Validate a Google Ads CSV Upload

Bid columns expect bare numerics. The classifier accepts $2.50 and 1,000 but not a non-dollar symbol or text like auto, so those cells are flagged against the inferred number profile.

Header:
Campaign,Ad group,Keyword,Max CPC

Rows:
Brand,Core,running shoes,2.50        OK
Brand,Core,trail shoes,$3.00         OK ($ stripped)
Brand,Core,sneakers,auto             type_mismatch (text)
Brand,Core,boots,£1.80               type_mismatch (non-$ symbol)

Validator output (2 issues):
Row 4: Expected number based on the column profile, but found "auto".
Row 5: Expected number based on the column profile, but found "£1.80".

Fix: replace text bids with numeric values, and strip non-$ symbols with
     csv-find-replace regex [£$€¥] -> (empty) before re-importing.

Headlines and descriptions often contain commas. If the cell isn't quoted, the row gains a column and every field after it shifts. The validator reports a row_width error.

Header (4 cols):
Campaign,Ad group,Headline 1,Final URL

Row (5 cols — comma inside the headline):
Brand,Core,Fast, free shipping,https://example.com
                ^ this comma split the headline cell

Validator output:
Row 2, Col 1  (row width)
  type: row_width  severity: error
  detail: "Row has 5 columns; header has 4."

Fix: re-export from Google Ads Editor (it quotes ad copy correctly), or
     wrap the Headline column in double quotes before saving.

A duplicated edit row referencing the same entity ID can create conflicting changes. Because the column name contains the id token, the repeat is flagged as a unique-key collision.

Header:
Campaign ID,Campaign,Budget

Rows:
101,Brand,50.00
102,Generic,30.00
101,Brand,75.00          <- same Campaign ID, conflicting budget edit

Validator output:
Row 4, Col 1  Campaign ID
  type: duplicate_key  severity: warning
  detail: 'Value "101" also appears on row 2.
           This column looks like a unique key.'

Fix: decide which edit wins and de-dupe with csv-deduplicator on the
     Campaign ID column, or inspect first with csv-duplicate-finder.

International campaigns carry non-Latin headlines. A CP1252 save read back as UTF-8 turns those into the replacement character, flagged as encoding errors that would also trip Google's character validation.

Header:
Campaign,Headline 1

Rows (CP1252 decoded as UTF-8):
DE,Schnelle Lieferung f\uFFFDr alle    encoding error (was 'für')
FR,Promotions d'\uFFFDt\uFFFD          encoding error (was 'été')

Validator output:
Row 2, Col 2 (Headline 1): replacement/control character — encoding problem.
Row 3, Col 2 (Headline 1): same.

Fix: re-save as 'CSV UTF-8 (Comma delimited)', or re-export from Ads Editor
     in UTF-8. csv-cleaner can normalise the text if you only have the
     corrupted file.

If a bid column is at least 85% populated and a row's bid is blank, the validator flags it as an empty_cell warning. Whether that's a problem depends on whether the bid is meant to be inherited — verify before bulk-filling.

Header:
Keyword,Max CPC

Rows (bid column mostly populated):
running shoes,2.50
trail shoes,3.00
sneakers,                 <- blank bid in a mostly-filled column
boots,1.80

Validator output:
Row 4, Col 2  Max CPC
  type: empty_cell  severity: warning
  detail: "Cell is empty in a mostly populated column."

Fix: if the bid should be set, fill it; if it's intentionally inherited
     from the ad group, the warning is informational — verify, don't
     bulk-fill blindly.

The number classifier strips only a leading $ (with an optional sign) before testing, so $2.50 passes but £1.80, €1.50, and inline codes are flagged as type_mismatch in a numeric bid column. Google Ads bulk sheets expect bid values in the account currency as bare decimals, so strip the symbol in source with csv-find-replace before upload. The flag is correct — a symboled bid would not import as the number you intend.

Headlines, descriptions, and URL parameter strings frequently contain commas. In a comma-delimited file an unquoted comma adds a column, so the validator reports row_width with the column-count delta. Re-exporting from Google Ads Editor is the cleanest fix because it quotes text fields correctly; if you're editing by hand, wrap ad-copy and URL columns in double quotes before saving. Don't strip the comma from the copy itself unless you actually want it gone.

Google Ads bid columns are often legitimately blank when a bid is inherited from a higher level. The empty_cell warning only fires when a bid column is at least 85% populated, so a column that's blank for many rows (over 15% empty) stays quiet, while a mostly-filled column with a few blanks flags them. Treat those warnings as informational — verify each blank is an intentional inheritance, not a forgotten value, before bulk-filling.

This is the most important honesty caveat for Google Ads files: the validator profiles only the columns present and has no knowledge of Google's required-column schema. If your bulk sheet is missing a column Google requires (a bid column, a Campaign column for a new entity), the validator won't flag the absence — it will validate whatever columns exist. Compare your header row against a fresh Google Ads Editor export of the same entity type to confirm you have the right columns before upload.

When non-Latin or accented ad copy is corrupted by a CP1252 save, the validator flags \uFFFD replacement characters as encoding errors. The same corruption trips Google's own character validation at upload, so clearing these errors here pre-empts a rejection. Fix by re-saving as CSV UTF-8 or re-exporting from Editor in UTF-8; the csv-cleaner tool can normalise text if you only have the corrupted file.

The classifier removes commas before the numeric test, so 1,000 reads as 1000 and a budget column full of thousands-separated figures is correctly inferred as number with no type_mismatch. Be aware this means an EU comma-decimal like 1,50 also has its comma removed, becoming 150 — a different value — and would still parse as a number, so it won't be flagged. For non-US locales, double-check that decimals use a point, because a silently-misread budget is worse than a flagged one.

Free tier caps at 2 MB and 500 rows. A whole-account Google Ads Editor export with thousands of keyword and ad rows will be blocked before validation, with the limit message shown. Use Pro (100 MB / 100,000 rows) or split the export with csv-row-splitter — for example by campaign — and validate each segment. Splitting also makes it easier to isolate which part of a large upload has a structural problem.

The validator is structural: it does not enforce Google's ad-copy character limits, advertising policies, or landing-page validation, and it does not fetch or parse Final URL values. Those are all handled by Google at upload. Use the validator to guarantee the file is well-shaped and the bid/budget columns are clean, then let Google Ads do the policy and content validation — clearing this report reduces, but doesn't eliminate, upload-time rejections.

The reported issue list is capped at 250 to keep the browser responsive; the Issues tile shows that capped count. A systemic problem — every bid carrying a symbol, or a mis-detected delimiter across a large sheet — can exceed 250 issues. Fix the single root cause and re-validate; the count collapses to a handful. The downloaded JSON report contains the same capped list, so for very large counts the workflow is fix-and-re-run, not scrolling through 250 entries.

The file is read with the browser's own File.text() and parsed by PapaParse client-side; keyword lists, bids, budgets, and landing-page URLs are never transmitted. The tool is marked 100% client-side, and only an anonymous run counter is recorded for signed-in dashboard stats. Since a full bulk sheet effectively encodes your account strategy, on-device processing keeps that competitive data off any server, including JAD's.