IAI CB3 common mistakes
IAI CB3 · Understand the traps and how to avoid them.
Treating the CB3 assessment as a pure multiple-choice knowledge test and never producing written or spreadsheet-backed answers to the case study, so the candidate memorises definitions of actuarial control cycles but cannot apply them to a scenario.
Candidates coming from CB1/CB2 assume CB3 is another recall exam and revise by re-reading notes rather than practising the applied, output-based tasks the module actually requires; the format is unfamiliar and under-advertised.
What to do instead
Obtain the current CB3 syllabus and specimen assessment, then build a practice schedule where at least half of study time is spent producing written analyses and spreadsheet models from short case scenarios under time pressure, not re-reading.
Reading the case study exhibit once and immediately starting to write, without extracting the data, assumptions, and stakeholder constraints into a structured list first.
Time pressure and the desire to fill the page create a rush to prose; candidates underestimate how much of the marks depend on demonstrably using the specific figures and facts in the scenario rather than generic theory.
What to do instead
Spend the first five to ten minutes of any practice task building a one-page skeleton: key data, stated assumptions, stakeholders, and the decision to be supported. Write only from that skeleton, and mark each paragraph against an item in it.
Producing a technically correct model or calculation but presenting it with no narrative linking the numbers to the business decision, so the answer reads as a data dump.
Candidates with strong quantitative backgrounds treat the calculation as the deliverable and assume the reader will draw the conclusion; CB3 rewards communication of results to a specified audience, which is a separate skill rarely practised.
What to do instead
Adopt a fixed output template: purpose, key results in plain language, sensitivity or uncertainty, recommendation, and limitations. Practise rewriting every numerical result as a one-sentence implication for the named decision-maker.
Failing to state and justify assumptions explicitly, leaving the reader unable to see which mortality, lapse, expense, or discount assumptions were used and why.
In shorter exams assumptions are often given; in applied work candidates silently import default values or textbook figures instead of deriving them from the case, and they fear that stating an assumption exposes them to criticism.
What to do instead
Maintain a visible assumptions table in every answer with columns for assumption, value, source or rationale, and sensitivity. Rehearse justifying each one in a single sentence tied to evidence in the case.
Ignoring the specified audience and format, for example writing a technical memo when the task asks for a board briefing or a client-facing summary, or omitting the requested word limit and structure.
Candidates skim the task instructions and default to the writing style they are most comfortable with; the audience specification is often a single line that is easy to miss under exam pressure.
What to do instead
Before writing, underline every instruction verb in the task and convert it into a checklist (audience, format, length, required sections, deliverables). Verify the finished answer against the checklist line by line.
Running out of time because the candidate builds an over-elaborate spreadsheet model with unnecessary tabs, formatting, and formulas before drafting any written commentary.
Spreadsheet work expands to fill available time and gives a false sense of productivity; candidates who enjoy modelling delay the writing, which carries substantial marks, until it is too late.
What to do instead
Practise with a hard time budget split, for example one third modelling and two thirds writing, and set interim checkpoints. Build models with the minimum structure needed for the required outputs and no cosmetic formatting.
Presenting a single point estimate with no sensitivity analysis, scenario testing, or discussion of uncertainty in the recommendation.
Candidates treat the central calculation as the answer and are not used to framing results as ranges; exam-style questions often have one right number, whereas applied advice requires exploring how the answer changes.
What to do instead
For every recommendation, add at least two alternative scenarios or a simple sensitivity table on the two most material assumptions, and state explicitly how the recommendation would change if those assumptions moved.
Writing recommendations that are vague or non-actionable, such as 'monitor experience' or 'consider the risks', without specifying who does what, by when, and against which trigger.
Candidates hedge to avoid being wrong and lack practice in converting analysis into operational actions; generic recommendations feel safe but score poorly against criteria demanding decision-relevant advice.
What to do instead
Rewrite each recommendation in an action-owner-trigger format: the action, the responsible role, the monitoring metric, and the threshold that prompts review. Delete any sentence that could apply unchanged to a different case study.
Ignoring professional and ethical dimensions such as conflicts of interest, regulatory constraints, data limitations, or the need to communicate uncertainty honestly, even when the scenario clearly raises them.
Candidates focus narrowly on the technical calculation and treat professionalism as a separate topic; the case study embeds these issues subtly, and they are easy to overlook when time is short.
What to do instead
Add a standing final section to every practice answer covering professional standards, regulatory and data constraints, and communication of uncertainty, and rehearse identifying these triggers in past or specimen scenarios.
Practising only full-length past papers in silence and never seeking feedback, so recurring weaknesses in structure, assumption justification, and communication persist unrecognised.
CB3 has limited official past material and marked scripts are scarce, so candidates self-mark generously and cannot see where a marker would deduct; studying alone removes the external check.
What to do instead
Exchange marked practice answers with peers or a tutor against the published marking criteria, keep an error log of recurring weaknesses, and redo the same task after feedback to confirm the fix has been absorbed.