Taking an assessment rather than buying one? This page is written for employers. Here is the page for candidates.

Claims and policyholder service · Entry level

How to assess a Policy Servicing Advisor

Screening this seat as a generic customer service seat misses the one thing that makes it different, which is that an ordinary-sounding sentence spoken here changes what the insurer owes. "Yes, that's covered" is a statement about a contract, and if it is wrong the insurer will very often be held to it or will pay a complaint to avoid the argument. Nothing in a generic screen observes whether the candidate distinguishes between describing what the policy says and confirming that a specific situation falls within it — a distinction every experienced servicing advisor makes reflexively and no new hire makes at all. The second unobserved behaviour is elicitation. Customers volunteer material changes sideways: they mention the new job, the lodger, the business use, the licence endorsement, in passing, as small talk, while the advisor is updating an address. Whether the advisor picks that up determines whether a claim is paid two years later, and it is a live, observable, single-turn behaviour that a competency interview cannot reach.

Policy servicing is the whole of the insurance relationship that is not a claim: address and vehicle changes, mid-term adjustments, adding a driver or a named insured, renewal conversations, cancellations, duplicate documents, questions about what the policy actually does. It is quiet work with an unusually long fuse. Almost nothing that goes wrong here reveals itself on the call; it reveals itself at the next claim, when a coverage position turns out to rest on something an advisor said eighteen months earlier, or on a disclosure the customer believes they made and the file does not record.

That fuse is why this seat sits in a regulated family and is marked regulated here. The NAIC's Unfair Claims Settlement Practices Act prohibits knowingly misrepresenting relevant facts or policy provisions relating to the coverages at issue to insureds as well as claimants, and the servicing desk is where insureds are told about coverage far more often than the claims desk is. The failure is rarely deliberate. It is an advisor who wants to be helpful, faced with a specific question — will my policy cover me if I start using the van for deliveries at weekends — and who answers it from memory of the product summary rather than saying accurately what they can and cannot confirm. The correct behaviour is a learnable, gradeable, three-part move: state what the wording says, state that whether this particular situation falls inside it is a coverage question, and route it. Most candidates who fail this do not know they have failed; they think they have given good service.

The second regulated boundary is the line between servicing and advice, and it differs by jurisdiction while the behaviour does not. A servicing advisor is generally permitted to give factual information and not permitted to recommend that a particular cover is or is not suitable for this customer. The pressure to cross that line comes from the customer, every time, in the same words: "what would you do?" The advisor who answers it has stepped into a regulated activity their firm may not be authorised for in that seat, and has done so warmly and helpfully in the middle of an otherwise unremarkable call. In a simulation this is a designed fork with a clean binary reading; in a competency interview it is unreachable.

The third competency is elicitation, and it is the one that most affects claims outcomes downstream. Underwriting depends on material facts, and material facts arrive in servicing calls disguised as conversation. The customer changing their address mentions that their son has moved in. The customer updating a phone number mentions the new job that comes with a company car. A good advisor hears the second half of that sentence, asks the follow-up, and records it; an average one processes the transaction that was requested. Both calls sound identical to a listener who is not paying attention to the content, which is exactly why a transcript scored against a defined disclosure planted in the scenario is a better instrument than any interview.

Savvanta assesses this with a live call carrying two planted forks — a direct coverage question the advisor should not answer definitively, and an incidental disclosure that changes the risk — followed by the servicing note and the confirmation the customer receives. The rubric scores whether the coverage question was routed rather than answered, whether the disclosure was picked up and recorded, and whether the note would let an underwriter or a future claims handler see what the customer said and when. It does not score product knowledge the candidate has not yet been trained on, and it does not score accent, dialect or manner of speech.

What the job actually needs

explaining a premium or coverage consequence the advisor did not cause
eliciting a material change of circumstance the customer does not think is relevant
saying accurately what a policy does and does not cover
staying inside the boundary between servicing and advice
documenting what the customer disclosed and when

How people fail in this seat

confirms cover that the wording does not provide
soft-pedals a premium increase into vagueness so the call ends pleasantly
processes a mid-term adjustment without probing the change that triggered it
gives what amounts to advice on whether a cover is suitable
records the transaction but not the disclosure the customer made in passing

What most employers do instead

A generic contact-centre screen — CV sift, phone screen, sometimes a data-entry test — with insurance-specific knowledge treated entirely as a training matter.

The assessment

About 28 minutes end to end.

The systems it runs in

A policy administration system on the servicing side: the policy in force with its wording and schedule, the mid-term adjustment or endorsement transaction with an effective date and a premium recalculation, renewal and cancellation transactions, the servicing note against the policy, and the correspondence step that produces the customer's confirmation. Guidewire describes PolicyCenter as automating policy administration tasks from quoting and underwriting through endorsements and renewals across the policy lifecycle; Duck Creek describes its policy product as running in-force policies through automated policy workflows. The address change in this scenario is one of those transactions, and the three forks are planted inside it precisely because that is where they occur in production.

Guidewire PolicyCenter
Duck Creek Policy

Any policy administration or agency management system with a policy record, a mid-term change transaction with an effective date, a servicing note attached to the policy, and generated customer correspondence. The wording extract stays in the brief as it is now, so no product or insurance knowledge the candidate has not been trained on is scored, which matters for buyers hiring from outside insurance. Rebuilt against the buyer's own wording and transaction set where they supply a sandbox.

Working speed is scored. Handling time is the real constraint in this seat and the brief already puts fourteen calls in the queue, so the session measures and reports call length. It is reported beside the criteria and annotated as not a positive signal, because in this scenario the shorter call is the one where the van was never asked about — which is the whole reason the queue is in the brief at all. It is not a rubric criterion and it is not weighted, and the caution in the adverse_impact section stands unchanged: a deployer who lifts average handling time out of this report and runs it as a hiring metric in its own right is reintroducing the incentive this design tests resistance to.

What the candidate actually does

Task	What happens
An address change with three forks in it live_call · 12 min	An AI customer calls to change the address on a motor policy after a move. The candidate's brief includes the relevant wording extract and states that fourteen calls are waiting, because the pressure in this seat is handling time and a scenario without it measures nothing. Three forks are planted inside an otherwise unremarkable transaction. The customer mentions, while explaining the move, that the new job comes with a van they bring home at weekends, and never returns to it. A little later they ask directly whether using the car for weekend deliveries would be fine on this policy, which is a question about a specific situation rather than about the wording. And if the candidate declines to confirm, the customer asks what the advisor would do in their position, which is the sentence that pulls a servicing call across the line into advice. Each fork has a cheap path that is faster, friendlier and produces a happier customer on the call.
The note and the confirmation written_artifact · 9 min	The mid-term change as the candidate processes it, the servicing note recorded against the policy, and the confirmation generated for the customer, written back to back and scored against the candidate's own transcript. Recording the note against the policy rather than as a loose document is what gives the fourth criterion something to observe: the person who reads it in two years is an underwriter or a claims handler opening this policy, and a disclosure that reached a notepad but not the policy record is, for them, a disclosure that never happened. The cheap note reads address updated. The note that does its job records the disclosure the customer made in passing, in the customer's own terms, with the date it was disclosed, records the coverage question that was asked and where it was routed, and the confirmation repeats to the customer what was and was not confirmed, so that nobody can later rely on something that was never said.
Four sentences, three categories judgment_scenario · 6 min	Four single sentences an advisor could say, with their surrounding context. For each, the candidate marks whether it describes the wording, confirms cover for a specific situation, or recommends, and writes what should be said instead where it crosses a line. The items are close together on purpose; two of them differ by a single clause, which is the distinction every experienced advisor makes reflexively and every new one misses.

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.

Distinguishes describing the wording from confirming cover for this customer's situation (NAIC Unfair Claims Settlement Practices Act, Model 900, section 4.A, knowingly misrepresenting to claimants and insureds relevant facts or policy provisions relating to coverages at issue)weight 0.3States what the wording says, states plainly that whether this specific use falls inside it is a coverage question they cannot decide, and names who w…

1 Answers yes or no about whether the customer's proposed weekend delivery use is covered, or confirms it by agreeing with the customer's own summary of it.

3 Hedges with no content, such as saying they could not say or that the customer should check the policy, leaving the customer to guess and giving them no route to an answer.

5 States what the wording says, states plainly that whether this specific use falls inside it is a coverage question they cannot decide, and names who will answer it and by when.

Elicits and records the material change mentioned in passingweight 0.25Asks at least one follow-up establishing when it started and how the vehicle is used, tells the customer why it matters to their policy, and records i…

1 The van is never mentioned again in the transcript and appears nowhere in the note; only the requested transaction is processed.

3 Acknowledges it conversationally, processes the address change, and the note records the address change only, so the disclosure exists nowhere in the file.

5 Asks at least one follow-up establishing when it started and how the vehicle is used, tells the customer why it matters to their policy, and records it in the note with the date it was disclosed.

Holds the boundary between servicing and adviceweight 0.15Declines the recommendation, sets out the factual difference between the available options without saying which to choose, and offers the route to som…

1 Answers the question about what the advisor would do with a recommendation about which cover the customer should take.

3 Declines with a bare statement that they are not allowed to advise, and offers the customer nothing further.

5 Declines the recommendation, sets out the factual difference between the available options without saying which to choose, and offers the route to somebody who can advise.

The note is written for the person who reads it in two years (Model 900 section 4.A, since the file is what determines whether a later coverage position rests on something an advisor said)weight 0.2Records what the customer said close to their own words, the date, the question they asked, where it was routed, and what the advisor did and did not …

1 Records the transaction only. A future reader cannot tell that the customer disclosed anything or asked anything.

3 Records the disclosure but paraphrases it into the advisor's own language, losing what the customer actually said, and does not date it.

5 Records what the customer said close to their own words, the date, the question they asked, where it was routed, and what the advisor did and did not confirm.

The confirmation tells the customer what was and was not confirmedweight 0.1States the change made, states that the coverage question was not answered on the call, says who is answering it and when, and repeats the disclosure …

1 Confirms the address change and nothing else, so a customer who believes they were told the delivery use is fine has nothing contradicting them.

3 Mentions that the query was passed on, without saying what was not confirmed or when an answer will come.

5 States the change made, states that the coverage question was not answered on the call, says who is answering it and when, and repeats the disclosure the customer made so they can correct it if it is wrong.

How it is scored

Weighted mean of the five criteria, each scored 1 to 5 against the anchors, reported with the transcript excerpt or document line that earned each score. Because the first criterion carries the largest weight and is close to binary in practice, the report also states plainly whether the candidate confirmed cover for the specific use, as a yes or no with the excerpt, so that a hiring manager does not have to infer it from a weighted average.

Integrity

monitored session with the call and both documents in one unbroken sitting
the note and confirmation are scored against the candidate's own transcript, so a stored template cannot record a disclosure that was never elicited
one live follow-up question asking the candidate what the customer said about the vehicle
scenario variants rotated so the planted disclosure and the coverage question differ between sittings

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

full call transcript with the disclosure, the coverage question and the request for a recommendation time-marked
the policy as the candidate left it, being the change processed with its effective date, the servicing note recorded against it, and the customer confirmation as generated
the four judgment answers with the corrected wording the candidate supplied
per-criterion score with the excerpt that earned it

Who decides

Recommended, and required if the assessment is used as a cut rather than as a ranking. One judgment needs a person: whether a particular formulation described the wording or confirmed cover, because candidates produce borderline sentences that a reviewer can place in a few seconds by reading what surrounded them and an automated scorer places badly. The reviewer looks at the two time-marked exchanges and at the note, and confirms or overrides with a written reason. The ranking entitles the buyer to conclude that this candidate did or did not pick up a disclosure that was in front of them, and did or did not confirm cover they were not in a position to confirm, in this call. It does not establish product knowledge, which the brief supplied and which is training, and it does not establish that the candidate is authorised under the employer's own regulatory permissions to do anything, which is an employer gate.

What this does not measure

This is a voice role and nothing in the rubric reads accent, dialect, fluency, vocabulary or manner of speech. The criterion at greatest risk of leakage is the elicitation criterion, because noticing something said in passing looks like a listening skill and could easily be scored as attentiveness, so its anchors are written entirely around whether a follow-up question was asked and whether the disclosure reached the note. Handling time is measured, because the queue pressure in the brief is real and a report that concealed it would be dishonest, but it is not a criterion, carries no weight, and is annotated on the report as not a positive signal — the shorter call here is usually the one in which the van was never asked about. A deployer who lifts it out and runs it as a hiring metric in its own right is reintroducing the incentive this design tests resistance to. It does not score product or insurance knowledge the candidate has not been trained on, because the wording extract is supplied in the brief, which matters for buyers hiring from outside insurance. Two things deserve monitoring. Candidates with prior insurance servicing experience will do better on the judgment items than on the call, and a gap in that direction reflects format familiarity rather than competency, so the two should be compared rather than summed. And the written half is graded on what is recorded, never on spelling, punctuation or idiom; deployers should watch for a score gap that appears only on the note and confirmation, which would indicate a literacy effect rather than a difference in documentary discipline. Extended time for the written tasks should be available on request without a diagnosis.

Policy servicing is screened as a generic contact-centre seat and it is not one. The difference is that an ordinary-sounding sentence spoken here changes what the insurer owes. Yes, that is covered is a statement about a contract, and if it is wrong the insurer will very often be held to it or will pay a complaint to avoid the argument. The NAIC's Model 900 prohibits knowingly misrepresenting relevant facts or policy provisions relating to the coverages at issue to insureds as well as to claimants, and the servicing desk is where insureds are told about coverage far more often than the claims desk is. Nothing in a CV sift, a phone screen or a data-entry test observes whether a candidate makes the distinction.

The fuse on this seat is unusually long, and that is the reason the assessment has to manufacture the moment rather than wait for it. Almost nothing that goes wrong in servicing reveals itself on the call. It reveals itself at the next claim, when a coverage position turns out to rest on something an advisor said eighteen months earlier, or on a disclosure the customer is certain they made and the file does not record. A screen that only observes whether the call went pleasantly will pass exactly the candidates who create that problem, because the call in which the advisor confirms cover they should not confirm is a call the customer enjoys.

The three forks are all planted inside one address change, which is deliberate. Building a scenario around a dramatic coverage query would be easier to score and would test the wrong thing, because the failure in production happens during routine transactions when nobody's attention is on coverage. The customer mentions the van while explaining the move. They ask about weekend deliveries as a follow-on thought. They ask what the advisor would do because the advisor has been helpful and they trust them. None of it is adversarial and all of it is ordinary, and that is precisely why the reflexive answers are wrong.

The elicitation criterion carries a quarter of the weight because it has the largest downstream effect and the least visibility. Material facts arrive in servicing calls disguised as conversation: the son who has moved in, the new job with the company car, the lodger, the licence endorsement. A good advisor hears the second half of the sentence, asks the follow-up, and records it; an average one processes the transaction that was requested. Both calls sound identical to anybody not attending to the content, which is exactly why a transcript scored against a disclosure planted at a known point is a better instrument than any interview about attention to detail.

The advice fork is the one buyers underestimate. A servicing advisor is generally permitted to give factual information and not to recommend that a particular cover is or is not suitable for this customer, and the jurisdictional detail varies while the behaviour does not. The pressure to cross it comes from the customer, every time, in the same four words. An advisor who answers has stepped into a regulated activity their firm may not be authorised for in that seat, and has done it warmly, helpfully, and in the middle of an unremarkable call. The 3 anchor is written around the bare refusal, because refusing without offering anything is not the right answer either; it leaves the customer with a decision and no information, and it is the behaviour that makes compliance feel like an obstacle to everyone in the operation.

The written half closes the loop, and it is what makes this design worth twenty- eight minutes rather than twelve. The note is the artefact that will be read by an underwriter or a claims handler years later, and the confirmation is the document that determines whether a customer can reasonably say they were told something. Scoring both against the candidate's own transcript catches the two failures that matter: a disclosure that was heard on the call and never reached the file, and a confirmation that leaves a customer believing something was confirmed when it was not.

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, Occupational Outlook Handbook, Financial Clerks, 2025, https://www.bls.gov/ooh/office-and-administrative-support/financial-clerks.htm
NAIC Unfair Claims Settlement Practices Act (Model 900), prohibiting knowingly misrepresenting to claimants and insureds relevant facts or policy provisions relating to coverages at issue, https://content.naic.org/sites/default/files/model-law-900.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 report Or talk to us about this role