage one, one function only, is 73.1% of companies hiring post-sale roles. Stage two, two functions, is 21.1%. Stage three, three functions, is 4.7%. Stage four, four or five functions, is 1.1%, which is twenty two companies out of 1,927.

The first thing to take from that shape is that differentiation is rare. If your post-sale organisation is one undivided team, you are not behind the market. You are the market. Nearly three quarters of it is in the same position, and the published models that place that team at stage one of five are describing the normal condition of the industry as a failure.

The second is that the gaps between stages are enormous. Moving from stage two to stage three means joining about six companies in a hundred. Moving to stage four means joining about one in a hundred. Any model that presents the climb as a steady progression every team should be making is describing something that almost nobody does.

The four stages, and what each one hires

The stages are not just counts. Each one has a characteristic shape, and the order in which functions appear is consistent enough to be useful.

Stage one: one function, usually customer success. Among the 1,408 companies here, 49.1% are hiring customer success, 23.9% onboarding and implementation, 22.2% account management, 4.0% technical account management and 0.8% operations. The median company at this stage has one open post-sale role. Worth noting that almost a quarter of stage one companies are hiring only onboarding, and another fifth only account management, so stage one is not always a CSM and nothing else. It is one post-sale idea, whichever one the company started with.

Stage two: the first split, usually onboarding. At two functions, 80.6% are hiring customer success and 57.0% onboarding, up from 23.9%. Account management reaches 49.9%. The median company has three open roles. This is the stage where a company decides that getting a customer live and keeping them are different jobs, which is the single most common first separation in the data.

Stage three: specialists appear. At three functions, customer success is 94.4%, onboarding 82.2% and account management 77.8%, and technical account management jumps to 34.4% from 9.6%. Operations reaches 11.1%. The median company has five open roles. The pattern here is that the generalist CSM role has been split along the axis the product demands, usually technical depth.

Stage four: the operation becomes a function. Among the twenty two companies at four or five, every one is hiring customer success and onboarding, 95.5% account management, 81.8% technical account management and 40.9% operations. The median company has ten open roles. This is a small and unrepresentative group, and the numbers below the top two rows are thin enough to read as direction rather than measurement.

The order matters more than the counts. Onboarding separates first, the technical specialist second, the operations role last. A team that buys an operations hire before it has separated onboarding is doing something unusual, and should be able to say why.

How to place yourself, in four questions

Four questions, all of which have checkable answers, unlike the ones in the published models.

Does one person own both getting a customer live and keeping them there? If yes, you are stage one on this measure, along with most of the market.

Is there somebody whose job title says onboarding or implementation, and who does not carry ongoing accounts? If yes, you are at least stage two. If the same person does both, you are not, whatever the title says.

Is there a role split out for technical depth or for commercial ownership, a technical account manager or a dedicated account manager owning renewals? If yes, stage three. The boundary between customer success and account management is worth getting right rather than assuming, and it is worked through in customer success versus account management.

Is there anybody whose whole job is the system rather than the accounts? Somebody who owns the data, the reporting, the tooling and the process, and who carries no book. If yes, that is the fourth separation, and it puts you in the top one percent of the market by this definition.

Answer honestly and most readers will land at one or two. That is the correct answer for most companies, and the rest of this page is about what to do with it rather than how to feel about it.

What changes between stages, and what does not

Here is the part that disagrees with every published model.

If maturity were really about becoming proactive, measured and data-driven, you would expect the language of adverts to change as companies differentiate. Mature teams should be asking for playbooks, health scores, reporting and forecasting at a higher rate than immature ones.

They are not. Across the 4,090 adverts read in full, one duty rises cleanly with stage: automation and working at scale, named in 37.4% of adverts at stage one, 44.7% at stage two, 48.9% at stage three and 54.0% at stage four.

Everything else is flat or falls. Playbooks are named in 30.1% of stage one adverts and 22.6% at stage four. Health scores fall from 6.0% to 1.8%. Reporting and dashboards go 30.1%, 27.7%, 35.4%, 17.9%, which is noise rather than a trend. Forecasting barely moves, 19.7% to 26.6%. Segmentation and tiering sit between 3.2% and 4.4% at every stage.

Two readings of that are possible and both are useful. One is that the vocabulary of customer success is now universal, so every advert says playbook and health score regardless of whether the team runs either, which makes those words worthless as a maturity signal. The other is that mature teams stop advertising for these things because the system already exists and the new hire is not being asked to build it.

Either way the conclusion for a leader is the same. Grading your team on whether it uses the vocabulary tells you nothing, because the vocabulary is evenly distributed across companies that look nothing like each other. The thin cells at stage four deserve care here: 274 adverts carry those percentages, so a difference of a few points is not a finding.

The operations threshold

One structural signal does rise, steeply, and it is the one the published models put in the middle of the journey rather than at the end.

A dedicated customer success operations role appears at 0.8% of stage one companies, 2.9% at stage two, 11.1% at stage three and 40.9% at stage four. Cut the same data by hiring volume instead of function count and the curve repeats: among companies with three or more post-sale openings the rate is 5.9%, at five or more it is 9.1%, and at ten or more it is 13.8%.

One number to reconcile, because we have published a different one. Our customer success operations page reports 29 of 876 companies hiring an operations role, which is 3.3%, against 2.2% across the 1,927 companies here. Different pulls, different dates and a wider company base explain the gap, and that page also counts a second group that names the function without hiring for it, which lands it at about one in eight on a deliberately broader measure. Treat that page as the authority on how many companies have the role and this one as the authority on where in a company's structure it appears.

That is the real threshold in post-sale organisations, and it arrives late. Operations is 1.07% of all post-sale hiring, which makes it the rarest of the five functions by a wide margin, behind technical account management at 5.66%.

The practical reading is that almost nobody has this role, so the work it describes is being done by somebody else, badly, in the gaps between their actual job. What that role owns, when to hire the first one and what to do if you cannot is covered in customer success operations, which is the page to read if your answer to question four was no and you suspect it should be yes.

Against that, the line chart in the previous section shows that naming a customer success platform does not rise with stage at all: 7.5%, 8.1%, 6.6%, 9.1%. Salesforce is named at a steady fifth of adverts at every stage, 21.2% to 23.3%. Buying the software that the top of the published ladders describes does not appear to be what separates a differentiated post-sale organisation from an undifferentiated one.

Does being more mature actually produce better retention?

Every published model implies that it does, and none of them shows it. This one cannot show it either, and the reason is worth stating because it applies to all of them.

A job board records structure. It records nothing about whether the customers stayed. Linking a company's stage to its retention would need outcome data that no public source carries, so any model claiming that stage three retains better than stage one is asserting it.

There is also a confound large enough to swallow most of the claim. Stage four companies have a median of ten open post-sale roles against one at stage one. They are bigger, better funded and usually later stage. If their retention is better, differentiation is one candidate explanation sitting among several stronger ones, and nobody separating those effects has published the work.

What survives the scepticism is a narrower mechanism. Separating a function does not improve an outcome by itself. It creates hours that were previously being taken from something else, usually from the proactive work that gets dropped first when a launch goes badly. The gain is capacity, and whether capacity becomes retention depends entirely on what the freed hours get spent on.

That is also the answer to which metrics belong at each stage, which is the other thing the published models hand you as a generic list of customer satisfaction, churn, retention rate, lifetime value and net promoter score. Those are company metrics. They do not change when you separate a function, and watching them for evidence that a structural change worked will tell you nothing for two or three quarters.

The stage-specific question is narrower in each case. At stage one, what is being dropped each week, and by whom. At stage two, how much information the new handoff loses, because a separation always creates one. At stage three, whether the specialists are now queueing on each other. At stage four, cost per retained dollar, which is the only point on this ladder where the organisation is complex enough for that number to mean anything, and which is set out in customer retention cost.

Leadership is the other thing that moves

The second structural signal is who is being hired to run it.

The share of companies hiring at least one director-level or executive post-sale role goes 11.7% at stage one, 24.3% at stage two, 36.7% at stage three and 54.5% at stage four. It doubles between stage one and stage two, and triples by stage three.

This is less surprising than the operations finding, but it is worth stating because of what it implies about sequence. Separation and seniority arrive together. A company that splits onboarding out of customer success usually does it around the same time it hires somebody senior enough to decide where the boundary sits.

For a leader at stage one with no senior post-sale voice above them, that is the honest read on why the structural change keeps not happening. The decision to separate a function is made a level above where the pain is felt.

Maturity is measured on four axes, and this page is only one of them

Worth being explicit, because the word maturity gets used for several different things and the stage counts do not match.

Structure, which is this page: how many post-sale functions exist, and whether anybody owns the system rather than the accounts. Four stages.

Coverage: how accounts are assigned to people, from everyone getting the same treatment through to a deliberate mix of high touch, pooled and digital tiers. That is a different ladder with its own stages, and it is in digital customer success.

Tooling: what a team should be running at its scale, which is a three-stage ladder in customer success platforms for startups, built for the buying decision rather than the org design one.

Value proof: how well the team can demonstrate the outcome it produced, which is the ladder in the value realization framework.

These move independently, which is the useful part. A single-function team can run a sophisticated coverage model and prove value well. A four-function organisation can be crude at both. If you are using one of these as a proxy for the others, you will get the next decision wrong.

What moves you up a stage, and what does not

Four things move a company up this ladder, based on the order the functions appear in.

Separate the moment the two jobs start competing for the same hours. Onboarding separates first in the data because it is the clearest case: a new customer needs attention on a schedule the renewal book does not respect. If your CSMs are dropping check-ins to run implementations, that is the signal, and it arrives well before anybody feels ready. The seven-stage flow in customer onboarding best practices is the practical version of what gets separated.

Split along the axis your product demands, not the axis the org chart suggests. Technical account management appears at stage three because some products need depth more than they need coverage. If your escalations are technical rather than commercial, that is your next split regardless of what a generic model recommends.

Hire the operations role once the system is bigger than any one person's memory. The data puts this last, and the volume cut suggests the trigger is somewhere around five open post-sale roles, which is a proxy for a team large enough that the process no longer fits in the lead's head.

Fix coverage before fixing structure. If the problem is that each CSM holds too many accounts, separating functions will not help, and the answer is a coverage model rather than a new function. The ratio to start from is in the accounts per CSM calculator, and the tiering decision itself sits in the coverage model playbook.

What does not move you up, on this evidence, is buying a platform. The platform mention rate is flat across all four stages. That is not an argument against buying one, and the case for a platform is a cost argument rather than a maturity argument, which is worked through in the customer success platform business case. It is an argument against treating the purchase as a stage promotion.

The cost of climbing, which no model states

Every stage above one is a headcount decision, and headcount is the whole cost of a post-sale organisation.

A separated function is at minimum one more loaded salary. The loaded cost of a single customer success manager, with the multiple applied and the assumptions shown, is worked through in cost to serve per account. Adding a function is that number again, and the second one does not come with a second book of revenue attached.

This is why stage one is so crowded and why the climb flattens so sharply after stage two. It is not that companies do not know the model. It is that every rung costs a salary, and the return on the operations hire in particular is indirect: it makes other people more effective rather than carrying revenue itself.

If you are building the case for the next rung, build it as a cost and coverage argument rather than a maturity argument. What the whole retention effort already costs, and how to work out whether another head changes it, is set out in customer retention cost.

Use this table once, then stop grading yourself

One table, to make the call and move on.

The right-hand column is the one most teams need. At every stage the common mistake is to buy the artefact that the next stage is supposed to have rather than make the structural change that produces it.

Most teams will stay one function for years. That is not a failure.

When a lower stage is the right answer

Three situations where staying at stage one or two is correct.

Your product does not generate two distinct jobs. If implementation takes two days and the ongoing relationship is light, separating onboarding creates a handoff where none was needed, and handoffs lose information. The split should follow the work, not the model.

You are small enough that the coordination cost exceeds the specialisation gain. Two people who each do everything will usually beat two specialists who each wait on the other. The separation starts paying somewhere above that, which is why the median stage two company has three open roles rather than one.

Your churn is concentrated somewhere a new function will not reach. If accounts leave because the product has a gap, a new post-sale function adds cost without touching the cause. Structure is the answer to a capacity and ownership problem, not to a product problem.

The case where a low stage is genuinely a problem is the quiet one: one team doing four jobs, nobody owning the system, and a leader who cannot say which of the four is being dropped this week. That is the state this measurement exists to make visible.

What reading maturity from hiring cannot tell you

The limits, because the model above is built on one kind of evidence and it has edges.

Hiring is a snapshot of what a company is adding, not a census of what it has. A company with a mature, fully separated post-sale organisation that happens to be hiring one CSM this month reads as stage one here. That makes the ladder reliable upwards and weak downwards: a company hiring into four functions at once is demonstrably differentiated, while a company hiring into one might be early, or might simply be quiet.

The sample is also skewed towards companies that are hiring at all, which in a given quarter is a minority of the market and tends towards the ones that are growing.

Stage four is twenty two companies. Treat every number attached to it as direction rather than measurement, including the 40.9% operations figure, which is nine companies.

And the duty rates are mention rates. They say what an advert asks for, not how an hour gets spent. A team with excellent playbooks that does not mention playbooks in its adverts counts as not mentioning them, which is part of why we read the flat lines as evidence that the vocabulary is worthless rather than as evidence that mature teams do less.

Those limits are why the four questions above are framed around your own organisation rather than around a percentile. The distribution tells you what is normal. Only you can see your own structure.

The ladder is shorter than it looks, and the rungs cost salaries

Most teams meet a maturity model while looking for a sequence, and leave it with a grade instead. The grade is worth very little, because it is self-assessed against behaviours everybody performs sometimes, and because nobody can tell you what the distribution is.

The structural version is less flattering and more useful. Three quarters of the market has separated nothing. The artefacts the published models grade you on, the playbooks and the health scores and the platform, are spread evenly across companies at every level of differentiation, so having them proves nothing. What actually distinguishes a differentiated post-sale organisation is that somebody decided a second job existed and paid for it, and eventually that somebody should own the system rather than the accounts.

Decide which of those two is your next move, write down why, and ignore the grade.