Troubleshooting

The problems that actually come up, and the order to check things in.

Four checks first

Most “the platform is broken” reports resolve at one of these four, so start here every time:

Which workspace are you in? The top-bar switcher scopes almost every screen. Calls uploaded to one workspace never appear in another.
What is your date range? The dashboard defaults to the last seven days and remembers whatever you last set — including a range from a fortnight ago.
Is a filter still applied? A saved filter or an advanced filter left in place will quietly hide most of the list.
Has it finished processing? Uploaded is not processed. Check the Running and Queued counts on Usage.

Failed calls

Every call that fails processing is listed on the Failed Calls screen — reachable from the FAILED status card on Usage — with a date filter and a workspace selector.

Column	Use
Call ID	The identifier to quote to support.
Filename	Which recording it was, so you can find it in your storage.
Uploaded Time	When it arrived — useful for correlating with a bad batch.
Reason	Why it failed. Read this before re-processing.
Action	Re-process runs the call again.

Read the reason first. Re-processing helps for transient failures; it will not help a file that is genuinely unreadable or empty.

Pattern in the reason	Likely cause	Fix
Format or codec related	The file isn’t one of the supported types, or is a container we can’t decode.	Re-export as MP3 or WAV and upload again.
Duration related	The call falls outside the workspace’s minimum or maximum duration limits.	Adjust processing limits if the call should have been processed.
Empty or silent audio	A dropped call, or a recording that captured nothing.	Nothing to fix — set a minimum duration to stop these consuming minutes.
Access related	The connected source’s credentials no longer allow reading the file.	Re-authorise the source, then re-process.
Balance related	Insufficient minutes at the time of processing.	Check remaining duration on Plan & Billing, then re-process.

Re-processing consumes minutes

A re-run costs the call's duration again. Before re-processing a large failed batch, fix the underlying cause on a single call first and confirm it succeeds.

Calls are not appearing

Nothing at all in Recent Calls Check workspace, then date range, then filters, then the Queued and Running counts on Usage. The empty state says "No Calls" whether the workspace is genuinely empty or your filter excluded everything.
Some calls missing Search by filename or call ID — if it's found by search, a filter or the date range was hiding it. If it isn't, check Failed Calls.
One agent's calls missing Almost always attribution — see Calls on the wrong agent.
Bookmarked calls missing Bookmarks respect the date range. Widen it before concluding a bookmark disappeared.

Upload problems

File rejected at selection — the type isn’t supported. Accepted: MP3, WAV, WEBM, MP4, M4A, OPUS.
“Please add audio file to upload” — the selection was cleared before the button was pressed; add the files again.
Uploaded to the wrong workspace — the Uploading to control on the upload screen sets the target. It follows the workspace switcher, so check it after switching.
Upload seems stuck — the floating tray tracks progress and survives navigation. Large batches over a slow link take a while; the tray confirms completion.
Uploaded but not in the list — that’s processing, not uploading. Watch Running and Queued.

A source stopped scanning

Look at Last Scan On Manage › Integrations. Not Scanned means it has never run; a stale timestamp means it stopped.
Check the schedule A frequency of None only ingests on demand. Set 4, 8, 12 or 24 hours, or use Run Now.
Press Run Now An immediate scan is the fastest diagnostic — it either works or fails now rather than in four hours.
Re-authorise the source OAuth grants for Google Drive, Dropbox and Box can be revoked provider-side or expire. Reconnect from the source's card.
For S3, check the bucket, not the schedule S3 is push-based and shows Not Applicable under Action. Verify the event notification still points at the SNS topic and the bucket policy is intact — see Amazon S3.
For GCP and Wasabi, check the credentials Rotated keys and deleted service accounts are the usual culprits. Re-enter them and run a scan.

Calls on the wrong agent

Calls attribute to agents by name. When an agent’s calls show as blank, unknown, or under a near-duplicate spelling:

Compare the exact spelling in All Agents against what your recording system writes.
Add missing agents to the roster — ideally with your Org Agent ID, which keeps the mapping stable back to your WFM.
Check the agent is assigned to the workspace the calls land in.
Fix the roster before a bulk import; already-processed calls keep the attribution they were given.

Missing screens & Access

Symptom	Cause	Fix
A sidebar item a colleague has is missing for you	Permission gating hides what your account can’t use.	An admin adjusts your role, workspaces or permission set.
Access Denied on an action	The specific permission for that action isn’t held.	Ask an admin to perform it, or to grant the permission.
An agent lands on Agents Performance, not the dashboard	They have Performance Access only.	Grant Calls Access too if they should see their calls — see access levels.
An agent sees nothing	No Access, the default.	Grant access from All Agents, individually or in bulk.
A workspace is missing from the switcher	The user isn’t assigned to it.	Add them under Users assigned.
Invite link says invalid	Expired or already used.	Resend the invite from All Users.

The numbers look wrong

KPI rings show no goal No targets are set on the workspace. Set them under Goals & targets.
Every call's duration is red The average-duration goal is set too low. It's a target, not a limit — set it slightly above your median handling time.
Scores changed without new calls Someone changed the opening / handling / closing weightage. It alters how existing scores combine, which is why it should be set once per review period.
Silence metrics look extreme Check the dead-air threshold (3–10 seconds). At 3 seconds, normal pauses register as dead air.
Analytics chart is empty No data in its relative window, or the selected agents had no calls. Widen the window or reselect agents.
Fewer calls processed than sent Check processing limits — minimum and maximum duration and the percentage-to-process sampling all exclude calls by design.

Maintenance & Error screens

Maintenance page — a planned window is in effect. Processing continues; reload afterwards.
Not Found — a mistyped or stale URL. Navigate from the sidebar.
Access Denied — you reached a screen your account can’t open.
Invalid Link — an expired invite, verification or reset link. Request a fresh one.
A screen fails to render — reload once. If it persists, note what you were doing and contact support; an error boundary catches the failure rather than blanking the app.

When to contact support

Reach us via the contact form. Include:

the call ID (copy it from the call row) and the workspace name;
the date range you were looking at;
the exact Reason text for a failed call;
what you expected versus what you saw, and whether it affects one call or a whole batch.

That set of details is almost always enough for us to answer on the first reply rather than the third.