Blog. Audit evidence.
A sensitive-data assessment needs a coverage record
Co-founder and CEO · November 6, 2026 · 3 min read

On this page
A sensitive-data assessment is the size of what it read, and a summary slide rarely says how big that was. Before anyone repeats a result, ask which sources were included, when each was read and which sit outside the conclusion. A coverage record answers all three on one page, next to the findings.
The slide speaks for the estate, and the read covered part of it
A software company holds customer data in five places: a CRM database, a billing database, a shared drive where exports land, a data warehouse and a support platform. An assessment runs. The warehouse connection fails on the day, and the support platform is held back because its owner wants a contract review first.
The summary slide says no unprotected card numbers were found. What a reviewer can conclude is narrower: three named sources had none on the day they were read. The warehouse is fed by both databases, so it is the copy that matters most, and nobody read it.
| Source | Owner | State | Read on | Limitation | Next action |
|---|---|---|---|---|---|
| CRM database | Head of sales operations | Assessed | November 3, 2026 | None recorded | Re-read after the next schema change |
| Billing database | Finance systems lead | Assessed | November 3, 2026 | Archive schema outside the scope | Decide whether the archive joins the scope |
| Shared drive, exports | IT operations manager | Assessed | November 3, 2026 | Two password-protected archives not opened | Owner supplies the passwords or confirms the contents |
| Data warehouse | Head of data platform | Connection failed | Not read | Copy of CRM and billing data | Fix the account grant and retry by November 13, 2026 |
| Support platform | Customer support director | Not approved | Not read | Owner requires a contract review first | Decision due November 20, 2026, with the approver named |
Nothing on the slide is false. The slide is larger than the evidence behind it, and the table is the smaller, truer statement. It also says what to do next.
The inventory and the assessed scope are two lists
The first list is the estate inventory: every system that holds, or may hold, sensitive data. NIST’s Cybersecurity Framework 2.0 expects organizations to maintain “inventories of data and corresponding metadata for designated data types” (ID.AM-07).
The second list is the assessed scope, the sources an assessment actually read. NIST’s draft guide to data classification practices, SP 1800-39, describes a tool that “is provided access to where the data is located.” A tool reads the locations it is given, and the rest sits outside the result.
The same draft says organizations “may not know the locations of all their data.” That is why the two lists differ, and why the difference has to be written down.
Five states settle every source on the inventory
- Assessed means the source was read on a stated date, and its findings link back to it.
- Connection failed means it was in scope and attempted, and the record holds the error and the owner of the fix.
- Not approved means someone decided to leave it out, and the record names who decided and why.
- Stale means it was read before a change that matters, such as a new export job.
- Unknown means nobody yet knows whether it holds sensitive data.
One rule outranks the labels. A source nobody read contributes no zero to a total. “No findings” and “not assessed” look alike in a sum and mean opposite things.
Formal assessments write the coverage into the plan first
NIST SP 800-53A gives each assessment method a coverage attribute, which “addresses the scope or breadth” of the work. The depth and coverage of an assessment are “assigned by the organization and specified within the assessment plan.” Scope is decided before the work starts.
A sensitive-data assessment can follow the same order: build the table above before the scan and complete it after. If a row needs a paragraph to explain its limitation, that limitation is a finding in its own right.
The reader works through limitations before findings
Whoever relies on the result goes through the record in this order.
- Read the limitation column before the findings.
- Check that the inventory is current. Has a system or an export been added since the last assessment?
- For each failed or excluded source, accept the gap with a reason or schedule the next attempt.
- Sign and date the record.
When the result moves into a questionnaire answer, a board paper or an audit file, the record moves with it. Otherwise a narrow result grows into a broad statement along the way.
Each scan records what it covered and lists what it couldn’t read, with the reason. Unread tables stay unclassified, and unknowns show as unknown, never zero. Capabilities vary by edition.
The audit evidence page shows the record. The source coverage register guide lays out the columns for a register you keep yourself.
The gaps define the next assessment
Retry the failed connection, settle the approval and re-read the stale source. That is a short list a team can finish, and it leaves no gap unnamed.
Start with last quarter’s summary. List every source it does not name, and give each one a state from the list above. The result is the first draft of your record. The glossary entry on scan coverage defines the terms.
Sources
- NIST: The NIST Cybersecurity Framework (CSF) 2.0, CSWP 29, subcategory ID.AM-07 (February 26, 2024; opened November 9, 2026)
- NIST: SP 1800-39, Data Classification Practices, initial public draft, section 2.2 and the workflow description (February 2026; opened November 9, 2026)
- NIST: SP 800-53A Rev. 5, Assessing Security and Privacy Controls, section 2.4, the coverage attribute (January 2022; opened November 9, 2026)
Start with a free assessment of one database you choose.
Get a free assessmentRelated resources
Protect your brand. Start with your data.
Book a demo, or start with a free assessment of one database.


