Careers
Part of Social Media Community Building: Purpose and Practice
Social media community building case studies: the mechanism behind seven named cases and the evidence limit of each
Social media community building case studies: the mechanism each of seven named cases used, the evidence limit on each, FTC rules for rewards, and a 12 week test for one mechanism.
Use a case study to decide what question to test, not to borrow its result. Follow the branch that matches the evidence in front of you: member fit, source interest, result quality, missing context, test capacity, or public claims. If no branch supports a careful decision, keep the case at idea stage.
If the case reflects a need your members have
Choose a case only if its described activity responds to a need that members in your own group share. A published example can suggest a mechanism, but it does not show that the mechanism changed behavior or will suit your community.
The cited accounts describe Sephora member advice, plus two platform-published accounts used here as worked examples, Discord's use of peer resources and LinkedIn's Lemonade involving employees and leaders replying to people who write in; use those strategy case comparisons only to identify candidate mechanisms. The source set also names LEGO Ideas, Notion, Figma, Salesforce, Duolingo and Glossier, without giving outcome figures for them.
Discord vs LinkedIn evidence
Discord account
- Publisher
- Platform
- Mechanism
- Peer resources, staff space
- Useful next step
- Test repeated-problem support
- Main limit
- No independent outcome test
LinkedIn account
- Publisher
- Platform
- Mechanism
- Named participation, response
- Useful next step
- Test expert response workflow
- Main limit
- Vendor-selected customer story
Use the comparison as a way to describe activity, not to rank brands. Match the member problem with the mechanism and with the capacity your team can sustain. Content planning examples provide another reference point, but a recognizable name or attractive format cannot establish local fit.
- Rule of thumb: Choose for the member need, not the brand name.
What each named case did
These are descriptions of public programs, drawn from the companies' own pages and secondary write-ups. None supplies an audited outcome, so read each as a mechanism to test.
- Sephora Beauty Insider Community.Launched in 2017 for loyalty members. Members ask and answer questions in a forum and on product pages, and highly engaged members can become community ambassadors. The mechanism is peer answers. Limit: the company describes its own program, and no figure for answers or ambassadors has a denominator.
- LEGO Ideas.Fans submit set designs and gather supporters. An idea that reaches 10,000 supporters goes to a LEGO review board, and only some become sets. The mechanism is public voting with a staff gate. Limit: 10,000 is an entry rule, not a result, and no conversion rate to released sets is given.
- Notion.A volunteer leader program, reported as renamed Notion Champions, in which power users host meetups, make templates and mentor others. The mechanism is member-led teaching. Limit: the renaming comes from secondary reports, and no count of leaders, events or templates is given.
- Figma.Friends of Figma is the official user group program, with local chapters whose leaders run events. Figma Community lets users publish files and plugins that others duplicate or install. The mechanism is local leaders plus shared files. Limit: no chapter count, attendance or download figure comes with a comparison period.
- Salesforce.Trailblazer Community pairs an online forum and groups with volunteer-led local groups that meet at least quarterly, tied to Trailhead learning. The mechanism is learning plus local leaders. Limit: the quarterly meeting is a group requirement, not a measured outcome.
- Duolingo.Ran discussion forums, events and clubs. Reports say the forums closed in March 2022 and the events program was later retired, though sources differ on dates. The case shows that a program can be stopped. Limit: the dates differ by source, and no effect on learners is documented.
- Glossier.Grew from the Into The Gloss blog, where reader comments informed products, then added a Facebook group and a rep program for loyal customers who earn commission. The mechanism is feedback plus paid advocates, which is why disclosure matters. Limit: no sales, retention or reach effect is given for the group or the rep program.
Two further accounts, Discord's and LinkedIn's, are not among the seven. They show the same reading on platform-published material. Discord's peer-resources account and LinkedIn's account are both published by the platform that ran the activity, so each is interested evidence. Neither supplies a denominator, comparison period or data owner for its outcomes. Both can suggest a mechanism to test, such as members answering each other, and neither can be cited as a benchmark.
If the publisher also ran the activity
Read the publisher's account as a description rather than independent verification. The publisher may still identify a mechanism worth investigating, but its involvement alone cannot confirm that reported outcomes are accurate.
Before relying on a result, identify who made the change, who controls the underlying records, and whether anyone outside the work checked the outcome. If those roles are unstated, mark the account as interested evidence. Use measurement repairs to separate what a reported metric can show from what remains unverified.
- Rule of thumb: Use an interested source to find ideas, not validate outcomes.
If a reported result seems strong
Do not use a persuasive-looking result as a benchmark until you know what the number represents. An isolated figure without a basis for comparison cannot show how much changed or whether the result applies to your circumstances.
The named case pages do not supply an audited outcome figure with a checkable denominator. Record that limit rather than infer missing detail. Compare growth case accounts for reporting context, and use the social analytics checklist to keep each metric connected to the decision it is meant to inform.
| If you need to decide | If the account omits | Then do this |
|---|---|---|
| Can the outcome be checked? | A denominator, period, comparison, or data owner | Do not use it as a benchmark |
| Can the source be located later? | A publication date | Record your access date |
| Does the activity fit members? | A recurring question is evident | Consider peer answers or member advice |
- Rule of thumb: Without a denominator and comparison, do not treat a result as a benchmark.
If important context is missing
If member value, group boundaries, or operating details are absent, treat each gap as unanswered. Do not fill it with assumptions; missing context limits your ability to judge whether the activity could fit your community.
Check whether the account addresses roles, a first useful action, programming, safety, privacy, measurement, paid support, verification, and limits on transfer. Record its publication date when stated; otherwise note the date you accessed it. Keep these questions together using common analytics questions, rather than silently completing the source's account.
Case-study review template
- Member problem and promised value
- Community boundary and platform context
- Member, staff, moderator, partner roles
- Onboarding and first useful action
- Programming and contribution mechanism
- Safety, privacy, rules, and appeals
- Measurement period, denominator, comparison
Use the checklist to keep limits visible, not to award a score. An omitted detail is a reason to pause or ask a question, not permission to supply an answer yourself. The record should make clear which parts come from the source and which remain unknown to your team.
- Rule of thumb: Treat omissions as evidence limits, not invitations to guess.
If you can test one mechanism locally
Begin with a problem you have already observed and set the boundaries before activity starts. This lets you assess local results without presuming another organization's experience will transfer unchanged.
For example, a twelve-week peer-answer trial could select a repeated question, choose peer resources or member advice, specify an outcome and window, name the data owner, disclose contributor terms, and log failures as well as successes; keep that scenario a local test, supported by benchmark planning.
Include moderation hours, because the test can fail if the team lacks staff time. Keep the selected measure and its owner visible through the trial. Growth questions can surface issues to investigate, but they do not replace a defined test or a record of what happened.
- Rule of thumb: Name the measure and its owner before the first contribution.
If rewards or public claims are involved
Pause before inviting contributions or publishing results. Separate what the case says from what you can substantiate about your own activity, and make selection, payment, and conflicts visible before contributions begin.
For endorsement questions, consult the FTC guidance on endorsements, influencers, and reviews, and for general claims consult the statutory text on unfair or deceptive acts. That is Section 5 of the Federal Trade Commission Act, 15 U.S.C. 45, while Section 12, 15 U.S.C. 52, covers false advertisements. The FTC's Guides Concerning the Use of Endorsements and Testimonials in Advertising are published at 16 CFR Part 255. Influencer record planning can help structure related records, but neither the case nor a planning resource determines which requirements apply.
- Rule of thumb: Make relationships clear and support claims before publication.
