Back to all articlesReports and dashboards

Explain Why a Report Changes After the Reporting Period Ends

Alactic Editorial10 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

A manager opens last month's report and finds a different total from the one circulated earlier. The reporting period has not changed, but the records available to the report have. Without an explanation, the reader cannot tell whether the original was wrong, the dashboard is unreliable, or someone quietly changed the rules.

Historical dashboard totals change when late records, corrections, or revised calculation rules alter what qualifies for a past period. The practical response is not to promise that historical data will never change. It is to distinguish the period being measured from the cutoff for information included, preserve circulated versions, and explain material revisions.

A useful report therefore answers two questions: What period does this cover? and What information was included when this version was produced? A revision policy makes the relationship between those answers predictable.

1. Separate the reporting period from the information cutoff

The reporting period describes when the measured activity happened. The information cutoff describes how much of the available record history the report includes. They are different boundaries.

For a count of completed service requests, the period might be determined by the completion date. A request completed during that period can still be recorded or corrected afterward. If the report includes records according to completion date, that later information can change the historical total.

Use three distinct timestamps or labels in the reporting specification:

  • Activity date: the date that assigns a record to a reporting period.
  • Information cutoff: the latest point through which accepted records and corrections are included.
  • Publication time: when the report version was issued to readers.

A refresh time is not necessarily an information cutoff. A report refreshed this morning might still contain an incomplete set of accepted records. Conversely, an archived report opened today might intentionally show only information accepted before its original cutoff.

Also state the timezone and period boundary. Prefer an explicit start-inclusive, end-exclusive interval in the specification. This removes ambiguity about activity at midnight without requiring every reader to understand the underlying query.

2. Diagnose the kind of change before explaining it

Not every historical change has the same meaning. Label the cause before deciding whether to revise a report, issue a notice, or treat the result as a new comparison basis.

  • Late entry: a qualifying activity happened during the period but entered the accepted record set after the earlier cutoff.
  • Record correction: an existing record's date, status, ownership, or other relevant field was corrected.
  • Duplicate removal: multiple records were found to represent the same qualifying activity.
  • Definition change: the rule for counting activity changed, such as excluding a category that was previously included.
  • View change: filters, permissions, grouping, or scope differ between the two reports.

Start by confirming that the versions use the same metric definition, period, timezone, and scope. A team-only view and an all-team view are not successive versions of the same total. Nor should a new metric definition be described as merely a late-data adjustment.

For genuine revisions, reconcile the difference to identified changes. A plausible explanation is not enough. If the stated adjustments do not bridge the original and revised totals, the report is not ready to be presented as fully reconciled.

3. Choose what provisional and settled mean

Provisional should mean that the report is available for use but remains inside a stated routine revision window. It should not mean that nobody is responsible for checking it.

Settled should mean that routine updates have closed under the reporting policy. It should not mean that a known error will be preserved forever. Corrections after settlement should take an explicit exception path, with a new version and an explanation.

Three approaches are useful, depending on the reader's decision:

  • Always-current historical view: past totals incorporate accepted changes continuously. This supports current analysis, but readers need change visibility and saved snapshots to reconstruct earlier decisions.
  • Provisional window followed by settlement: routine revisions continue until a stated cutoff, after which changes require review. This balances early reporting with a stable reference version.
  • Fixed publication snapshots: every issued version remains unchanged, while later publications show updated knowledge. This preserves exactly what readers received but can leave several valid versions in circulation.

These approaches can coexist. A live dashboard can show the latest accepted total while a circulated monthly report remains a fixed snapshot. The important requirement is to label them differently and show which version supersedes which.

Choose the routine window from observed reporting delays and the decisions the report supports. A longer window may reduce later revisions, but it also delays a stable reference. A shorter window provides earlier closure at the cost of more explicit exceptions. Neither choice removes the need for correction.

4. Worked example: a filled report revision log

Fictional worked example: All dates, roles, counts, thresholds, and explanations below are hypothetical. They illustrate a proposed operating practice, not a product feature or customer result.

A service operations team reports completed requests for March. Its metric counts each unique request once, according to its completion timestamp. The reporting interval is 1 March 2026 at 00:00 UTC up to, but not including, 1 April 2026 at 00:00 UTC. The team circulates an early provisional report, then settles it after a short review window.

Report: March 2026 completed service requests
Report owner: Operations reporting lead
Metric definition: Unique requests completed within the reporting interval
Scope: All service teams; test records excluded
Reporting timezone: UTC
Reporting interval: [2026-03-01 00:00, 2026-04-01 00:00)
Routine settlement cutoff: 2026-04-05 12:00 UTC

Version 1 — Provisional
Published: 2026-04-02 10:00 UTC
Information cutoff: 2026-04-02 09:00 UTC
Original total: 1,240
Archived reference: March completed requests — v1

Version 2 — Settled
Published: 2026-04-05 14:00 UTC
Information cutoff: 2026-04-05 12:00 UTC
Starting total: 1,240
Late-entry adjustment: +36
Reason: Requests completed in March, accepted after the v1 cutoff
Duplicate correction: -8
Reason: Eight duplicate records removed from the original count
Net adjustment: +28
Corrected total: 1,268
Net change from v1: +2.3% rounded
Reviewed by: Service operations manager
Archived reference: March completed requests — v2
Supersedes: v1 for current period comparisons

Reader-facing explanation:
March completed requests have been revised from 1,240 to 1,268.
The revised total includes 36 March completions accepted after
our earlier cutoff and removes eight duplicate records.
The reporting period and counting definition have not changed.
Version 1 included accepted information through 2 April 2026
at 09:00 UTC. Version 2 includes accepted information through
5 April 2026 at 12:00 UTC and is the settled reporting version.
Use v2 for current comparisons. Retain v1 when documenting
what was available for decisions made from the earlier report.

The arithmetic is visible: 1,240 + 36 − 8 = 1,268. The explanation identifies both the additions and the removals rather than hiding them inside a net increase. It also gives readers a specific instruction about which version to use.

The earlier version remains a valid record of what was published at its cutoff. That does not make its total the best current estimate. Preserving both facts is more useful than declaring one version simply trustworthy and the other worthless.

5. Write a revision policy that someone can actually apply

A practical policy can fit on one page. It needs decisions, not broad assurances about data quality. For the hypothetical report above, the following rules would make the log repeatable:

  1. Publication status: label the first issue provisional and display its information cutoff beside the total.
  2. Routine closure: accept routine revisions through the stated settlement cutoff, then publish a reviewed settled version.
  3. Version retention: preserve every circulated version and its metric definition, scope, cutoff, and publication time.
  4. Change review: require the reporting owner to reconcile adjustments and a designated reviewer to approve settlement or post-settlement corrections.
  5. Notification: notify the original recipients when a revision meets the communication threshold or changes a decision-relevant interpretation.
  6. Post-settlement exceptions: log later discoveries, assess their effect, and issue a numbered correction when needed rather than silently replacing the settled report.

Define communication thresholds separately from accuracy rules. In this hypothetical policy, a change of at least 20 requests, at least 1%, or any change that reverses a target assessment triggers a direct notice. Smaller revisions remain visible in the log. Those thresholds govern notification; they are not permission to ignore known errors.

A percentage-only threshold can miss an important change in a small team. An absolute-only threshold can miss a meaningful rate change. Decision impact should remain an explicit override.

6. Put the explanation where readers encounter the number

A revision log is necessary, but it is not sufficient if nobody sees it. Put the status, version, and information cutoff near the report total. Place a short change note beside a revised figure, with the full reconciliation available in the accompanying report or log.

A useful notice contains five elements: the old total, the new total, the reason, the information cutoff, and the action readers should take. Avoid vague statements such as an update to the data without identifying what changed.

Use the same distribution route as the original report when a material revision affects its readers. Updating a live dashboard does not update a downloaded attachment, a copied slide, or a screenshot. Name the affected version in the notice so recipients can identify stale copies.

If a revised total changes an interpretation, say so directly. If that assessment has not been completed, state that it is still under review. Do not claim there is no decision impact simply because the numerical movement looks small.

Keep the language neutral. Late entries describe the timing of accepted information; they do not, by themselves, establish who made a mistake.

7. Recover when totals changed without a record

If readers have already spotted an unexplained difference, resist the urge to manufacture a clean history. First preserve the currently visible report and identify any reliable earlier exports or circulated copies.

  1. Confirm that the two versions have matching definitions, filters, periods, and scope.
  2. Reconcile the difference using available record history and documented corrections.
  3. Separate confirmed adjustments from any unexplained remainder.
  4. Publish a correction note with what is known, what remains uncertain, and when readers will receive the next update.
  5. Start version retention and cutoff labeling for subsequent publications.

If the earlier report cannot be reconstructed, say that plainly. An old export may establish the previously published total without establishing every record that produced it. Do not present a complete record-level reconciliation unless the evidence supports one.

A definition change needs a different recovery path. Either recalculate comparison periods on the new basis or flag a break in comparability. Silently applying a new rule to only part of a trend can make an operational change look like a change in performance.

8. Test the policy on the next reporting cycle

Choose one recurring report that readers already use. Before its next issue, assign an owner, define its activity date and timezone, select a routine cutoff, and prepare a revision log using the fields above.

At settlement, measure three things: the adjustment between provisional and settled totals, the share of that adjustment explained by logged causes, and whether every material revision received a notice. Also record whether the revision changed a decision or target assessment.

Use those observations to adjust the policy. Repeated large late-entry adjustments may justify a later cutoff. Frequent unexplained differences call for stronger reconciliation before settlement, not a longer disclaimer. A stable process should let a reader connect each published total to its scope, cutoff, and documented changes without asking for a private explanation.

ReportingDashboardsData qualityRevision policy

Put the next conversation to work.

Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.

Explore Communications