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

Sales and revenue · Mid level

How to assess a Insurance Sales Agent

Everything that goes wrong in this role becomes visible after the sale — at first claim, or when the policy lapses months later — so a process built entirely around the point of sale cannot see it. Screening tests whether the candidate can move someone to yes today; the job requires them to say the sentence that makes yes less likely. Nobody in the current process ever watches a candidate deliver an exclusion, or recommend the cheaper policy when the fact-find points there. The state licence compounds this by being mistaken for a competence gate: it examines statute recall, so it establishes that a person may sell and never that they will sell suitably.

An insurance sales agent spends the day in a sequence of consumer conversations that each have to do two contradictory things at once: move a person who is mildly reluctant towards a decision, and make sure that person understands what they are buying well enough that it still looks like a good decision when something goes wrong. Most of the hours are unglamorous — following up quotes that went quiet, re-running a fact-find because a circumstance changed, explaining for the fourth time why the cheaper policy has a higher excess. The compensation is usually commission-weighted, which BLS notes is the most common arrangement especially for experienced agents, and that structure sits directly on top of the tension in the work.

The gap between the median agent and the top quartile is almost never charm and almost always the fact-find. A median agent asks the questions on the form, hears the answers as data to be entered, and then recommends from a shortlist of two products they are comfortable explaining. A strong agent hears the answer that does not fit — the second property, the recently changed job, the adult child still on the policy — and follows it, because that is where suitability actually lives. The second separator is what happens at the exclusion. Every policy has a clause that will disappoint someone, and the agent who states it plainly before the signature loses a fraction of sales and keeps almost all of them; the agent who lets it slide closes more this month and generates a complaint, a lapse, or a denied claim later, when the cost lands on someone else's desk.

What a hiring manager is really trying to predict, then, is persistency and complaint rate, not first-month production. An agency principal can see production immediately and sees the other two a year later, which is why the role is so often hired on the wrong signal: enthusiasm, network and a strong audition. Because the seat is licensed, there is also a permanent temptation to let the licence do the assessing. It cannot. Licensure is an employer-verified legal gate that establishes eligibility, and it should be stated on any job advert as exactly that; ranking selling ability is a separate exercise, and a simulation should never claim to verify a credential. The assessable question is narrower and more useful: given a consumer whose stated request does not match their circumstances, does this candidate change the recommendation, and can they explain the limit of the cover in words the consumer repeats back correctly?

What the job actually needs

• needs analysis before recommendation
• plain-language explanation of exclusions and limits
• suitability judgment across a product set
• disclosure discipline under commission pressure
• persistence across a multi-touch consumer cycle

How people fail in this seat

• recommends the product they know rather than the one that fits
• glosses the exclusion that will later deny the claim
• writes cover the customer cannot sustain and the policy lapses
• defers the awkward disclosure until after the signature
• treats a fact-find as a form to complete rather than a diagnosis

What most employers do instead

CV plus a state licence check, a sales-aptitude or personality questionnaire, a manager-run role-play, and in captive channels a recruiting conversation that quietly selects for the size of the candidate's personal network.

The assessment

About 40 minutes end to end.

The systems it runs in

An agency management system holding the client file, with a quoting surface beside it. Applied Systems describes Applied Epic as an agency management system giving real-time access to policy information across the policy lifecycle with customer interactions synced back to it, and that is the shape the fixture uses: the consumer in t1 already exists as a client record with a household, existing policies and an activity history — which is where the forgotten former-employer cover is discoverable by a candidate who looks and invisible to one who does not. The fact-find is completed as the fields on that record rather than as free text, so an unanswered question stays visibly unanswered. Quotes for the two suitable products are pulled through a rater or carrier quoting screen, with the commission difference stated on the comparison rather than in the brief. And t2 is the decisive change: the record of the recommendation is written into the client file as the documentation that sits behind the sale, in the same place a compliance reviewer or a carrier audit would go looking for it, rather than as a free-standing essay about a recommendation. The task's existing basis in the NAIC's revised Model #275 is unchanged by this — naming the surface only puts the documentation where the obligation expects to find it.

• Applied Epic
• Vertafore AMS360
• EZLynx
• HawkSoft
• Salesforce Financial Services Cloud

Any agency or brokerage management system with a client file, policy records, an activity log and somewhere durable to attach the record of advice, plus whatever quoting surface the buyer actually uses — a comparative rater in personal lines, an illustration or carrier portal in life and annuity. Captive agents working a single carrier's own platform get the same exercise with one quote path instead of a comparison, and the conflict in t1 is then expressed as a difference between products rather than between carriers. Rebuilt against the buyer's own product set, their own fact-find fields and their own suitability form where they can provide a sandbox.

What the candidate actually does

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.

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 a regulated seat. Half the weight sits on the two criteria that measure whether the candidate made the sale harder for themselves — following the inconvenient answer, and disclosing the limit before the yes — because everything that goes wrong in this role goes wrong after the sale and only these two behaviours are observable before it. And the consumer's paraphrase of the exclusion is reported verbatim next to the candidate's disclosure, because a disclosure that was made and not understood is the specific pattern that produces a denied claim, and the pairing shows it where a score cannot.

Integrity

• none of the four hidden circumstances appears in the brief, and each requires a distinct question to reach
• the yes at twelve minutes is scripted regardless of what the candidate has established, so the disclosure moment cannot be avoided by pacing
• per-candidate variation of the product pair, the hidden circumstances and the exclusion
• keystroke and paste timing on the written record
• a follow-up question asking which of the consumer's statements the recommendation rests on

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 each hidden circumstance marked at the turn it became discoverable
• the disclosure passage and the consumer's paraphrase, reported together
• the written record as filed
• the four judgments with the stated disclosures
• per-criterion score with the excerpt that earned it

Who decides

Required, and constrained in a way the other designs in this corpus are not. This assessment does not verify a licence and must never be presented as contributing to that verification. Licensure is a legal prerequisite that the employer verifies through the state department of insurance, separately, before any offer — BLS records that agents must be licensed in the states where they sell, with separate licences for life and health and for property and casualty. A reviewer with compliance responsibility at the deploying firm reads the disclosure passage and the written record for every shortlisted candidate, and is the person who decides whether the disclosure standard reflected in these anchors matches the firm's own product lines, jurisdictions and supervision procedures. Two specific corrections. A candidate may decline to recommend anything on this call and ask for a second conversation once they have the existing policy documents, which is a strong answer that several anchors mark as incomplete. And firms differ on when remuneration must be disclosed; where a firm's own rule is stricter or looser than the anchor, the anchor is changed before the assessment runs rather than overridden after it. All overrides carry a written reason. No candidate is rejected on the composite alone.

What this does not measure

This design does not measure, and must not be read as measuring, whether a candidate is licensed, licensable, or compliant with any particular jurisdiction's rules. It ranks the observable components of suitable selling and nothing else. The regulatory frame it borrows — care, conflict-of-interest disclosure, and documentation — comes from the NAIC's revised Model #275, which binds annuity recommendations specifically and has been adopted by most but not all states. It is used here because it is a published, checkable articulation of what a suitable recommendation looks like, not because it governs every line of business. A firm deploying this for property and casualty, or outside the United States, must map the anchors to the standard that actually applies to them before use. No criterion scores accent, dialect, fluency, pace, warmth or persuasiveness. In a consumer-facing sales seat this exclusion carries real weight, because agents are routinely assessed on how they sound to a customer and that is a direct route for accent and dialect to be scored under the name of rapport. The anchors here point at questions asked, recommendations changed, disclosures made and their timing, and what a consumer was able to repeat back. Note that the paraphrase criterion scores the candidate's disclosure and their correction of a wrong paraphrase — never the consumer's fluency, and never the candidate's own accent as a supposed cause. The written record is judged on completeness and reconstructability, not on register, and non-native phrasing is not a deduction. No product or carrier knowledge is required; the brief supplies both products. What forty minutes cannot see is the outcome. The role page's argument is that everything that goes wrong in this seat becomes visible after the sale — at first claim, or when a policy lapses months later — and no session reaches that. Persistency, complaint rate and lapse behaviour are the metrics that matter and they take a year to appear. What is observed here is a single consumer conversation, its record, and four judgments. There is no evidence about volume, about prospecting, about whether the candidate's disclosure discipline survives a bad month against a commission target, or about their behaviour with vulnerable consumers, which requires a specialised design this one does not attempt. Deployers should back-test scores against thirteen-month persistency and complaint rate rather than against first-quarter production, which is the wrong signal and the one the status quo already over-weights. The live-call format may disadvantage candidates with speech, hearing or anxiety-related disabilities. A text-based run of the identical consumer conversation, with the same hidden circumstances, the same scripted yes and the same paraphrase mechanic, must be available on request, scored on the same anchors and unmarked on the result. Extra time on the written record and the judgment task must be available without disclosure.

The first thing to say about assessing a regulated seat is what the assessment is not. It is not a credential check. Licensure is a legal gate the employer verifies with the state department of insurance, and BLS is explicit that separate licences apply to life and health and to property and casualty. Savvanta cannot verify a credential and this design does not claim to. Any job advert using it should state licensure as an employer-verified requirement in its own right, and any vendor implying that a simulation contributes to that verification is misrepresenting what a simulation does. What remains after that gate is the question worth ranking: given a consumer whose stated request does not match their circumstances, does this candidate change the recommendation, and can they explain the limit of the cover in words the consumer repeats back correctly.

Both halves of that question are staged in a single call, and the design's most useful mechanic is the scripted yes. At around twelve minutes the consumer agrees to proceed, regardless of what has or has not been established. This is not realism for its own sake; it removes the candidate's ability to manage the timing. In a real sale the agent chooses when the disclosure happens, and the entire distinction the role page draws is about that choice — the agent who states the disappointing clause before the signature loses a fraction of sales and keeps almost all of them, and the agent who lets it slide closes more this month and generates a complaint, a lapse or a denied claim later, on someone else's desk. Forcing the yes to arrive on a fixed clock means every candidate faces the same moment at the same point, and what they do with it is comparable across a cohort.

The paraphrase is the single strongest observable in this design and it is worth explaining why it works. A disclosure can be made and not land, which is functionally identical to not making it and is the specific pattern that produces a denied claim two years later. Scripting the consumer to repeat the exclusion back in their own words converts an unobservable — did the consumer understand — into a transcript. If the paraphrase is wrong and the candidate lets it stand, that is visible. If the candidate hears the error and corrects it without retreating into policy language, that is visible too, and it is the behaviour the role page describes as the difference between an agent who sells suitably and one who sells.

The fact-find carries the other quarter of the weight, and the four hidden circumstances are chosen so that each requires a different kind of question to reach. The changed income is found by asking about work. The existing employer scheme is found by asking what cover they already hold, which many agents skip because the consumer has already said they have none. The dependent adult child is found by asking who else depends on this. And the affordability problem is only visible by connecting the premium to the income that changed, which is a composition question rather than a fact question. The role page's claim is that the gap between the median agent and the top quartile is almost never charm and almost always the fact-find, and this is the version of that claim that can be marked.

The written record exists because a recommendation that cannot be reconstructed is a recommendation that cannot be supervised. Its structure follows the obligations in the NAIC's revised model regulation — care, conflict disclosure, documentation — which is used here as a published articulation of what a suitable recommendation looks like rather than as a claim about every product line. What it produces for a hiring manager is a candidate-written page showing what they considered and rejected, which is the part of the sale that never survives into a CRM and never appears in an interview.

Finally, the conflict is put on the table explicitly: the brief says one product pays the agent less and fits better. Most assessments of commercial roles leave that tension implicit, which lets a candidate be suitable without cost. Naming it means the recommendation is a decision with a price attached, and a decision with a price attached is the only kind worth observing.

Sources

Every figure on this page is traceable. Where a claim could not be sourced it is stated qualitatively instead.

1. US Bureau of Labor Statistics, Occupational Outlook Handbook, Insurance Sales Agents, 2025 (572,600 jobs; 43,100 annual openings; 3 percent growth; 18,800 additional jobs; median pay $62,280; state licensure required, separately for life/health and property/casualty), https://www.bls.gov/ooh/sales/insurance-sales-agents.htm