Profile Google Sheets Columns Before Accepting Cleanup Suggestions

Use a column profile to decide what needs investigation, not which records to delete. A large value may be correct, and repeated values in one field do not prove that whole records are duplicates.

Google’s desktop help describes Data → Column Stats for distributions, value frequencies and column summaries. Data → Data cleanup → Cleanup suggestions opens potential cleanup actions; highlighted cells help locate what a suggestion concerns. The sidebar arrows move between columns. Column Stats and Cleanup Suggestions

These ordinary Smart Cleanup tools were announced for personal Google accounts as well as business accounts. This exercise does not require a paid AI feature or external service. Google’s availability announcement

Build an independent control before opening the sidebar

Use a disposable native Google Sheet on a computer. Enter this synthetic data in A1:C7, with no other records on the tab:

Record ID	Category	Units
1001	Cable	2
1002	Cable	2
1003	Adapter	3
1004	Adapter	3
1005	Stand	4
1006	Bulk	20

Keep A1:C7 intact while inspecting it. Put Check in E1 and Result in F1. Label E2:E6 as Count, Sum, Minimum, Maximum and Median, then enter the following formulas in F2:F6 in that order. These controls stay outside the three profiled data columns:

Formula	Expected result
=COUNT(C2:C7)	6
=SUM(C2:C7)	34
=MIN(C2:C7)	2
=MAX(C2:C7)	20
=MEDIAN(C2:C7)	3

The average is 34 ÷ 6, approximately 5.6667, not the median 3. The category frequencies are Cable 2, Adapter 2, Stand 1 and Bulk 1. These are independently calculated controls for the exact six records, not promises that every sidebar displays every statistic with these labels or precision.

Inspect the current column and scope

Select a cell in the Units column and open Data → Column Stats. Confirm which column the sidebar describes. Compare its distribution with the six known values, and inspect any displayed counts or summaries against the bounded controls above. Use the sidebar arrows to inspect Category, where repeated labels are expected.

If a displayed statistic differs, first inspect its scope, treatment of the header and any extra cells in that column. Do not alter data simply to make a sidebar number agree. The formulas deliberately reference C2:C7 so their inclusion rule is explicit. A sidebar profile and a range formula may be examining different inputs.

If a frequency display shows two Cable entries, verify IDs 1001 and 1002 before calling them duplicates. In this fixture they are two separate records. Likewise, a count of two values equal to 3 does not authorize deleting either Adapter row.

Review suggestions without manufacturing an error

Open Data → Data cleanup → Cleanup suggestions. If suggestions appear, inspect the indicated cells and the proposed action before applying anything. No particular suggestion is guaranteed for this small fixture. An empty suggestion list is not proof that every value is correct.

For this exercise, the fictional source record explicitly confirms 20 units for ID 1006. Retain 20 even if its size attracts attention: it represents a legitimate bulk quantity. The total 34 includes that record. Reducing it to 2 merely because 2 is common would create an unsupported total 16.

Do not accept a whole-row removal based only on repeated Category or Units values. Decide which fields define a record first. For a separate worked cleanup job with an explicit two-column key, use split, trim and remove duplicates.

Test a correction only when a source justifies it

Now make a separate working copy of the six-record fixture. In this second scenario, a corrected source says ID 1006 should contain 5 units, not 20. Record that reason before changing C7 to 5.

Recalculate the controls. Expected results are count 6, sum 19, minimum 2, maximum 5 and median 3. The average becomes 19 ÷ 6, approximately 3.1667. Record IDs and category frequencies must remain unchanged. This verifies a value correction rather than accidental row deletion or category editing.

Return to the original copy and confirm it still contains 20 and total 34. Preserve both labelled scenarios if retaining the exercise. In real work, keep the authoritative source and a change log; do not use the tutorial’s fictional correction to justify changing an unexplained outlier.

Troubleshooting and verification limits

If the numerical count is below 6, inspect whether a units entry is text or blank before changing formats. A number-like display does not establish numeric storage. Check COUNT and SUM for input behavior. If you only want a different visible subset, use filter views rather than deleting records.

This guide covers the decision and verification workflow, not an exhaustive catalogue of suggestion heuristics. Google documentation was checked October 21, 2026; the controls, frequencies and changed scenario were independently calculated. No signed-in Google Sheets session, sidebar output, suggestion acceptance, saved-file reopening or screenshot was performed. Mobile behavior and paid AI analysis are outside scope.