Reading nonprofit social media benchmarks without fooling yourself

The Facebook calculation is a mismatch test

The constructed Facebook example gives 0.95%: 95 comments, shares, and saves divided by 10,000 followers across 30 days, excluding likes and reach. Published benchmarks define engagement differently, so a median is comparable only when its numerator (which interactions count), denominator (followers or reach) and period match yours. Rival IQ's industry benchmark report, for example, divides all interactions by total follower count. Quote the edition year and the exact definition beside any median you cite.

Any gap between your rate and a published median is directional, not a verdict: audience, duration, formula, and page maturity differ, and a published sample may include much older pages. Before using the contrast to set a threshold, compare local cost per engaged follower and verified downstream donations.

Attach permission to beneficiary evidence

A beneficiary story belongs in a benchmark claim only when documented permission exists. If that record is missing, leave the story out rather than treating consent as an afterthought.

Keep the permission record with the person responsible for corrections, so any later issue has an owner. If a problem appears in an internal or public output, preserve the affected record, identify the material to repair, and correct it where it appeared. Consent and correction responsibility belong alongside the claim, not in an unrelated file.

Trace a published statistic to its inputs

The figure is usable only to the extent that its inputs, arithmetic, and exclusions can be checked. Recreate the result from source records, then ask whether the proposed use matches the purpose for which the data were collected.

The Government Accountability Office's guide, Assessing Data Reliability (GAO-20-283G), links data quality with intended use and a documented assessment. Apply it to a published median gift. Suppose a report states a median online gift and you want to compare it with your own. List your gifts for the same period and gift type, such as one-time online gifts only. Sort them: $15, $20, $25, $25, $35, $50, $50, $100, $250. The middle value is $35, so your median is $35, while the mean is about $63, pulled up by one large gift. Comparing a published median with your mean would mislead, and so would mixing monthly gifts into the list when the report excludes them. If a report withholds its inputs, label the figure unverifiable.

Your own records set another boundary: an outside donor acquisition benchmark can frame a threshold, while a retention benchmark cannot certify gift-record accuracy. External reliability guidance helps structure review; it does not establish that a program change caused your outcome. GAO's Designing Evaluations explains how evaluators test for cause.

Let local economics set the action line

The sector median is context, not a target. For an outside reference point, the Fundraising Effectiveness Project's Q4 2025 report, drawing on 15,102 nonprofits, put overall donor retention at 43.3% for 2025, up from 43.1% in 2024, while the number of donors fell an estimated 3.6%. A donor who arrives through social media is judged against that kind of figure only after you compute retention the same way. Use your own cost per donor, retention, and monthly giving churn to decide which result deserves attention. Record the threshold owner and decision date before a campaign starts, rather than converting the published number into a quota.

An underperforming reactivation result calls for diagnosis, not an automatic failure label. Compare cost per reactivated donor with expected lifetime value, and keep the decision rule in a reactivation campaign. A strategy review checklist can preserve the question, evidence, and rationale before results arrive, reducing post hoc threshold changes.

Narrow the channel comparison before investing

Use a published channel result only as directional evidence when your audience, calendar span, formula, and operating maturity are reasonably similar. Where a definition diverges, narrow the conclusion and write down the mismatch instead of treating the figure as a local performance verdict.

Audience fit concerns who the measure represents; the period concerns season and duration; engagement requires the same calculation; maturity reflects time operating with current staffing. These limits should inform a marketing strategy choice, including which social channel receives investment.

Make expansion reversible

Expansion needs an evidence case and a stop rule before spending grows. Consider ordinary and failure paths, record uncertainty, and avoid describing an association between activity and results as proof that the activity caused them.

Keep the engagement definition, downstream action, cost, source limits, owner, and review date in one decision record. Add claims, inputs, versions, alternative explanations, and unresolved comparison limits. Write evaluation questions before choosing a method, and keep a strategy benchmark beside the assumptions it depends on.

The external framework can sharpen the questions and method choice, but it cannot replace local evidence. Make the file show what is known, what remains uncertain, and what finding would trigger a pause or reconsideration. This keeps expansion tied to evidence the organization can inspect, rather than to confidence created by a comparison alone.