Troubleshooting

Check three knobs. (1) Time window: if your contacts' last contact is older than the configured window (default 5 years), they are not analyzed; the scan stats show how many contacts sit outside. Raise the window under Settings → Database scan. (2) Signal keywords: overly narrow or unusual keywords skew prioritization; the default list is a good start. (3) Excluded contact types: whatever is excluded under Settings → Qualification is also skipped by the scan. One more thing: the scan reads activities. Without an imported activity history there is little to find (check the Synchronization tab).

The DB-Leads add-on includes a fixed quota of full scans per month; once it is used up, the next full scan starts next month. The regular (incremental) database scan is unaffected: it still analyzes everything new, any time.

Most likely an optional AI limit per run is set under Settings → Database scan, which is not an error. The next scan continues exactly there; the freshness check skips what was already analyzed. Without a limit (the default) one run works through the entire database.

Most likely the time window was reduced. Only findings whose lead has a last contact inside the window are shown. Nothing is deleted: widen the window and the finding reappears. Alternatively it was snoozed.

You loaded sample data in the setup assistant. A notice above every view says so, and the assistant removes them with one click. Once you connect onOffice and start a sync they disappear automatically, so they never sit among your real contacts.

That is exactly the case: onOffice is not set up yet. Connect it under Settings → Integrations; then the same button imports your real addresses. This used to load sample data silently, which looked like a successful import of your own database.

Large imports continue in batches on the background worker. If a run is interrupted, the next sync detects it as stale via the heartbeat and finishes it. Re-syncing is always safe (idempotent, no duplicates).

In the Synchronization tab you'll see the failed records with their external id and reason. The most common cause: a missing or duplicate customer number (KdNr).

Individually triggered briefings are billed against your credit balance (or directly at your provider with BYOK). The database scan's analyses are included in the add-on, as are all CRUD, board, and sync operations and the scan prefilter.

Your credit balance is used up. Top up under Settings → Credits, or use the database scan: its analyses are included in the add-on and handle the bulk. On the Max plan and up you can alternatively switch to BYOK under Settings → REOS AI.

Without a provider key, REOS AI uses a deterministic fallback briefing, so the tool works even without a key. The lead-scoped chat in the customer view, however, requires a configured provider.

A sync does not start a scan. Set the auto-scan under Settings, Database scan to daily: every automatic run syncs first and then rates only what changed. Or start the scan by hand under Findings, it syncs first as well. The analyses are included in the add-on.

That's intentional: the number is the real total of the stage; only the newest cards load first. Use "load more" to fetch more per column. That keeps the board fast even across thousands of leads.

Bulk delete and archive require the db-leads.delete permission, assign requires db-leads.assign. Without the permission the UI hides the action, and the backend rejects it as well.

Starting a scan requires the scan permission (db-leads.scan), accepting a finding requires the qualify permission (db-leads.qualify). Ask your workspace admin.

Three possible reasons. First: the contact has no briefing from a run with the current analysis yet. Trigger Re-qualify on the contact. Second: the sentence does not evidence ownership. A mentioned address alone is not enough, it has to be recognisable that the property belongs to this person. Third: the email bodies are missing from the history, so the analysis has nothing to read. See onOffice import.

Edit it via the pencil icon on the card or remove it via the bin. Both need the db-leads.update permission. A correction turns the entry into your own: from then on no scan overwrites it.

That happens when neither entry carries a street, for example “flat in Augsburg” and “3-room flat in Augsburg”. Without a street two entries cannot safely be recognised as the same property, and merging two different properties would be the worse mistake. Add the address to one, or remove the other.

It needs the Edit contacts permission (db-leads.update), because every outcome writes something. If the view is filtered empty, the button is additionally disabled: no visible contacts, no stack.

The stack lives in the browser. After a reload it is rebuilt from the visible order, and if the data changed in between it can differ. Outcomes already recorded are unaffected.

Moving on only happens once the entry is really written. If you stay on the contact, the error message names the cause, usually an interrupted onOffice connection. That is deliberate: otherwise you would be on the next contact while the previous one had nothing in its history.

They are preserved, including findings and scan history. Actually deleting the data is a separate, explicit step. Nothing disappears unnoticed.

No. The AI presents evidence-backed findings and assessments; the final decision is always yours (AGG-compliant, no discrimination scoring). Signals come exclusively from statements in the activities, never from personal attributes.