Taking an assessment rather than buying one? This page is written for employers. Here is the page for candidates.
Customer success and account management · Senior level
How to assess a Client Partner
This seat is hired almost entirely on chemistry, and chemistry is precisely the trait that produces its characteristic failure. The client partner who is adored by the client and whose accounts run at negative margin is a familiar figure in every agency, and their interview is outstanding. The commercial half of the job — noticing that the fourth small favour this month is a project, and saying so while the client is still in a good mood — appears in no current step, and it is the whole difference between a profitable account and a busy one. The second unobserved thing is which direction the candidate faces under pressure. Given a client demanding something the delivery team cannot do, some candidates advocate for the client internally and some construct a third option; the interview cannot separate them because both describe themselves as the client's champion. Third, unlike a software renewal, a services relationship regularly has to survive work that was genuinely bad, and the recovery conversation — accepting fault precisely, without conceding scope or fee, and putting a specific remedy in writing — is a narrow, observable skill that no portfolio or client reference reveals.
A client partner sits between a paying client and the team that does the work, and is accountable for both the relationship and the margin. That dual accountability is the entire character of the role and the reason it does not share a mark scheme with a software customer success manager. In a subscription business, cost of service is roughly fixed and the CSM's job is adoption. In a services business the cost is the team's hours, which means every concession made in a client conversation converts directly into someone's evening, and an account can be simultaneously beloved, renewing, and losing money.
Day to day the work is status calls, a weekly negotiation about priorities, one or two escalations, a stream of requests that arrive informally — "while you're in there, could you also..." — and a periodic conversation about next year's retainer. The requests are the crux. Individually each is small, refusing any one of them looks petty, and the aggregate is how service businesses lose their margin. The behaviour that separates the top quartile is not refusal; it is conversion. A strong client partner receives the fourth informal request and says, in the same friendly register, that these have added up to about a fortnight of work and proposes handling them as a scoped addition — which the client, who has not been counting either, usually accepts. A median client partner says yes, absorbs it, and mentions the strain to their delivery lead instead of to the client.
The second separator is the direction of advocacy. Under sustained client pressure, some client partners begin to represent the client's interests inside their own company, which feels like loyalty to the relationship and is in fact the beginning of the end of the account, because the delivery team stops trusting them and the client stops meeting resistance. The strong version holds a third position: name the constraint honestly to the client, name the client's real need honestly internally, and construct an option neither side had.
The third is recovery. Agencies ship bad work sometimes. The conversation that follows is a specific craft — accept the fault in precise terms rather than general contrition, avoid the reflexive offer of free work that resets the fee permanently, and commit to a named remedy with a date and put it in writing the same day. It is learnable, it is observable, and it is almost never assessed before hire, because it requires putting a candidate in front of a client who is genuinely displeased rather than in front of a panel that is trying to be welcoming.
What the job actually needs
- holding scope without souring the relationship
- converting an out-of-scope request into a change order in the same conversation
- representing a delivery constraint as a shared problem rather than a refusal
- recovering a relationship after work that was genuinely poor
- growing a retainer with no obvious sales motion
How people fail in this seat
- goes native and argues the client's case against their own delivery team
- absorbs small favours until the account is unprofitable
- apologises for a delivery failure with free work
- learns about a competitive review when the brief arrives
- sells work the team cannot staff
What most employers do instead
A portfolio of client names on a CV, a chemistry-led interview, sometimes a credentials or case-study presentation, and references from former clients.
The assessment
About 45 minutes end to end.
The systems it runs in
A professional services automation system, because in this seat the commercial truth is not in the CRM — it is in hours booked against a fee. t1's creep log is the account as a PSA presents it: the retainer value, the hours budgeted by role for the month, the time actually entered against the account by role and rate, and the eleven informal requests sitting as unbudgeted work with rough estimates against them. Kantata's published material describes PSA as the layer that adds resource management, margin tracking and forecasting on top of project management, and that is precisely the layer the question "is this account profitable" needs and a project tool alone cannot answer. The remedy in t3 is not described in prose either: the scoped addition is raised as a change order against the account, with hours, roles and a value, in the same system that will bill it. A candidate who writes a warm client note and raises nothing has left the account exactly where they found it, and the record shows it.
- Kantata
- Certinia PSA
- Rocketlane
- Scoro
- Harvest
- HubSpot Sales Hub
Any system holding a fee, budgeted hours by role, actual time entered, and a scope-change or change-order object. Agencies run this on everything from a full PSA to a timesheet tool plus a spreadsheet, and the fixture works on either as long as time is attributable to a rate — without rates there is no margin question and t1 collapses into a workload complaint, which is the exact failure the task is built to distinguish itself from. Rebuilt against the buyer's own role rate card and change-order template where they can provide a sandbox.
What the candidate actually does
| Task | What happens |
|---|---|
| The creep log data_task · 12 min | Six weeks of an account's records — the retainer value, the hours budgeted per month, the hours actually booked by role, and a log of eleven informal requests that arrived by email, in status calls or in a chat channel, each with a rough estimate against it. Individually every request is trivial and refusing any one of them would look petty. Together they are close to a fortnight of senior time and the account has moved from healthy margin to losing money, which no single line reveals. The candidate answers whether this account is profitable, which requests are the problem and why, and what they propose to do about it. The fork is between reporting that the team is under pressure and establishing that the account has an unrecognised project inside it. |
| The recovery conversation live_call · 18 min | An AI client marketing director. The agency shipped work last week that was genuinely poor — it missed a stated part of the brief and arrived two days late — and the client is unhappy in the controlled way senior clients are unhappy. Two things are hidden. There is an internal review of agency spend next month and this person needs something they can defend the relationship with, which they will reveal if the candidate asks what pressure they are under rather than only apologising. And inside their complaint sits a request for additional work as a remedy, which if granted quietly becomes the twelfth item on the creep log. The forks are general contrition against precisely accepted fault with a dated remedy; the reflexive offer of free work against a remedy that does not reset the fee; and whether the candidate raises the accumulated requests in this conversation or decides the timing is wrong and says why. |
| Two notes, in opposite directions written_artifact · 15 min | The note that goes to the client the same day — fault accepted, remedy named and dated, and the accumulated requests proposed as a scoped addition in a register the client will accept rather than resent, raised as a change order against the account with roles, hours and a value on it rather than described in prose. And a short internal note to the delivery lead whose team did the poor work and is already over its hours. The pairing is the test. A client partner who has gone native writes a warm client note and an internal note that reads as an instruction to work harder; one who is defensive writes the reverse. The third position — honest about the failure to the client, honest about the client's real need internally, and constructing an option neither side had — has to be visible in both notes at once. |
The mark scheme
Each criterion is scored 1 to 5 against written anchors, and every score is reported with the excerpt that earned it. A criterion marked floored is reported as a finding rather than averaged into the total. The first is open; open any other to read its anchors in full.
Converts accumulated requests into a scoped addition, in registerweight 0.25States what the requests have added up to in a specific quantity, does it in the same friendly register as the rest of the conversation, and proposes …
Accepts fault precisely without conceding feeweight 0.2States exactly what was missed and what was late, without excuses and without expanding the fault beyond what happened, and makes clear that the remed…
The remedy is specific and datedweight 0.15Names the corrected deliverable, the date it lands, who owns it, and the one change being made so it does not recur — and the same commitment appears …
Holds the third position rather than advocating either sideweight 0.2Names the constraint honestly to the client as a shared problem, states the client's real need honestly to the delivery lead without editorial, and pr…
Establishes the account's actual profitabilityweight 0.1States that the account is running at a loss, attributes it to the accumulation rather than to any single request, and identifies which requests accou…
Reads what the client is under pressure aboutweight 0.1Surfaces the upcoming spend review, and shapes the remedy so the client has something concrete to take into it.
How it is scored
Weighted mean of the six criteria, 1-5 against the anchors, each attached to the excerpt that earned it. Two rules specific to this seat. The two written notes are always reported side by side rather than sequentially, because the contradiction between them is the evidence and it is invisible when they are read apart. And any offer of unpaid work made in the call is reported verbatim with the sentence that preceded it, since the reflexive discount-as-apology is the single most expensive habit in this role and it typically appears within the first four minutes.
Integrity
- the spend review and the embedded extra-work request are absent from the brief and surface only under specific prompts
- per-candidate randomisation of the request log so the accumulation sits in different work types and totals
- keystroke and paste timing on both notes
- the notes are written against the candidate's own call, and a follow-up question asks which of the eleven logged requests they judged to be the largest and why
- identity continuity across the data task, the call and the written notes
The log describes what happened. It does not produce a cheating verdict — the follow-up conversation is the control, because a statistical accusation is not something we would ask a reviewer to defend.
What you receive
- the profitability answer with the candidate's reasoning and their nominated problem requests
- full call transcript with any offer of free work and any scope conversion marked
- both notes as written, presented side by side
- per-criterion score with the excerpt that earned it
Who decides
Recommended, and pointed at one thing above all. A reviewer reads the paired notes for every shortlisted candidate, because the direction-of-advocacy criterion is the one this design exists to expose and it is a judgment about two documents rather than a countable event. Two corrections a person should be prepared to make. A candidate may reasonably decide that the recovery conversation is the wrong moment to raise accumulated scope, and defer it while saying explicitly when and how they will — that is a defensible commercial judgment which the anchors mark down, and a reviewer should credit it where the deferral is specific rather than avoidant. And an agency with a genuine service guarantee may want an offer of credit to score well rather than badly, which is a legitimate configuration change to make before the assessment runs rather than an override afterwards. All overrides carry a written reason. The composite ranks; a person decides.
What this does not measure
This design does not score charm, likeability, chemistry or rapport, and that exclusion is the reason it exists. The role page's argument is that chemistry is both the trait this seat is hired on and the trait that produces its characteristic failure — the beloved client partner running an unprofitable account — so a mark scheme that rewarded it would reproduce the problem it is meant to solve. Nor does any criterion score accent, dialect, fluency, pace or manner of speech, directly or through composites such as "polish", "presence" or "client-facing gravitas", which in an agency context are near-universal euphemisms for class and speech markers and must not be added. What is scored is what was quantified, what was proposed, what was committed to and by when, and what the two notes say. Both notes are judged on whether their named reader could act on them, never on idiom or register beyond the single behavioural requirement that the scope conversion is made in the same friendly tone as the rest of the conversation — which is scored as whether the request was framed as an accusation, not as a question of style. Non-native phrasing is not a deduction. The data task requires arithmetic rather than finance training, and no sector or agency experience is required, since the brief supplies the account. What forty-five minutes cannot see is the relationship. A client partner's performance is a year of accumulated small decisions, and margin erosion of the kind this design puts in a log takes months to build in reality. The session observes one recovery conversation, one reading of an account, and one pair of notes. It gives no evidence about whether the candidate holds scope consistently on the fortieth request rather than the eleventh, about their ability to grow a retainer over a year, about how they behave when the delivery team is genuinely at fault repeatedly, or about new-business capability, which is a separate skill this design does not touch. Deployers should back-test against account gross margin and retainer growth over four quarters, and should drop any criterion that does not predict either. The live-call format may disadvantage candidates with speech, hearing or anxiety-related disabilities. A text-based run of the identical conversation, with the same hidden pressures and release conditions, must be available on request, scored on the same anchors and unmarked on the result. Extra time on the data task and the notes must be available without disclosure.
The reason this role gets its own design rather than sharing one with the customer success manager is a difference in cost structure that changes what should be observed. In a subscription business the cost of serving an account is broadly fixed, so the customer success manager is assessed on adoption and risk. In a services business the cost is the team's hours, which means every concession made in a client conversation converts directly into someone's evening, and an account can be simultaneously beloved, renewing and losing money. That last sentence is the whole assessment. Two of the three tasks exist to make it visible.
The creep log is the least dramatic task here and possibly the most predictive. Eleven requests over six weeks, each small enough that raising it would look petty, and an account that has quietly crossed from profit into loss. The role page's claim is that the separating behaviour is not refusal but conversion — the strong client partner receives the fourth informal request and observes, in the same friendly register, that these have added up to about a fortnight of work, and proposes handling them as a scoped addition, which the client, who has not been counting either, usually accepts. Testing that requires giving the candidate something to count. A candidate who reads the log and concludes the team is stretched has described a resourcing problem. A candidate who concludes there is an unrecognised project inside the retainer has described a commercial one, and those are different people.
The recovery conversation is included because a services relationship, unlike a software renewal, regularly has to survive work that was genuinely bad. The scenario does not hedge this: the agency missed part of the brief and delivered late, and the client is right. That is uncomfortable to design and it is the only honest version, because the skill under examination is what a person does when they have no case. The reflexive move — an offer of free work, a credit, a discount on the next invoice — feels like accountability and permanently resets the fee, and it usually appears within four minutes, which is why the design extracts it verbatim whenever it occurs. What the top anchor describes instead is narrow and learnable: name exactly what was missed, do not expand the fault beyond what happened, and put a corrected deliverable with a date and an owner against it.
Two hidden elements sit inside the call, and both reward curiosity rather than technique. The client has an internal spend review next month and needs something to defend the relationship with, which reframes the entire conversation for a candidate who asks what pressure they are under — the client is not looking for an apology, they are looking for ammunition. And buried in the complaint is a request for extra work as a remedy, which is the creep log's twelfth entry arriving live. Granting it quietly is the most natural thing in the world at that moment, and it is the failure the first task was designed to make legible.
The paired notes are the design's sharpest instrument. The role page identifies direction of advocacy as the thing an interview cannot separate, because every candidate describes themselves as the client's champion and the ones who have gone native describe themselves that way most convincingly. Asking for two documents written within minutes of each other, to audiences with opposed interests, makes the position falsifiable. A candidate facing the client writes one thing; a candidate facing their own team writes another; a candidate holding the third position writes two notes that are consistent with each other and with what actually happened. Reading them side by side takes a hiring manager about ninety seconds and is worth more than any reference from a former client.
Sources
Every figure on this page is traceable. Where a claim could not be sourced it is stated qualitatively instead.
- US Bureau of Labor Statistics, 2018 Standard Occupational Classification Definitions (contains no 'client partner' or 'account manager' occupation title; the seat is not separately measured in US occupational statistics), https://www.bls.gov/soc/2018/soc_2018_definitions.pdf
See what the employer actually receives. A full report for one role, with every score shown beside the excerpt that earned it, conduct findings reported rather than averaged, and a reviewer sign-off required before any decision. No form.
Read a sample reportOr talk to us about this role