How to tell if your social media manager is doing a good job

How should you choose a fair evaluation window before looking at results?

Evaluate at least one complete production and reporting cycle, and use a longer comparison window when the business has low lead volume, a long buying cycle, strong seasonality, or an offer that changes during the period. Record the starting conditions: active offer, audience, locations, channels, posting cadence, website or booking path, typical lead volume, approval speed, and any operational constraint. A manager cannot be evaluated fairly against a baseline that was never recorded.

Separate controllable service duties from uncertain market outcomes. The manager can control whether researched posts are completed, facts are sourced, approvals are recorded, links work, scheduled items publish, routine issues are handled within scope, and reporting arrives on time. The manager influences attention and customer action, but the offer, price, capacity, reputation, season, platform distribution, landing page, and sales follow-up also affect the result. The audit should expose those dependencies instead of turning every weak month into either an excuse or an accusation.

How should you give delivery integrity 20 points before rewarding reach?

Award up to five points for a complete planned-versus-delivered record, five for on-time approval and publishing states, five for correction and failure handling, and five for usable source files and post identifiers. The record should show each planned asset, current fact source, approver, destination, scheduled profile, publication status, live link, and any delay, correction, substitution, or cancellation. A total post count without this chain cannot show whether the agreed work actually happened.

Deduct for repeated concepts counted as separate originals, silent substitutions, expired offers, missing accessibility text, broken destinations, unexplained scheduling gaps, or performance screenshots that cannot be tied to published work. Do not deduct when the manager correctly pauses an item because the owner did not verify a volatile price, date, permission, or claim. A safe pause with a documented reason is stronger delivery than an on-time mistake.

How should you give evidence and claim quality 20 points?

Award five points each for current business facts, claim support, proof provenance, and complete approval context. Sample at least five posts from the month. For every price, service detail, deadline, comparison, result, testimonial, image, and offer condition, locate the source and confirm it supported the customer-facing impression before publication. FTC guidance requires truthful, non-deceptive advertising and a reasonable basis for objective claims; informal social creative is not exempt from that standard.

Check the complete unit, not only the caption: visual, on-image text, disclosure, accessibility text, CTA, destination, profile, and timing. Review customer proof for its original meaning, permission, incentive or material-connection context, and reuse limits. A manager who catches an unsupported claim and returns it for review is protecting the business. A manager who produces polished but unverifiable content is creating hidden correction and trust costs.

How should you give customer-path measurement 25 points?

Award up to five points for native platform definitions, five for consistent campaign tags, five for working destinations, five for qualified-action reconciliation, and five for explicit attribution limits. Preserve the original source and date range. LinkedIn, for example, distinguishes estimated impressions, members reached, clicks, reactions, comments, reposts, and engagement rate. Those measures can diagnose distribution and response, but they should not be merged with unlike metrics or relabeled as leads.

Connect content to the next observable step. Google Analytics documents source, medium, campaign, and content parameters for distinguishing referral traffic and creative variants. Google Business Profile can expose calls, directions, website clicks, bookings, searches, views, and other applicable interactions. Reconcile those signals with the business’s own quote, appointment, order, or call record where possible. Report correlation and known gaps honestly; a tagged visit is not automatically a qualified lead, and a booking is not proof that one post alone caused the sale.

How should you run the scorecard on one hypothetical local-service month?

Consider a clearly labeled hypothetical contractor campaign with twelve planned posts. Eleven publish; one pauses because the business never confirms a seasonal offer. Ten posts have complete source and approval records, one needs a corrected destination, and all published links use the documented campaign convention. The platform exports show 8,400 impressions and 116 tagged website sessions. The business log shows 18 estimate-form starts, seven qualified calls associated with campaign landing visits, and four booked estimates, but it cannot reliably connect closed jobs to an individual organic post.

A defensible score might be 17 of 20 for delivery, 18 of 20 for evidence, 20 of 25 for customer-path measurement, 16 of 20 for decisions, and 13 of 15 for control, totaling 84. The missing post should not be counted as provider failure because the pause rule worked. The more important corrections are destination testing and closed-job reconciliation. The next decision is not post more; it is to fix those two evidence gaps, keep the current cadence for another cycle, and compare qualified actions per content job. This example demonstrates the method, not an expected Lumora or customer result.

How should you give useful decisions 20 points?

Award five points each when the report identifies what changed, diagnoses the most plausible constraint, proposes one bounded test, and names the decision owner and review date. A useful manager should distinguish at least five failure locations: the content did not reach the intended audience, the message did not earn a relevant action, the destination failed to convert interest, the business did not follow up, or the offer itself was weak or unavailable. Each diagnosis requires different evidence and a different response.

Reject recommendations that are detached from the record, such as post more, make it viral, or try more video without naming the customer decision, source material, channel, success measure, cost, and stop rule. Strong reporting may recommend less content when approval quality, source supply, destination readiness, or reply capacity is failing. The manager’s value is partly the ability to narrow uncertainty and protect the business from scaling a broken path.

How should you give owner control 15 points and count hidden workload?

Award five points for business-controlled accounts and recovery, five for direct access to analytics and durable files, and five for a workload and handoff record. Confirm that the business can reach its profiles, remove provider roles, export source data, retrieve final and contractually transferable assets, find approval history, and understand open customer conversations without relying on one provider-owned dashboard. Account access is operational resilience, not an administrative detail.

Measure the work the owner still performs: sourcing facts and photos, rewriting captions, correcting designs, chasing approvals, publishing, triaging replies, interpreting reports, and repairing links. Compare that retained work with the promised scope. A lower monthly fee may be expensive when the owner supplies most of the production system; a higher fee is not automatically good value when evidence, access, or decisions remain opaque. The score should reflect the workflow the business actually receives.

How should you turn the total into a renew, repair, or transition decision?

Use 85 to 100 as evidence to continue and optimize, 70 to 84 as a requirement for a written correction plan, 50 to 69 as a short remediation period with tighter access and evidence checks, and below 50 as a reason not to renew without first recovering business control and missing records. The thresholds are an operating framework, not an industry benchmark. Adjust them only before the review, document why, and never add points because a provider is likable or subtract points because one uncontrollable result disappointed the owner.

For a repair plan, choose no more than three material gaps, assign an owner and deadline, preserve the current baseline, and review one more complete cycle. For a transition, add the replacement role before removing the outgoing provider when it is safe to do so, export the operating record, reconcile scheduled work and conversations, test publishing and recovery, then remove obsolete access. The conclusion should be a concrete next move: continue the tested system, fix named constraints, narrow the scope, or change providers without losing the business’s evidence and accounts.