How to Explain What Candidates May and May Not Use During an Assessment
Hiring teams should first define the capability an assessment is meant to reveal, then set resource rules that preserve the evidence needed to evaluate that capability. For every tool or source—from notes and calculators to generative AI—state whether it is allowed, allowed with limits or disclosure, or not allowed. Give candidates these rules before they begin, include concrete examples, and provide a clear way to ask questions.
The goal is not to create the longest possible rulebook. It is to replace assumptions with explicit criteria, relevant instructions, and practical checkpoints that candidates and reviewers can follow.
Start With the Capability and Evidence the Assessment Should Reveal
Before deciding what candidates may use during an assessment, answer two questions:
- What capability are we trying to evaluate?
- What evidence would help us evaluate it?
An assessment designed to test independent recall may reasonably restrict reference materials. A task designed to resemble everyday work may allow documentation, internet research, calculators, code libraries, or writing tools. Neither approach is automatically better. The appropriate choice depends on the role and the purpose of the exercise.
For example, consider two assessments that both involve spreadsheet work. One may test whether a candidate understands basic formulas without assistance. The other may test whether the candidate can investigate messy business data, select suitable tools, document assumptions, and explain a recommendation. Allowing internet research would undermine some of the evidence sought in the first exercise but could make the second exercise more representative of the job.
Write a one-sentence purpose statement before creating the resource rules. For example:
> This exercise evaluates your ability to identify the main operational problem, analyze the supplied information, make reasonable assumptions, and explain a practical recommendation.
That statement becomes the decision filter. If a resource replaces the capability being evaluated, consider limiting it. If it supports the candidate in demonstrating that capability as they would at work, consider allowing it. If its effect depends on how it is used, allow it with defined limits or a disclosure requirement.
Classify Each Tool or Resource Using Three Clear Categories
A three-category system is easier to interpret than vague instructions such as “work independently,” “use common sense,” or “do not cheat.” Those phrases do not tell a candidate whether checking documentation, using spellcheck, or consulting an old project is acceptable.
Use these categories:
- Allowed: The candidate may use the resource without special approval, subject to the assessment’s normal time and submission rules.
- Allowed with limits or disclosure: The candidate may use the resource only within stated boundaries or must explain how it contributed to the submission.
- Not allowed: Using the resource would prevent the assessment from showing the capability the team intends to evaluate.
The following table is an illustrative decision aid, not a universal policy:
| Resource | Questions for the hiring team | Possible classification |
|---|---|---|
| Personal notes | Does the task test recall, or the ability to apply knowledge? | Allowed for an application task; restricted for a recall task |
| Reference materials | Would employees normally consult manuals, policies, or documentation? | Allowed, possibly limited to named sources |
| Internet search | Is research part of the capability, and should candidates cite sources? | Allowed with source links or prohibited for a closed-resource exercise |
| Calculators | Is manual calculation important, or is interpreting the result more important? | Allowed unless arithmetic itself is being assessed |
| Code libraries or development tools | Is selecting and applying existing tools part of the role? | Allowed with limits, version details, or attribution |
| Generative AI | Would it support the work or produce the very output the candidate is expected to create independently? | Allowed, allowed with disclosure, or prohibited depending on purpose |
| Spellcheck or grammar tools | Is writing mechanics central to the assessment? | Often allowed; limits may apply to language-focused tasks |
| Communication with other people | Must the submission represent the candidate’s independent work? | Usually prohibited unless collaboration is intentionally assessed |
| Reuse of prior work | Can candidates bring their own templates or code, and how should they identify reused elements? | Allowed with disclosure or prohibited when original work is required |
Avoid treating the resource name alone as the decision. “Generative AI allowed” is incomplete if candidates do not know whether they may use it to brainstorm, edit, research, write, debug, or produce a final answer. Likewise, “internet access allowed” does not explain whether copying an existing solution is acceptable. Define permitted uses and boundaries in terms of behavior.
Compare Realistic Tool Use With Consistent and Independent Evaluation
Resource decisions usually involve several competing considerations:
- Workplace realism: Allowing tools used in the role can make the task more representative of everyday work.
- Independent-skill evidence: Restricting a tool can make it easier to observe what a candidate can do without that support.
- Consistency: Giving candidates the same clear rules supports more consistent administration, although identical instructions do not make every candidate’s circumstances or results identical.
- Reviewability: Disclosures, citations, working notes, or a short explanation can help reviewers understand how the result was produced.
The hiring team should decide which consideration matters most rather than trying to maximize all four at once. A developer exercise might allow documentation and standard libraries because choosing and applying them is part of the job. The same exercise might require candidates to identify copied code and explain any generative AI assistance so reviewers can discuss the candidate’s reasoning.
For a short writing task, grammar tools may be acceptable when the team cares primarily about message structure and judgment. They may be restricted when unaided language mechanics are the capability being evaluated. The rule should follow the intended evidence, not an assumption that a particular tool is inherently good or bad.
Disclosure is a useful middle option, but it should be specific. Instead of asking candidates to “declare AI use,” say what to report: the tool category, the part of the task where it was used, and any output incorporated into the submission. A disclosure offers context; it should not be treated as proof of every action taken.
Write the Rules Candidates Need Before the Assessment Begins
Send the complete instructions before the assessment clock starts. Candidates should not have to infer important restrictions from a later email or discover them after submitting their work.
Effective instructions address:
- The capability or outcome being evaluated
- The resources that are allowed
- Any limits on how those resources may be used
- Resources or forms of assistance that are not allowed
- What candidates must cite or disclose
- The time limit and deadline
- The required file type, link, or submission format
- Whether working notes, prompts, sources, or calculations must be included
- Where candidates can ask questions
- How broadly a clarification will be shared when it affects other candidates
Use examples to resolve predictable ambiguity. If generative AI may be used for brainstorming but not drafting the final response, say so. If candidates may search official documentation but not reuse a published solution to the same problem, state that distinction directly.
Questions should go to a named person or monitored channel. Record material clarifications and provide the same answer to candidates working under the same rules. Where an employer and candidate connect through MeeBoss, Chat to Apply can support a direct conversation with the hiring team about unclear instructions. MeeBoss supports the conversation; assessment rules and hiring decisions remain with the employer and its people.
Apply the Framework to a Realistic Employer Scenario
Consider a hypothetical software company hiring a customer operations manager. It asks candidates to review a fictional support backlog and propose a plan for the next two weeks.
- Define the capability. The team wants to evaluate prioritization, operational judgment, assumptions, and communication—not unaided recall of support-management terminology.
- Identify the evidence. Reviewers need a prioritized plan, a brief rationale, stated assumptions, and a simple explanation of how the candidate would monitor progress.
- Classify resources. Notes, calculators, and general internet research are allowed. Generative AI is allowed for brainstorming or editing if its use is disclosed. Asking another person to create or revise the recommendation is not allowed. Prior templates may be used only if the candidate identifies them and adapts the analysis to the supplied scenario.
- Review the tradeoffs. Allowing research reflects how the manager might work on the job. Requiring assumptions and disclosures gives reviewers more context about the candidate’s process. Restricting outside collaboration keeps the submission focused on the candidate’s own judgment.
- Write the instructions. The company provides the resource categories, a two-hour recommended effort, the deadline, the requested document format, and the disclosure method.
- Answer a candidate question. A candidate asks whether they may use a spreadsheet template to calculate workload. The team answers yes, provided the candidate submits the calculations and identifies the template. Because the answer could affect everyone, the team shares the clarification with all active candidates.
- Complete a consistency check. Before reviewing submissions, interviewers agree that disclosed brainstorming assistance is permitted and should not be treated as a rule violation. They evaluate the recommendation using the same role-related criteria established at the start.
This approach gives the employer a traceable path from assessment purpose to written rules. It does not depend on reviewers making different assumptions after seeing a candidate’s work.
Give Candidates a Simple Instruction Template
The following template is an editable starting point. Replace every bracketed item and remove examples that do not fit the assessment.
> Purpose > This assessment evaluates your ability to [name the capability and expected evidence]. > > Allowed > - [Personal notes] > - [Named reference materials] > - [Calculator, spreadsheet, or specified development tools] > > Allowed with limits or disclosure > - You may use [internet research] if you provide links to the sources that influenced your answer. > - You may use [generative AI or writing tools] for [permitted uses]. Do not use them for [restricted uses]. Disclose [tool category, purpose, and affected part of the submission]. > - You may reuse [templates, code, or prior work] only if you identify what was reused and what you changed. > > Not allowed > - Do not ask another person to produce, rewrite, or materially improve your submission. > - Do not use [specific prohibited resources or behaviors]. > > Timing and submission > - Recommended or maximum working time: [time] > - Deadline: [date, time, and time zone] > - Submission format: [file type, link, or platform] > - Include: [sources, working notes, disclosures, prompts, or calculations] > > Questions > If any instruction is unclear, contact [person or channel] before [cutoff]. Asking for clarification will not be treated as an attempt to gain an unfair advantage.
Neutral wording matters. Describe the conduct the employer expects without making assumptions about a candidate’s intent or character.
Review the Rules for Clarity and Consistent Application
Before sending the assessment, run a short review with the recruiter, hiring manager, and at least one intended reviewer:
- Does every rule connect to the role and the stated assessment purpose?
- Have notes, reference materials, internet search, calculators, development tools, generative AI, writing aids, outside help, and prior work been addressed where relevant?
- Could a candidate understand the difference between permitted research and prohibited copying?
- Are limits and disclosure requirements concrete enough to follow?
- Do the deadline, time expectations, submission format, and question channel appear in one place?
- Would two reviewers interpret a disclosed use of a tool in the same way?
- Is there an owner for answering questions and sharing material clarifications?
After each hiring round, teams can review practical signals such as the number of clarification questions, repeated points of confusion, incomplete disclosures, submission-format errors, and disagreements among reviewers about whether a resource was permitted. These signals do not prove that an assessment is effective, but they can identify instructions that need revision.
Accommodation-related requests should be directed through the employer’s applicable process and handled separately from the general resource rules. That keeps the assessment instructions clear while ensuring individual requests reach the appropriate person or team.
A reliable policy is one candidates can understand before they start and reviewers can apply without improvising afterward. Define the goal, classify each resource, explain the tradeoffs, publish concrete instructions, and check that answers remain consistent.
Join MeeBoss and start a real hiring conversation.