General

Why can't I open Reports or use one of its features?

Reports uses two layers of access:

Workspace role permissions control whether you can open Reports and perform actions. Separate permissions cover access, creating and editing, deleting, scheduling deliveries, exporting CSV, chart drill-in, and managing workspace folders. An admin manages them in Settings → Members and teams
Report access controls what you can do with an individual saved report. Full access, Explore only, and View only do not override a missing workspace role permission

For example, exporting from a shared report requires both enough access to that report and the Can export CSV role permission. See Report sharing and access controls for the full model.

Creating and saving reports

Why did my filters or other report changes disappear?

Changes made while viewing or exploring a saved report are temporary. This includes the date range and report-level filters, so you can investigate without changing what teammates see. Use Edit and then Save when you want to update the saved report.

If you edit a report and select Cancel or leave without saving, Featurebase discards the unsaved changes, including chart configuration and layout changes. Teammates with Explore only access can explore but cannot save changes.

Why can't I add a report-level filter?

The report-level filter picker only shows attributes supported by every chart in the report. An attribute may be available as a filter on one chart but disappear at report level when another chart uses a dataset that does not support it. Apply it as a chart-level filter instead, or check its exact dataset support in Reporting metrics and attributes.

Why can't I use an attribute as a breakdown?

Not every attribute supports every use. Some are filter-only, some are breakdown-only, and some appear only as drill-in or export columns. The picker only offers attributes supported by the selected metric and use; the attribute dictionary lists each attribute's supported uses and datasets.

Data availability and counting

When does reporting data start?

Reporting data starts on July 12, 2026, with no backfill from before that date. Date pickers do not allow earlier dates, relative ranges that reach further back only show data from July 12, 2026 onward, and scheduled deliveries and exports follow the same boundary.

How fresh is reporting data?

Reporting data updates near real time, but brief delays can occur. If a number looks behind, reload the report and check again shortly.

The Last updated value in the reports list tells you when someone last edited the saved report. It is not a data-freshness timestamp.

Which timezone does Reports use?

Saved reports use your workspace timezone by default, and you can change it from the report header. Drill-in uses the report timezone shown above its table. Standalone dataset exports use the local timezone shown on the export page, while a scheduled delivery runs in the timezone shown in its schedule editor.

When comparing two results, check both the date range and timezone. A timezone difference can move activity near midnight into a different day.

How are conversations counted?

New conversations has two timestamp variants. Conversation created at counts every conversation when it is created, including conversations with no customer reply. Conversation started at counts inbound and outbound conversations when their first customer message is received.

Response and resolution metrics use narrower eligible populations. A conversation with no visible teammate reply has no first response time, and a conversation never closed by a teammate has no time to close. This is why a volume chart can show more conversations than a response-time or resolution chart over the same period. Filter by attributes such as Has customer reply or Conversation initiator type when you need a narrower volume population.

Why don't Reports numbers match what I see in the Inbox?

The Inbox and Reports answer different questions. Inbox views usually describe the current workload, while a report includes rows based on the selected metric's timestamp, eligible population, date range, and filters. For example, a conversation can be open now but still contribute a reply or assignment event to an earlier reporting period.

Unexpected results

Why does a chart show no data?

Start with the date range, timezone, report and chart filters, and the chart's info card. The selected metric may also require a narrower population, such as conversations with a visible teammate reply or a teammate close. Follow the custom report troubleshooting checklist to isolate the cause.

Why do two similar charts show different numbers?

Charts that look alike can still use different:

Metric variants or timestamp bases
Aggregations, such as Average instead of Median
Office-hours settings
Chart-level filters on top of the report-level filters
Date ranges or timezones

Compare the charts' info cards and report headers side by side. Reporting metrics and attributes defines each metric variant and timestamp.

Why did a past reporting period change?

User, company, and custom attributes preserve the value captured for each historical reporting row. Later profile or company edits do not rewrite earlier results.

Current-state attributes, including current assignment, team membership, conversation tags, priority, and conversation state, can move historical rows between groups when their value changes. Close-dependent metrics can also appear later when a conversation closes. Troubleshooting custom reports explains which values can change and when to use action-anchored attributes instead.

Why do breakdown groups add up to more than the total?

Multi-value attributes can place one row in several groups. A conversation with two Conversation tags, for example, counts once under each tag; Teammate replied to can similarly place one conversation under every teammate who replied. This is expected. Use a single-value attribute such as Channel when the group totals need to reconcile to the overall total.

Why are values missing from a breakdown?

Breakdowns show the number of leading groups selected under Show top, rather than every unique value. Increase Show top or narrow the report with filters. Donut charts automatically combine the remaining values into an Other group. Custom report chart types and options explains these controls.

Why don't my filters seem to work?

Report-level filters apply to every chart, while chart-level filters apply only to one chart. The two levels combine, so a row must match both to be counted. Temporarily clear the report-level filters and then the chart-level filters to identify which condition is excluding the rows.

Why can drill-in show more or fewer rows than the chart value?

Some charts count unique conversations while drill-in shows the individual event rows behind them, so one conversation can appear several times. Drill-in can also show fewer rows than a chart because it respects your Inbox conversation permissions, even when the aggregated chart includes conversations you cannot open.

Explore report data with drill-in explains both cases and how drill-in columns, filters, and CSV exports work.

Why does my CSAT score show 100% or 0% after filtering?

CSAT score is the percentage of positive ratings out of all included ratings. Filtering the chart to positive ratings makes the score 100%, while filtering it to negative ratings makes it 0%. To compare CSAT across a population, filter by conversation properties such as team or channel instead of filtering by the rating itself.

Metrics and time

Where can I check exactly what a chart measures?

Hover the info icon next to the chart title. The card shows:

The metric description
The Timestamp that decides which reporting period receives the value
The Aggregation
Report-level and chart-level filters
The office-hours state when the adjustment is enabled

The full definitions and dataset applicability live in Reporting metrics and attributes.

Why are there several versions of the same metric?

Some variants change where a clock starts; others change which period receives the result. First response time (started, including bot) measures the full wait from conversation start, including time in the Bot inbox. Excluding bot variants start when the conversation permanently leaves the Bot inbox for teammate handling, or at conversation start if it never entered the Bot inbox. First replied variants bucket the result by when the first reply was sent instead of when the conversation started.

Choose the variant that matches the question you are answering. The metric dictionary defines every variant.

Should I use Average or Median?

Use Average when every value should influence the result. Use Median when you want the middle result and do not want a few very long conversations to dominate the chart. For response and resolution durations, compare both when investigating outliers.

Why do response or resolution times include nights and weekends?

Duration metrics use calendar time by default. For supported metrics, turn on Within office hours to count only time inside the applicable office-hours schedule. Office hours in reporting lists the supported metrics and explains how Featurebase chooses the schedule.

What are per-active-hour metrics?

Per-active-hour metrics compare teammate output with the time recorded as active in the Inbox. There are five:

Conversations closed per active hour
Conversations assigned per active hour
Conversations replied to per active hour
Conversations participated per active hour
Teammate replies sent per active hour

The first four divide a conversation count by teammate active time. Teammate replies sent per active hour divides visible replies by active time. Office hours and working schedules do not change the denominator.

Why does a per-active-hour result look unusually high or unstable?

Teammates with very little active time in the selected period can show large swings because the denominator is small. Widen the date range and break the chart down by teammate before comparing results. Active time is recorded automatically while teammates are active in the Inbox; there is no manual clock-in.

How is Teammate handling time measured?

Teammate handling time measures eligible time while a conversation is assigned to a teammate and becomes reportable when the conversation closes. Adjusted teammate handling time narrows this to time when the teammate is actively present in the conversation. Teammate performance reporting explains the exclusions, office-hours behavior, and when to use each metric.

Why can teammate or team filters return different conversations?

Teammate and team attributes answer different questions. Currently assigned attributes follow live ownership, Teammate is a member of uses current team membership, and action attributes preserve who replied, closed, handled, or owned the conversation when the action happened. The dynamic Teammate and Team filters adapt to each metric, so the same selected person can match a different role across charts. Use Teammate filters in reports when you need a specific attribution point.

Sharing and access

Who can see the reports I create?

Only you, until you share them. New reports are private by default. Use Share → Manage access to grant teammates or roles Full access, Explore only, or View only, or set a workspace access level. Report sharing and access controls explains what each level allows.

Why can't a teammate open a report I shared?

They need both the Can access Reports workspace role permission and access to the individual report. If the report is restricted, confirm that the teammate, one of their roles, or the whole workspace has been granted access. Actions such as editing, drill-in, exporting, deleting, and scheduling also require their matching workspace role permissions.

Does Featurebase notify teammates when I share a report internally?

No. Granting access does not send an email notification. Use Copy link in the sharing drawer and send the URL to the teammate yourself. The link does not grant access, so they still need the required role permission and report access.

Can people outside my workspace see a report?

Yes. You can share the report externally as a public read-only snapshot link or schedule report delivery to email fresh snapshots to external addresses. Recipients do not need a Featurebase account.

Anyone with a public snapshot link can open it without logging in, so review the report for sensitive data before sharing it.

Reports containing Table charts cannot be shared through a public link. Remove every Table chart before generating the link.

Does an external snapshot update when the report changes?

No. A public link is a frozen copy of the report at the moment it was generated. Editing the report or receiving new data does not change an existing snapshot. Each scheduled delivery generates a fresh snapshot using the report at that delivery time, so saved changes affect future deliveries only.

How long do external snapshot links work, and can I revoke one?

Snapshot links expire 30 days after they are generated. You can revoke an individual link early from Share → Manage access → External access → Active links.

Generating a replacement does not invalidate earlier links, so revoke any link you no longer want people to open. See Share reports externally for the full security and expiry behavior.

Exports and programmatic access

How much data can I export?

Chart and drill-in CSV exports include up to 10,000 rows. Manual and scheduled dataset exports include up to 100,000 rows. Export download links remain valid for 24 hours. Export your reporting data explains each export path.

Why did my export fail or why wasn't a scheduled dataset export sent?

An export can fail when it exceeds its row limit. A manual dataset export above 100,000 rows fails with an error, while an oversized scheduled dataset export is not sent and the schedule owner receives an email that {schedule name} could not be delivered. Featurebase does not automatically split oversized exports or send the overflow later. Recreate an oversized scheduled export with a narrower date range or filters. For a manual export, narrow the date range, add filters, or split it into smaller ranges.

For chart and drill-in exports, narrow the result below 10,000 rows or use a dataset export for a larger extract.

Why can't I see an export option?

CSV exports require the Can export CSV workspace role permission. Exporting or drilling into a shared report also requires Explore only or Full access to that report; drill-in separately requires Can access Chart drill-in. Ask an admin to check both your role and the report's access settings.

Can I export a report as a PDF?

No. Report data exports are CSV only. To share a formatted, read-only version of the report, create an external snapshot or schedule report delivery.

Can I access reporting data programmatically?

Yes. Use the public Reports API to discover the catalog, query metrics, retrieve filter values, and drill into reporting data. You can also connect AI tools through the Featurebase Reader MCP. Both are documented in the Featurebase developer documentation.

Still stuck?

What should I send to support when a report still looks wrong?

Include enough context to reproduce the result:

The report link and chart name
The metric, aggregation, and result you expected
The date range and timezone
Every report-level and chart-level filter
A screenshot of the chart and any error message
Whether the problem also appears after reloading the report

If you already have a Messenger conversation open, send the details there. Otherwise, email support@featurebase.app.