What is tenant isolation?
- Tenant isolation
- Tenant isolation is the separation of one customer's data from another's inside software that serves many customers from the same system.
What tenant isolation means in practice
Almost every business tool you use is shared. The alternative is a separate installation per customer, which costs more than most products are sold for.
Separation comes in degrees. A separate database per customer is strong and expensive; a shared table with a customer column is cheap and depends entirely on the code being right.
Failures are rare and memorable. A missing filter in one query shows someone else's records, and the shape of that bug is always the same.
The strength of the isolation also decides what happens when one customer's data has to be deleted or exported.
Backups inherit the same model. Data separated cleanly in the live system can end up interleaved in a nightly archive, which is where deletion requests usually get complicated.
What people get wrong
One missing line in one query
Say the scheduling product you use keeps every customer's appointments in one big table, with a column saying which business each row belongs to. Every query is supposed to end with a filter: only rows where the business matches the one that's logged in. A developer adds a "search by phone number" feature on a Thursday and forgets that line.
On Friday a salon owner in Tucson searches for the last four digits of a client's number, 0142. She sees her client. Below that sits a row from a chiropractor in Maine, whose patient shares those four digits. Nothing was hacked and nobody's password leaked. That software did exactly what the code said. Stronger designs put the rule in the database itself, which PostgreSQL calls row-level security. A forgetful query then returns nothing at all instead of everybody's rows, including yours.
Questions that get a real answer out of a vendor
You'll always get a yes to "Is our data isolated?", so ask these instead. Is separation enforced by the database or by application code? Do you run an automated test that signs in as customer A and tries to read customer B's records? Does it run on every release? When a support engineer opens our account, can they see ours alone, and is that logged?
Follow with the practical ones. If we leave, can you delete our data without touching anyone else's, and how long until the backups holding it expire? Can you export everything that's ours in one file?
Listen for specifics more than for a particular architecture. A vendor with a shared table, database-level rules and a cross-tenant test is in better shape than one with separate databases and no tests. All of this matters most when what you store would embarrass your customers if a stranger read it, and call transcripts qualify.
How GreetKeeper handles it
Your call records belong to your account, and we are not going to dress that up as an architecture claim with no audit behind it.
GreetKeeper holds no certifications, which is the relevant fact for any formal security review.
Retention and access are the settings that reduce your exposure regardless of how any vendor's separation is built.
Tenant isolation questions
Is shared software less safe?
Not inherently. Shared systems get more scrutiny and more patching; separate ones get forgotten. The failure modes differ rather than one being safer.
What should I ask a vendor?
How separation is enforced and what testing covers it. A specific answer tells you they have thought about the bug that matters.
Does it affect deletion requests?
Yes. Strongly separated data is straightforward to remove; a shared store needs careful queries and backups that expire.
Related terms
Hear it take one of your calls
Two minutes, your own scenario, no card.