Treat coverage as a source-by-source record, not a single yes-or-no feature. Separate public visibility, available retrieval routes and what your configured subscription actually collects.
Compare narrative intelligence platforms and software on our homepage, then validate the sources required by your brief. Keep unresolved coverage questions attached to the buying decision and the resulting reports.
Distinguish three layers of visibility
A post may be publicly visible. A supported source route may expose some public content under particular access conditions. A vendor then applies collection settings, package limits, filters and timing. These layers can produce different boundaries.
Use a known public example to investigate a gap, but do not assume its visibility guarantees inclusion through every route. Ask where a restriction occurs and how the limitation appears to the customer.
Build a coverage matrix for your actual brief
| Field | What to record | Why it matters |
|---|---|---|
| Named source | Specific platform, site, forum or feed | A category label is not a source commitment |
| Eligible content | Posts, replies, media or selected accounts | Different events may use different routes |
| Time scope | Historical window, start date and collection intervals | Defines what a timeline can show |
| Languages | Collection and analysis tested separately | Finding text is not proof of correct interpretation |
| Operational status | Connection owner and interruptions | Missing data can resemble inactivity |
| Retained evidence | Fields, archives, exports and retention period | Determines later reviewability |
Mark each requirement as verified, unconfirmed, unavailable or outside the brief. Avoid collapsing these states into one green checkmark.
Separate history, retention and preservation
Historical access describes content available before monitoring starts. Retention describes how long collected records remain accessible. Preservation describes what source evidence is retained if an item changes or disappears. None of these automatically establishes a complete edit history.
Ask for source-specific windows and an export sample. A set of URLs, a saved text record and a page archive are different outputs. Document publication time and collection time so a later import is not presented as earlier visibility.
Test language and media interpretation
Use representative examples from the languages in your brief, including local terms, mixed-language posts, quotation and sarcasm. Check narrative grouping and stance as well as whether the content was collected.
For images, video and audio, ask which text or signals are actually analyzed: captions, OCR, transcripts, visual features or another defined input. A product’s support for one media type does not establish equivalent coverage on every source.
Record manual reviewer needs and unresolved cases. NIST’s evaluation guidance supports documenting what cannot be measured or established, rather than hiding those gaps behind a single performance figure.
Investigate a missing result systematically
Check source eligibility, connection health, query rules, dates and package limits. Then inspect whether the item was collected but excluded by a display or classification filter. Record the cause rather than treating every absence as a collection failure.
If the cause remains unknown, retain an unresolved gap. Do not report a decrease in conversation until you have checked whether collection or configuration changed during the same period.
Carry coverage limits into the final report
State the sources, languages, collection period and material gaps beside the findings. Describe an ‘earliest observed source’ when the evidence does not establish the first publication. Qualify network relationships and narrative volumes by the dataset actually reviewed.
The report example shows how to keep scope visible. Revisit the selection checklist when a critical source is unsupported, and use the monitoring workflow to maintain the register after launch.
Frequently asked questions
Does ‘multilingual’ mean the platform interprets every supported language equally well?
Not necessarily. Ask how collection, translation, grouping and stance analysis are validated for your languages. Use reference examples and record where human review remains necessary.
Can a narrative platform recover deleted content from before monitoring began?
Do not assume it can. Request evidence of an available earlier record and its provenance. Historical access and an archive of previously captured content are different capabilities.
What should I do if a key forum is missing from a vendor’s source list?
Treat it as a material gap and ask whether a supported route exists. Evaluate an alternative source strategy or provider without assuming the platform’s general web coverage includes that forum.
How do I describe source gaps without making the report unusable?
State what was observed, the scope that supports it and which conclusions remain unresolved. A bounded finding can be useful when its limits are clear.
References & further reading
These references support the concepts or product-scope statements identified in the text. The checklists and illustrative workflows are editorial guidance, not certification requirements or independently measured vendor results.