What should you ask a social media manager before hiring?

What decision should the interview actually make?

The interview should decide which candidate has earned a small, reversible assignment under the business’s real conditions. It should not decide who tells the most persuasive story, recognizes the most platform trends, or presents the largest follower count without context. Write the business problem first: for example, produce and schedule eight monthly posts from approved service facts while the office manager retains pricing, claims, complaints, and lead follow-up. Then test the candidate against that exact boundary.

Give every finalist the same brief, questions, evidence requests, scoring rules, and hard stops. Separate work the provider will own from work the business will retain. If the owner has not named the current offer, customer decision, source-fact owner, media permissions, approver, customer destination, response owner, and success evidence, pause the hiring process and repair those inputs. An interview cannot transfer business decisions that have never been made.

Which three questions reveal whether the candidate understands the work?

Ask: Which customer decision would you help us make easier this month? Which deliverables and operating duties are included? Which inputs and decisions remain with us? A useful answer translates the brief into a customer path, countable original work, named formats, deadlines, revisions, approval states, publishing responsibilities, monitoring windows, reporting, exclusions, and owner inputs. A weak answer repeats post counts or promises engagement without explaining the operating system around the posts.

Request one annotated scope or campaign map with sensitive client details removed. Look for a clear distinction between original concepts and resized exports, content creation and account management, routine replies and sales or support decisions, organic work and paid media, and a provider report versus the business’s direct platform access. Score the evidence, not the terminology. Social media manager, strategist, creator, agency, and virtual assistant are not standardized scopes.

Which questions test source facts and advertising judgment?

Ask: Where will every price, service detail, date, result, comparison, testimonial, and image permission come from? What do you do when the source is missing, old, or contradictory? Show one example of a post you deliberately paused or narrowed. FTC guidance requires truthful, non-deceptive advertising and an appropriate basis for objective claims before publication. The candidate should be able to connect a public statement to a current business source and explain who is qualified to approve it.

Use a short correction exercise. Provide an attractive draft with one expired offer date, one unsupported superlative, one customer photograph without a permission record, and one working CTA. A strong candidate flags the three unresolved elements, preserves the valid destination, requests the minimum evidence, and keeps the draft out of the queue. Reject an answer that treats disclaimers, owner approval, or a satisfied customer as a substitute for claim support.

How do you test production, approval, and correction handling?

Ask: Walk us through one item from brief to live verification. How are versions, feedback, final approval, scheduling, and publication recorded? What happens when a fact changes after approval? The answer should name states rather than relying on a crowded message thread: source ready, draft, factual review, revision, approved final, scheduled, live checked, corrected, paused, or archived. The complete combination of image, caption, accessibility text, link, date, profile, and disclosure needs a visible final decision.

Request a redacted approval record and run a change drill during the paid pilot. Change a material detail after several formats have been prepared and observe whether the manager finds every derivative, pauses scheduled versions, revises from the source, obtains fresh approval, publishes only the accepted version, and verifies the live destination. Speed matters, but dependable correction is more valuable than a fast calendar that leaves stale claims in several places.

What should you ask about account access and connected tools?

Ask: Which exact platform roles do you need for each accepted duty? Which people, subcontractors, applications, schedulers, asset libraries, analytics systems, and AI tools will touch our material? How will we remove access without losing accounts or records? Meta separates content, messages, ads, insights, settings, and full-control capabilities. LinkedIn separates super admin, content admin, analyst, and paid-media permissions. TikTok Business Center distinguishes administrative and assigned asset access. The requested role should follow the duty, not provider convenience.

Keep primary ownership, recovery, billing, domains, analytics properties, administrator management, and durable source files under business control. Google lets each Business Profile user work through a separate account instead of sharing a password, and its third-party rules preserve client ownership, reporting access, transparency, and a practical termination path. A candidate who asks for a primary password or exclusive ownership before explaining a narrower supported role fails the access gate.

Which questions expose unsafe response and escalation promises?

Ask: Which comments, messages, and reviews may you answer without new approval? What categories are always handed to us, and how quickly? A useful scope lists approved public facts, hours, locations, booking instructions, availability rules, and simple acknowledgments, then separates complaints, refunds, disputes, threats, private information, regulated questions, custom quotes, employment matters, crises, and sales decisions. Always-on engagement is not meaningful when nobody can say what the manager is authorized to decide.

Request the response library, escalation matrix, coverage window, handoff record, and one example of an intentionally unanswered message. Confirm how customer information is stored, which tools or subcontractors can see it, how long it is retained, and what leaves the provider’s system after termination. A fast reply can still be wrong, disclose information, contradict current policy, or strand a qualified inquiry outside the business’s sales and support process.

What should a candidate prove about reporting and business outcomes?

Ask: Which decisions will your report support, where do the numbers come from, and what can the data not prove? Platform reports should preserve definitions and source access. Google requires third parties managing Business Profiles to distinguish Business Profile performance from other platform data and make reports accessible. Ask the candidate to show a redacted report that moves from delivery and customer actions to a specific continue, change, or stop recommendation.

Require a measurement map before the pilot: published items, failures, corrections, owner time, qualified clicks, calls, bookings, quote requests, orders, visits, or other accepted actions. Use consistent campaign parameters where practical, but do not let a tagged link become a promise that every later sale was caused by one post. Reject guaranteed followers, rankings, leads, or revenue when the candidate cannot name assumptions, baselines, attribution limits, and the business-controlled systems that confirm the result.

Which exit questions protect continuity before the relationship begins?

Ask: What exactly do we receive at the end of the pilot or contract? How quickly can we revoke your access, retrieve approved and source files, identify licenses, reconcile scheduled posts, preserve live links and reports, and transfer open customer handoffs? Request a sample offboarding checklist before signing. Exit quality is evidence of operational maturity because it depends on organized ownership, permissions, records, and deliverables throughout the engagement.

The answer should cover account and tool roles, source and final files, reusable components, captions, accessibility text, approvals, licenses, publication records, campaign parameters, analytics exports, response logs, unresolved corrections, customer handoffs, retention and deletion duties, and the last responsible owner. A provider-specific dashboard may be useful, but it should not be the only place the business can understand or recover its public operation.

How does the 100-point interview scorecard work?

Score five areas from zero to 20: scope and retained work; source facts and claim control; production, approval, and correction; access, response, and continuity; measurement and decision support. For each area, award zero to five points for a specific answer, zero to five for a relevant artifact, zero to five for a credible demonstration or example, and zero to five for fit with the business’s written brief. Record what was observed and what remains an assumption.

Interpret 85 to 100 as eligible for a paid pilot when no hard stop is present. A score of 70 to 84 can support a narrower pilot after the named gap is repaired. Below 70, do not grant live duties; the candidate or the business brief needs more work. Do not average away shared-password requests, exclusive account ownership, fabricated evidence, hidden subcontractors, unsupported guarantees, refusal to use approval, or no safe exit. One hard stop can outweigh an otherwise polished total.

What does a worked comparison look like for a local service business?

Consider a hypothetical plumbing company interviewing three providers for eight monthly posts, scheduled publishing, routine public replies, and a monthly report. Candidate A shows attractive visuals but cannot distinguish original concepts from adaptations, requests the owner’s Facebook password, and promises a fixed number of leads. The portfolio earns production points, but the password request and unsupported guarantee are hard stops. Candidate A does not advance.

Candidate B maps the customer decision, names source inputs, shows an approval record, requests task-level content and insight access, routes price and emergency questions to dispatch, and provides an export and removal checklist. The report example is clear, but the candidate has not defined weekend coverage, producing 88 points with one repair item. Candidate C scores 77 because its production process is sound but its report ends at reach and likes. Candidate B receives a paid pilot limited to weekday content, publication, and approved routine replies.

How should the paid pilot confirm the interview evidence?

Use two cycles with one audience, one current service or customer question, approved facts, usable media, one or two profiles, a named approver, a working destination, and only the permissions required for accepted duties. Cycle one tests ordinary delivery. Cycle two introduces one real material change, such as availability, price, image permission, or destination. Do not add inbox, advertising, finance, or administrator access merely to make the pilot feel complete.

At the review date, rescore only observed behavior. Choose continue, expand one duty, repair one control, narrow the scope, or stop and offboard. Compare owner time removed alongside delivery, corrections, live-link accuracy, qualified customer actions, report usefulness, file organization, and access removal. The interview earns the test; the pilot earns the recurring relationship.

What should the final hiring decision record contain?

Keep the written brief, all candidate scores, evidence links, hard-stop decisions, selected duties, exclusions, owner inputs, platform roles, tool and subcontractor list, approval states, response matrix, report definitions, attribution limits, deliverables, rights, retention, removal steps, pilot dates, and acceptance criteria. Attach the contract to this operating record instead of assuming the contract alone will make vague duties executable.

The next decision should be explicit: this candidate may run this bounded pilot because these observable controls passed, these gaps remain outside scope, and this business owner retains these decisions. If no candidate clears the gate, refine the brief or search again. Choosing nobody is better than granting broad access to a persuasive provider whose real workflow cannot be inspected.