Guides
Every section we cover, with the explainers that go with it.
Attribution and experiments
Model
- How do I measure dark social? Dark social traffic cannot be measured precisely, but you can estimate it by correlating direct-traffic spikes with social posting cadence and running on-site referral surveys.
- How do social media lift tests work? A social media lift test isolates incremental impact by comparing a holdout group against an exposed group, typically revealing platform attribution overstates contribution by 30-60%.
- How does social media attribution work? Social media attribution models assign conversion credit by rules and windows—platform self-attribution, GA4, and experiments each answer a different question.
- What are UTM parameters? UTM parameters are the five-parameter URL tagging convention for social attribution—govern them with a controlled vocabulary to prevent data fragmentation in GA4.
- Why do analytics tools disagree? Social media analytics tools disagree because of differences in metric definitions, time zones, identity resolution, cookie consent loss, and API data completeness—not errors.
Content analysis
Model
- Can sentiment analysis be trusted? Social media sentiment analysis is a probabilistic classifier, not a measurement—validate vendor scores against human-coded samples before trusting trends.
- How do I audit social media content? A social media content audit inventories posts, applies a taxonomy, and groups outcomes by comparable cohorts—without claiming one post's performance predicts another's.
- How do I identify evergreen content? Evergreen social content sustains organic reach beyond the 48-hour decay window—identify it by isolating organic from paid reach over 12 weeks minimum.
- How should posts be tagged for analysis? Design a social media content tagging scheme with mutually exclusive categories, documented definitions, and a governance process that prevents taxonomy drift.
- What makes a social benchmark fair? A fair social media benchmark matches platform, audience, format, period, and distribution—not just the metric—and treats vendor datasets as biased samples.
Metrics
Model
- Reach versus impressions: what is the difference? Reach counts unique users; impressions count total displays. Platforms count them differently—Meta uses pixel thresholds, X counts any serve—so cross-platform comparison is invalid.
- What does video completion rate mean? Video completion rate depends on what counts as a view. Platforms use 2-3 second thresholds; meaningful views use 15 seconds, creating a 3-10x gap.
- What is social media engagement rate? Engagement rate divides interactions by reach or impressions. The denominator choice changes the number materially; native platforms default to reach.
- Which follower metrics matter? Follower count is vanity; net change, audience quality, and engagement ratio matter. Demographic data is modelled from a subset, not census-level.
- Why do platform metrics change? Platform metrics change due to product updates, privacy rules, and API deprecations. Historical data may be retroactively recalculated against new definitions.
Privacy and data access
Model
- Do social analytics need consent? Whether social analytics need consent depends on jurisdiction, purpose, and data source—distinguish first-party native analytics from third-party listening.
- How do I evaluate a social analytics vendor? Evaluate a social analytics vendor on data access legitimacy, security certifications, export granularity, deletion controls, and verifiability of claims.
- How should social data be retained? Retain social analytics data based on purpose and minimisation principles, distinguishing between raw identifiable data and aggregated anonymised reports.
- What are the limits of social media APIs? Social media APIs are constrained by rate limits, access tiers, data completeness gaps of 5-15%, and policy changes that can remove access without notice.
- What personal data can social analytics contain? Map the identifiers, content, and inferred data that can constitute personal data in social analytics, from handles and device fingerprints to sentiment scores.
Reporting
Model
- How do I explain a performance change? To explain a social media metric change, decompose the metric into its components, compare against a relevant cohort, and check for platform definition changes before attributing cause.
- How often should social reports be updated? Match social media reporting frequency to decision cadence: daily for operators, weekly for marketing leads, and monthly or quarterly for executives, not to real-time data availability.
- How should uncertainty be shown? Show uncertainty in social analytics reports using ranges, caveats, and sensitivity analysis rather than false precision, and remove any chart without a decision owner.
- What belongs in a social media dashboard? A social media dashboard needs an operational layer for the team and a strategic layer for decision makers, with every chart tied to a named decision owner and labelled for data latency.
- What is a useful social media KPI? A useful social media KPI connects a controllable action to a business outcome, specifies a time horizon, and has a named decision owner who can act on it.