Resources · 88
Social platforms: compare metrics without false equivalence
Build a metric dictionary before comparing platforms, formats and campaigns.
· 3 min
What this guide helps achieve
- Retain the actual definition
- Write every ratio
- Separate formats and distribution
- Reconcile interfaces and exports
- Connect results to an action
Quick check
- Is there a universally good rate?
- Can unique people from several platforms be added?
- Does a high CTR guarantee more distribution?
Step-by-step method
- 01
Retain the actual definition
Record each indicator’s label, source, consultation date and export method. Views, impressions and people are different units. YouTube documentation specifies that impressions CTR covers registered impressions rather than all views. Do not generalize this scope to other platforms.
Deliverable: sourced metric dictionary.
- 02
Write every ratio
Record included interactions, denominator and exclusions. Engagement divided by followers is not directly equivalent to engagement divided by reach. Adding rates without their underlying counts produces misleading results. When definitions differ, show separate series.
Deliverable: formulas and comparison rules.
- 03
Separate formats and distribution
Distinguish short video, long video, posts and live streams; separate paid and organic distribution where available. Document observed duration, publication date and content age. A recent campaign can have only a partial observation window; older content may still receive views.
Deliverable: segments and observable windows.
- 04
Reconcile interfaces and exports
Compare identical content at a common cutoff. Note rounding, availability delays and access limitations. Estimated unique people should not be summed across platforms as distinct individuals because populations may overlap.
Deliverable: reconciliation and known limitations.
- 05
Connect results to an action
Identify the decision before interpreting results, such as changing format, clarifying a message or revising distribution. Association between engagement and sales does not establish causation. Include qualitative signals such as useful questions, resolved problems and feedback warranting corrections.
Deliverable: recommendation with evidence and limits.
Reusable worksheet
Complete with your authorised observations. These fields are a working template, not observed results.
| Field | Information to record |
|---|---|
| Definition | Exact label, documentation and unit |
| Ratio | Numerator, denominator, exclusions and formula |
| Scope | Format, distribution, period and cutoff |
| Decision | Proposed action, evidence and comparison limit |
Worked example
Illustrative situation
Fictional example: a report compares a follower-based engagement rate with an impression-based rate.
Decision and expected evidence
The team retains both measures, displays formulas and compares changes under stable definitions instead of ranking platforms by a single value.
Distinguish the mechanisms
| Mechanism | Purpose | Check or limitation |
|---|---|---|
| Impressions | Count platform-defined exposures | One person can generate several |
| Estimated unique people | Estimate distinct people within a scope | Do not sum without deduplication |
| Engagement rate | Divide interactions by a base | State formula and denominator beside the value |
Management indicators
| Indicator | What it measures | First action |
|---|---|---|
| Explicit definitions | Decisive indicators with visible units and formulas | Complete before publishing rankings |
| Export discrepancies | Explained differences for identical content | Align cutoffs and scopes |
| Justified actions | Recommendations linked to comparable observations | Remove unsupported conclusions |
Common pitfalls
- One person can generate several
- Do not sum without deduplication
- State formula and denominator beside the value
Frequently asked questions
Is there a universally good rate?
No. Formula, format, distribution and population change interpretation. Use comparable references rather than a context-free threshold.
Can unique people from several platforms be added?
The total may not represent distinct people. Show platform populations and disclose the absence of deduplication.
Does a high CTR guarantee more distribution?
No. Examine scope, duration and other signals. YouTube documentation notes that not every view originates from a counted impression.
Further reading and tools
Editorial resources to extend the method.
- WORLD SOCIAL NETWORKS — audience metrics External link
Extend your review of indicators to social-network use cases.
Official references
References consulted: . The method and worksheet propose checks to adapt to your context; they do not constitute certification.
Continue exploring
Related analysis
