How to Find the Personal Data Hiding Across Your Systems
A practical method for discovering personal data across databases, files and SaaS tools, and turning the results into a data map your team can use.
Ask most companies where their personal data lives and you will get a confident answer: the CRM, the HR system, the main production database. Then someone mentions the analytics warehouse, the support tool, the old marketing database nobody switched off, and the spreadsheet finance exports every month.
Personal data spreads. Finding it is the first step in protecting it, and it is usually harder than expected.
Why personal data ends up everywhere
- Copies are easy to make. Exports, backups, test environments and data pipelines all create new copies.
- Fields get reused. A free-text “notes” column can end up holding phone numbers, addresses or identity numbers.
- Tools multiply. Every new SaaS product that integrates with your systems may store its own copy.
- People leave. The person who built a database may be the only one who knew what was in it.
Step 1: List every place data could be
Start broad. Include:
- Production databases and their replicas
- Data warehouses and lakes
- Backups and archives
- Test and staging environments
- SaaS tools: CRM, support, HR, marketing, finance
- File storage: shared drives, object storage, email attachments
- Logs, which often capture more than intended
Do not filter yet. The goal is a complete list of candidates.
Step 2: Talk to the people who run them
For each system, ask the owner what it is for, what data goes in, who uses it and whether it is still needed. These conversations are often where you discover systems that should simply be switched off.
Step 3: Scan, do not just ask
Interviews tell you what people believe is in a system. Scanning tells you what is actually there. A good scan:
- Looks at real values, not just column names. A column called
ref_codemight hold identity numbers. - Understands local formats. Indian identifiers such as PAN, Aadhaar, GSTIN, IFSC codes and mobile numbers have recognisable patterns, and some have checksums that cut false positives.
- Samples rather than copies. It should read enough rows to be confident without moving the data anywhere.
- Explains each finding. Why does it think this is personal data? A pattern, a checksum, a column name, nearby fields?
Step 4: Keep discovery inside your perimeter
A discovery tool, by definition, reads your most sensitive data. That makes how it runs as important as what it finds. Look for:
- Deployment inside your own environment
- Read-only credentials
- Findings that contain masked samples, not raw values
- No outbound calls to outside services
Step 5: Turn findings into a map
Raw findings are not a map. For each system, record:
| Question | Example |
|---|---|
| What personal data is here? | Names, mobile numbers, PAN |
| Whose data is it? | Customers |
| Why do we hold it? | Billing and tax invoices |
| Who can access it? | Finance team, two engineers |
| How long do we keep it? | The period agreed with finance and legal |
Some of these answers need human judgment, especially purpose and retention, and some may need input from your legal team. A scanner can draft the factual parts and leave the judgment calls to you.
Step 6: Keep it current
A data map starts going stale the day it is finished. Re-scan on a schedule, scan new systems before they go live, and make “update the data map” part of launching any new product or integration.
What to do with the map
Once you know where personal data lives, you can:
- Answer requests from individuals quickly and completely
- Delete data you no longer need
- Protect sensitive fields with tokenisation or encryption (see tokenisation vs encryption vs masking)
- Give your legal advisers an accurate picture to work from
For how a data map fits into wider readiness work, see our DPDP compliance checklist.
Arbor Lens finds the personal data in your systems from inside your own perimeter, read-only.
Explore Arbor LensQuestions people ask
Why is a data map so important?
Almost every privacy task depends on knowing where personal data is: answering requests, setting retention, securing sensitive fields and assessing vendors. Without a map, each of these becomes guesswork.
Can we do data discovery manually?
For a small number of systems, yes. Interviews and spreadsheets work. As the number of databases and tools grows, automated scanning becomes the only practical way to keep the map accurate.
Is it risky to scan systems for personal data?
A scanner reads sensitive data by design, so choose one that runs inside your own environment, uses read-only access and reports findings without copying values out.