Reliable Twitter analytics begins with a consistent record of what you collected. A precise chart can still mislead if its posts, dates or metric definitions changed between periods.

Keep a small data dictionary

For every field, record its source and definition. Distinguish native impressions, public post views, likes, reposts, replies and website conversions. State whether a value is a captured count, a calculated rate or an estimate. Keep account and post IDs as strings so spreadsheet rounding does not change them.

Keep a collection log

  • Account and query or profile scope.
  • Collection and export dates, including time zone.
  • Selected publication period and captured post count.
  • Refresh schedule and known failures.
  • Missing fields, deleted/unavailable posts and access limits.

An archive documents what was captured or imported. It does not prove every matching post was available. Tweet Archivist’s storage and archive limits depend on the plan; retention duration is not unlimited historical collection.

Check quality before calculating

Deduplicate by stable post ID. Look for missing dates, rounded IDs, duplicate rows and zeros used in place of unknown values. Separate original posts from replies/reposts and paid activity from organic activity.

Review a few source posts against the export. A total that looks unusual can reflect a changed export schema or collection gap rather than an audience change.

Report with context

Use the same rate definition and comparable observation windows. Show the sample size, totals and typical per-post results. An average can be dominated by one unusual post. Label all illustrative numbers as hypothetical.

Do not replace your goal with a universal target such as a fixed monthly follower-growth percentage. A support team might measure resolved questions; a publication might measure qualified visits. State the chosen outcome and why it matters.

Make review repeatable

  1. Save the source export unchanged and protect personal data.
  2. Document cleaning and calculations in a separate copy.
  3. Write observations separately from explanations.
  4. Choose one action, owner and review date.
  5. Compare the next period without silently changing definitions.

Use the metric glossary, dashboard specification and safe Excel export guide when building the workflow.