Create an Onboarding Scavenger Hunt from Company Documentation with ChatGPT

New employees often receive a large collection of documents before they know which ones matter to their work. Reading everything in sequence can leave them with little practice finding the right answer when a real question arises. An onboarding scavenger hunt gives them a small set of purposeful tasks that develop that skill.

The output in this guide is an activity pack: learner tasks, source references, an answer key, and facilitator notes. It is an original workflow using ChatGPT, with fictional examples rather than a tested company program. Current capability documentation was checked on October 6, 2026. A company owner should review the pack against approved policies and the learner's access.

Choose useful learning outcomes

Start with the actions a new employee needs to perform during the first stage of onboarding. Examples include finding the expense submission instructions, locating a project brief template, and identifying where to ask an access question. Choose outcomes relevant to the role rather than trivia about the company's history.

Define what successful completion demonstrates. Finding a document proves navigation; explaining which section applies demonstrates interpretation. Submitting a real expense or changing an account setting is an operational action and should be handled through the company's approved process, not hidden inside a learning game.

Decide how the activity fits into the onboarding schedule. Specify the expected time, whether it is self-paced, and how learners can ask for help. Do not assume all learners have identical access, language fluency, or familiarity with the company's tools.

Assemble the approved documentation set

Select current, role-appropriate documents and record an owner, revision date, and source ID for each. Avoid obsolete policy copies and links available only to administrators. An unanswered task can reflect a permissions problem rather than a learner's failure.

Provide ChatGPT with the selected text or readable files. The current ChatGPT usage guide documents file attachment for extracting and reorganizing information. Verify that the supplied sections are readable, including any tables that describe responsibilities or process steps.

Keep sensitive or unrelated information out of the activity pack. Use approved examples instead of real personnel records or confidential incidents. The learner should practice finding useful guidance without needing access to information outside the role's normal scope.

Design tasks that require purposeful discovery

Each task should contain a situation, a question, and a clear completion criterion. “Find the handbook” is too broad. “A fictional colleague needs to request a project tool; identify the documented request route and its owner” creates a meaningful reason to explore.

Vary the activity types. Include locating a source, choosing between two documented procedures, identifying a prerequisite, and explaining when to ask for help. Keep every correct answer grounded in the supplied documentation. Avoid trick questions and tasks that rely on undocumented organizational customs.

Give each task a stable ID. That allows facilitator feedback to refer to the exact task even when its wording changes in a later version.

Ask ChatGPT for an activity pack

Supply the learning outcomes, source register, learner role, and format. Request the learner-facing pages separately from the answer key. Otherwise the model may put the correct answer immediately below the question and remove the discovery element.

Create an onboarding scavenger-hunt activity pack from the supplied documents.
Use only approved sources and the stated learner role.
For each task include an ID, situation, question, completion rule, and help route.
Do not require real submissions, account changes, or unauthorized access.
Produce learner tasks separately from the facilitator answer key.
In the answer key cite the source ID and section supporting each answer.
Flag missing documentation, contradictory rules, and access assumptions.
Keep estimated timings provisional until a facilitator reviews them.

Review whether the tasks cover the selected outcomes rather than simply following the handbook's headings. A useful pack might leave several documents unused because they do not help this particular learner at this stage.

Review a fictional task and answer

Suppose source D-05 contains an approved project brief template and states that a named team reviews completed briefs. A task can ask the learner to locate the template and identify the documented review role. Its answer key should refer to D-05's relevant section and preserve any submission prerequisite.

Do not ask ChatGPT to invent the actual reviewer when the source only names a role. “Project operations reviewer” may be supported; assigning a specific colleague without evidence is not. If the documentation lacks the route, the facilitator should clarify it before the activity is released.

For a task comparing procedures, explain why the fictional situation falls under one procedure. The answer key should show the deciding condition rather than merely state a label. This helps facilitators distinguish a correct guess from an understanding of the guidance.

Make the experience accessible and supportive

Use plain language and avoid relying on color, obscure icons, or spatial clues alone. Provide descriptive links and a text-based version of any visual task. The activity should teach document navigation rather than reward experience with a particular interface.

Allow learners to report that a source is unavailable or unclear. Include a facilitator response for those cases so people are not pressured to work around missing permissions. A blocked task can be marked for follow-up without counting it as an incorrect policy interpretation.

Review competitive elements carefully. A timed leaderboard may distract from learning and disadvantage people using assistive tools or working in a second language. If timing is useful, use it to improve the pack's workload rather than make unsupported judgments about employee performance.

Pilot the pack before rollout

Ask someone unfamiliar with the selected documents to attempt the proposed tasks in a review session. Record which links they could open, which instructions confused them, and which answers required help. This guide has not run that pilot; it describes what your facilitator should observe.

Check the answer key with document owners. Resolve contradictions and verify that the pack accurately reflects current procedures. Use a feedback-to-edit ledger if reviewers propose competing changes to a task.

Adjust the number and sequence of tasks based on actual pilot feedback. Place simple navigation tasks before interpretation tasks that depend on the same documents. Retain an optional extension for learners who finish early instead of making the essential pack unnecessarily long.

Maintain the source and task register

Record the pack's version, supported role, source revisions, approval date, and owner. When a policy or template changes, identify every task and answer that refers to it. Review those items before reusing the pack with the next onboarding group.

Collect learner questions as documentation feedback. Repeated confusion about the same step may indicate unclear company guidance, not a need for a harder activity. Route those findings to the document owner with the task ID and relevant source section.

Keep completion records limited to what the onboarding process requires. The activity pack should support learning and reliable navigation while making its own assumptions and maintenance needs visible.

Frequently asked questions