Find the Snowflake Tables Nobody's Using, and Safely Clean Them Up
Orphan Tables Insight
Every Snowflake environment accumulates tables nobody remembers creating: a one-off export, an old pipeline’s leftover output, a table someone made for a test that never got cleaned up. They just sit there, taking up storage and cluttering the catalog, and nobody wants to be the one who deletes a table that turns out to matter. Orphan Tables finds the ones that are genuinely safe to drop, backs that up with real evidence, and hands you the exact command to do it.
Seemore Orphan Tables Insight in Action
Confirmed dead by two independent checks, not a guess
Deleting the wrong table is a real risk, so “probably unused” isn’t good enough.
- A table only gets flagged if it has no connection to any tracked pipeline (dbt, Airflow, Fivetran, and others), and it has had zero actual queries run against it
- Both conditions have to be true at once, so a table that’s still wired into a pipeline, even a quiet one, won’t get flagged just because nobody’s queried it directly
- That combination is what makes an orphan table genuinely safe to consider dropping, not just a table that looks idle
Every finding comes with the exact fix, ready to run
Once you know a table is dead weight, you shouldn’t have to go write the cleanup yourself.
- Each finding shows the table’s row count and storage size, plus the actual annual dollar savings from freeing that storage
- You get the exact
DROP TABLEstatement ready to copy and run, no need to write it yourself or double-check the syntax - Mark it done once you’ve cleaned it up, and if that table ever becomes active again, the insight reopens automatically instead of staying silently closed
Built to avoid false alarms
A cleanup tool that cries wolf gets ignored, so this one is built to be trusted.
- The usage check looks back over a rolling window (21 days by default), not a single snapshot, so a table that’s simply quiet for a day doesn’t get flagged
- You can exclude specific service accounts or bots from counting as real usage, so a table only kept alive by an automated ping still gets surfaced instead of hiding behind it
- Every finding is cross-checked against other waste-reduction insights so the same table doesn’t show up flagged two different ways
Fits into how your team actually works
Finding a dead table is only useful if someone actually acts on it.
- Assign a finding to whoever should own the cleanup, and track it through Open, in review, or Done, right alongside a comment thread for context
- Anyone on the team can see the findings and their status; changing account-wide detection settings is limited to editors and above
- Nothing gets lost in a one-time report, it stays visible and tracked until it’s actually resolved
Frequently asked questions
What is an "orphan table"?
A table or view in your Snowflake environment that isn’t connected to any data pipeline and hasn’t had a single query run against it in the recent period Seemore checks. In plain terms: nothing is feeding it, and nothing is reading from it.
How does Seemore know it's actually safe to delete, and not something important?
It only flags a table when two things are true at once: it’s not part of any tracked pipeline (like dbt or Airflow), and it’s had zero real query activity. A table connected to even one active pipeline won’t be flagged, even if nobody’s querying it directly.
Will this flag a table that's only used occasionally, like once a quarter?
It looks at a rolling window of recent activity (21 days by default), not a single day. A quieter usage pattern than that is the kind of thing worth reviewing anyway, and the finding always shows you the evidence before you act.
What if a table is only touched by an automated bot or service account?
You can choose to exclude specific bots or service accounts from counting as “real” usage. That way, a table kept artificially alive by an automated ping still gets surfaced instead of hiding behind it.
Do I have to figure out how to delete the table myself?
No. Each finding comes with the exact command ready to copy and run, along with the table’s size and the annual cost you’d save by removing it.
What happens after I delete a table, or decide to keep it?
You mark the finding as Done once it’s handled. If that table somehow becomes active again later, the finding reopens automatically rather than staying incorrectly closed.
Can my whole team work through these together?
Yes. Each finding can be assigned to someone, tracked through a status (open, in progress, done), and discussed in comments, so cleanup is a tracked team task, not a one-time report that gets forgotten.
Does everyone on the team have the same level of access?
Anyone can view findings and mark them done. Changing account-wide detection settings (like which bots to exclude) requires an editor role or higher.
