Assess a creator campaign subcontractor with a paid, time-limited exercise built from wholly fictional records. Give every candidate the same brief, score their evidence and calculations, and inspect whether another person can continue their work. Keep client files, account permissions and creator contact out of the test. A passing result should lead to a scoped engagement review, not automatic access to live campaigns.
This is a recommended evaluation method. The payment terms, time limits and scoring thresholds below are illustrative agency choices, not platform rules or legal requirements.
Test the job you plan to delegate
Name one deliverable the specialist would own. A shortlist researcher should evaluate candidates. A reporting specialist should reconcile a fictional report. An outreach specialist can draft replies to simulated messages without sending them.
Modash's discussion of agencies and in-house teams describes dividing work around specific gaps in expertise or capacity. Apply that boundary to the test. Do not ask a research specialist to design an entire campaign merely because the exercise is available.
First settle the inputs and decision owner. The client intake brief guide can help you define those before translating the work into a fictional assignment.
Ask candidates to identify who will perform the test and who would perform the contract. If the subcontractor is a firm, evaluate the proposed delivery team. A senior sales contact's work tells you little about another person's handoff.
Build a packet without client information
Create fictional creator IDs, post counts, audience figures, quotes and availability notes from scratch. Do not copy an export and replace its name column. Also exclude real handles, contact details, unpublished creative, client budgets and account identifiers.
The US Federal Trade Commission's security guidance describes using fictitious information to avoid exposing personal information during training or development. Applying that approach to subcontractor assessment is a practical precaution. The same guide advises limiting contractor access to what the task requires.
Inspect the files you send, including hidden spreadsheet tabs, comments, document history and embedded links. Start with a new file if cleaning an old client workbook would leave uncertainty. Call the packet synthetic throughout; nobody should mistake its figures for real creator performance.
The exercise should contain a few realistic ambiguities. Tell candidates that some evidence is incomplete. Score whether they record uncertainty and ask for clarification. Do not reward guessing what the evaluator wanted.
A paid test brief you can adapt
The following is an illustrative shortlist-research exercise. Its data and time allowances are hypothetical.
| Term | Assignment |
|---|---|
| Goal | Recommend two fictional creators for a UK homeware tutorial campaign |
| Work limit | 90 minutes of preparation and a 20-minute handoff discussion |
| Payment | Agree a fixed fee and payment date in writing before starting; pay for the agreed work regardless of selection |
| Materials | Use only the six synthetic records below; no external research or account login |
| Criteria | At least 50% UK audience, availability in the campaign week, and a combined quoted fee no greater than GBP 1,000 |
| Output | One selection table, calculations, unresolved questions and a short handoff note |
| Communication | Send clarification questions to the evaluator; stop at the work limit and list unfinished items |
| Restrictions | No creator contact, publishing, client access, further delegation or uploads to unapproved services |
Agree confidentiality expectations for the test itself: who may see the packet, whether outputs may enter a portfolio, and when copies must be deleted. Agree a delivery date that allows the candidate to schedule the work. Tell them which local tools and AI assistance, if any, are permitted. Apply the same rules to everyone. If you add another exercise or revision, agree its fee and scope first. Decide in writing how the evaluation output may be used; do not quietly turn it into client work.
For this test, every quoted fee covers one tutorial video. The audience and availability fields are fictional supplied information, not independently verified facts.
| Record | Tutorial fit | UK audience | Campaign week | Quote |
|---|---|---|---|---|
| C01 | Clear demonstration format | 62% | Available | GBP 450 |
| C02 | Clear demonstration format | 54% | Available | GBP 500 |
| C03 | Clear demonstration format | Missing | Available | GBP 400 |
| C04 | Clear demonstration format | 71% | Unavailable | GBP 350 |
| C05 | Clear demonstration format | 48% | Available | GBP 300 |
| C06 | Clear demonstration format | 59% | Available | GBP 700 |
A defensible answer selects C01 and C02 for GBP 950, leaving GBP 50 against the fictional fee ceiling. C03 needs audience evidence. C04 fails the availability condition. C05 falls below the audience threshold. C06 meets the individual conditions, but cannot pair with another fully qualifying candidate within the budget.
That answer does not establish total campaign cost, creator interest, rights clearance or future results. The packet has no evidence for those claims. Ask candidates to name these limits rather than invent allowances or approvals.
For the handoff discussion, change one fact: C02's quote becomes GBP 600. The combined fee becomes GBP 1,050. The candidate should flag the breach and identify the next decision, such as checking C03's missing audience evidence or requesting a budget change. They should not silently lower the quote or waive the audience rule.
Score the files before the presentation
Use a rubric agreed before submissions arrive. Have two reviewers score independently when practical, then resolve differences against specific rows or statements. Keep both the score and the evidence behind it.
Here is a proposed 100-point rubric. It is an internal decision aid, not a validated prediction of future performance.
| Area | Points | Evidence for full marks |
|---|---|---|
| Evidence handling | 30 | Every decision cites a record ID; supplied facts, assumptions and missing information remain distinct |
| Accuracy | 30 | Eligibility checks and fee totals are correct; the changed quote triggers a revised decision |
| Handoff | 25 | Another reviewer can locate the selection, calculations, open questions and next owner without a call |
| Boundary handling | 15 | Candidate follows tool and contact limits, records unfinished work and explains the proposed access needs |
For each area, award all points when the evidence meets the description, half for a material but repairable gap, and zero when absent or unusable. Record the reason. A formatting preference should not count as an accuracy error.
A sample decision rule is at least 80 points, full accuracy marks and no boundary breach before considering a limited engagement. Fabricated evidence, unauthorized sharing or contacting a real creator should stop selection even if the total score is high. Investigate what happened before deciding whether further work is appropriate.
Discuss one weakness with the candidate. If you cannot distinguish a brief defect from a skill gap, improve the brief and offer a paid retest on equal terms. Avoid unpaid rounds of revisions disguised as clarification.
Separate selection from access approval
A synthetic test shows how someone handles that task under those conditions. It cannot establish their real creator relationships, workload capacity or security controls. Check references with permission, confirm availability and review the proposed working arrangements separately.
Before live work, document confidentiality, permitted recipients and tools, further subcontracting, incident reporting, retention and deletion. Confirm whether your client agreement permits the proposed delegation. Have qualified advisers review the contract and applicable privacy obligations for the relevant jurisdictions. The FTC guide supports written security expectations and oversight; it does not turn this checklist into a complete contract or legal advice.
Ask the specialist to propose the minimum access for the first deliverable. Review the actual platform permissions. For example, YouTube channel permissions allow delegation without sharing the owner's Google Account. Viewer includes revenue data, while Viewer (Limited) excludes it. Channel permissions also do not support YouTube APIs. These distinctions matter when someone requests access for reporting.
No platform access is necessary for the synthetic exercise above. For an approved live engagement, name the person who grants access, checks its scope and removes it. Use the client data separation guide to plan that boundary, and record later additions through a scope-change process.
Choose the next specialist task you need to delegate. Prepare one fictional packet, agree its fee and restrictions, and write the scoring rule before inviting candidates.
