Evidence brief 01Notice export dated July 30, 2026 · RLLV submissions through July 29 · complete months through June

June 2024 was extraordinary. The larger shift in DIR’s dated notice records had already begun.

New state records show a durable rise in dated notice records, a dramatic three-day concentration around reform week, and a steady standardization of allegation checklists. They do not show that violations, lawsuits, or recoveries increased. ^{Source 1 · Source 3 · Source 4 · Source 6 · Source 8 · Source 9 · Source 14}

The reading rule

Notice≠Lawsuit≠Proven violation

DIR lists claim notices and court complaints as separate submissions. The notice and RLLV fields analyzed here record allegations, not adjudications. ^{Source 1}

Independent analysisPublished August 2, 2026

Read the audit protocol

01 / Volume

The higher-volume period was evident in 2023—before reform week.

Distinct dated notice identifiers rose from 5,811 in 2022 to 7,900 in 2023, a 36.0% increase. Eleven of twelve months exceeded their 2022 counterparts. The series rose again to 9,437 in 2024, then eased only 1.5% to 9,295 in 2025. ^{Source 3}

Annual notice identifiers

The step-up begins before the 2024 reform

Distinct dated LWDA identifiers

2017

2018

2019

2020

2021

5,811

2022

7,900

2023

9,437567

2024

9,295

2025

2017: 5,109
2018: 5,713
2019: 6,421
2020: 6,513
2021: 6,498
2022: 5,811
2023: 7,900
2024: 9,437
2025: 9,295

567 · June 19–21 subset of the 2024 total

The 2023 total was 36.0% above 2022; 2025 remained 17.7% above 2023. ^{Source 3}

SourceExh. A, PAGA Notices - all time, B3:F3 (“LWDA Case Number”; “PAGA Notice Filing Date”), B15076:F51445 and B51446:F78203.

LimitCounts are distinct dated identifiers—not lawsuits, proven violations, or unique employers.

Period comparison500.9 → 739.8

Average distinct identifiers per month: 2017–22 versus 2023–25. That is a 47.7% increase. ^{Source 3}

Extreme exclusion test+44.9%

Later-period monthly average after removing the entire month of June 2024 and its 1,237 records. ^{Source 3}

Latest rolling year9,597

July 2025–June 2026, up 7.1% from the prior rolling year and the highest complete-month rolling total since 2017. ^{Source 4}

What survives scrutiny
DIR’s dated notice records entered a higher-volume period by 2023 and had not returned to the earlier level by June 2026. The files do not identify why. ^{Source 3 · Source 4}

02 / Reform week

A DIR-recorded three-day concentration was extraordinary—and too narrow to account arithmetically for the broader rise.

State officials announced a PAGA reform agreement on June 18, 2024. The enacted legislation was approved July 1, but its amendments generally apply to civil actions brought on or after June 19, with an exception tied to notices filed before June 19. ^{Source 5}

June 18Reform agreement announced

June 19General application dateNotice-based exception applies

July 1Legislation approved

Matched three-day periods

A DIR-recorded three-day concentration

5.25× average of matched 2023 and 2025 periods

2023matched Wed–Fri

105

2024June 19–21

567

2025matched Wed–Fri

111

Seven days before206

Beginning June 19681

Following seven days161

The 567 records equal 45.8% of all June 2024 records—but the concentration dissipates quickly. ^{Source 6}

SourceExh. A, PAGA Notices - all time, B3:K3 (“LWDA Case Number”; “PAGA Notice Filing Date”), B63823:F64389, B54832:F54936, and B73211:F73321.

LimitThe export contains no filer-motive, processing-lag, or batch-handling field; the chart does not establish causation.

Equal before/after windows

The difference rapidly converges

30 days740→1,135

+53.4%

60 days1,522→1,760

+15.6%

90 days2,361→2,453

+3.9%

180 days4,572→4,627

+1.2%

Post-window count versus the equal pre-window count immediately before June 19. ^{Source 6}

Not a causal counterfactual—an arithmetic stress test

9,437− 567 event-window records8,870

Even after deleting every June 19–21 record, the adjusted 2024 count remains 12.3% above 2023. The removed records represent 36.9% of the observed 2023-to-2024 increase. ^{Source 7}

What survives scrutiny
The concentration was extraordinary and short-lived. The export cannot distinguish strategic filing from administrative timing, processing lag, or batch handling—and cannot prove causation. ^{Source 6 · Source 7}

03 / Allegations

Allegation packages became broader—and increasingly standardized.

Because 3,893 LWDA identifiers have multiple RLLV rows, the primary comparison uses the latest create-date row per identifier, with the later worksheet row breaking date ties. It then counts 29 Boolean allegation fields and groups records by notice year. ^{Source 8}

Latest row per LWDA identifier · grouped by notice year

Seven recurring fields were co-selected in 62.6% of 2025 records

Share of latest-row cohort

2017: 14.5%674 / 4,654
2018: 19.7%1,058 / 5,376
2019: 28.7%1,827 / 6,357
2020: 35.6%2,313 / 6,495
2021: 45.5%2,945 / 6,472
2022: 48.9%2,818 / 5,768
2023: 53.2%4,193 / 7,886
2024: 60.5%5,707 / 9,438
2025: 62.6%5,812 / 9,287

2017: 14.5%
2018: 19.7%
2019: 28.7%
2020: 35.6%
2021: 45.5%
2022: 48.9%
2023: 53.2%
2024: 60.5%
2025: 62.6%

The combination is analyst-defined: it consists of the seven fields that each exceeded an 80% selection rate in the 2025 latest-row cohort. ^{Source 8 · Source 9}

SourceExh. B, RLLVS, P3:S73097 (“Minimum Wage”; “Overtime”; “Not paid for all hours worked”; “Not Paid Wages Due on Termination”), V3:V73097 (“PMNT Reimbursement of Empl Expenses”), Y3:Y73097 (“Meal and Rest Breaks”), and AE3:AE73097 (“Improper or Incomplete Wage Statements”).

LimitA selected field records an allegation, not a proven violation; later RLLV rows may be partial amendments.

Breadth of selected categories

The middle of the distribution shifts

0–3 selected22.8%5.1%

8–11 selected26.8%60.3%

12+ selected3.5%11.8%

2017 latest-row cohort compared with 2025. At every threshold from one through 28 categories, the 2025 share is higher. ^{Source 8}

The seven-field co-selection

Each field exceeds 80% in 2025

Improper/incomplete wage statements8,584 / 9,28792.4%

Meal & rest breaks8,571 / 9,28792.3%

Overtime8,332 / 9,28789.7%

Unpaid hours8,017 / 9,28786.3%

Termination wages8,012 / 9,28786.3%

Minimum wage7,816 / 9,28784.2%

Expense reimbursement7,710 / 9,28783.0%

Latest-row 2025 cohort, n = 9,287. Selection does not establish merit. ^{Source 9}

The direction survives three alternate counting rules.

Every raw submission rowEver selected per identifierCohort by RLLV create year

The distribution shift and seven-field co-selection rise under each alternative. Later rows may be partial, so true-to-false changes are not labeled “withdrawals.” ^{Source 8 · Source 9}

04 / Representative labels

More records are associated with recurring representative-field labels.

Using exact “Law Firm for PAGA Plaintiff” text after outer trimming, each year’s ten most frequent labels account for 17.9% of nonblank 2019 records and 29.8% in 2025. The year-specific top-25 share rises from 32.8% to 47.4%. ^{Source 11}

Each year’s top 102019

17.9%

202529.8%

Each year’s top 252019

32.8%

202547.4%

44.8%rise in total notice identifiers, 2019→2025

1,129→777outer-trimmed nonblank labels

7.3%largest single exact 2025 label

^{Source 11}

SourceExh. A, PAGA Notices - all time, B3:G3 (“LWDA Case Number”; “PAGA Notice Filing Date”; “Law Firm for PAGA Plaintiff”), B26132:G32574, and B68824:G78203.

LimitThese are year-specific exported text labels—not verified law-firm entities or market shares.

These are exported text-label concentrations—not verified law-firm entities or market shares. The field includes “In pro per,” and spelling, branding, affiliate, and formatting variants can split or combine activity imperfectly. ^{Source 11 · Source 12}

The evidentiary boundary

What these records cannot answer

Several tempting conclusions fail the source audit. They should not appear as public claims.

Not lawsuits

A claim notice and a court complaint are separate submissions. Notice volume cannot be described as lawsuit volume. ^{Source 1}

Not proven violations

A selected RLLV category records an allegation. It does not establish liability, frequency, severity, or merit. ^{Source 1 · Source 8}

No enforcement rate

The NOI and issue files are request-specific. Their BOFE-key linkage is incomplete and many-to-many; the issue export also contains an exact duplicate. They cannot support statewide investigation, citation, adjudication, or collection rates, or statewide dollar totals. ^{Source 14}

Not unique companies

Employer text does not reliably consolidate subsidiaries, DBAs, corporate families, or spelling variants. ^{Source 13}

Sources & methods

Built to be checked.

Every derived statistic identifies its cohort, calculation rule, and workbook range; every displayed ratio identifies its numerator and denominator.

ANotice export

81,783 distinct nonblank IDs; 2,310 excess exact-duplicate rows. ^{Source 2}

BRLLV export

68,708 distinct IDs; latest-row primary rule plus three sensitivity rules. ^{Source 8}

C–DEnforcement exports

Used to establish limitations—not to manufacture rates. ^{Source 14}

Audit protocol

The audit traversed all 4,395,313 positions inside the four used worksheet rectangles, including blanks, and inventoried every OOXML package member. Each workbook contains one visible sheet and no hidden rows or columns, formulas, comments, hyperlinks, external links, macros, drawings, or pivot tables. No result relies on hidden content or a workbook formula. ^{Source 16}

The safest time series begins in 2017. Fourteen earlier notice rows lack filing dates, DIR warns that older records may be destroyed under retention practices, and 2026 is incomplete. This analysis uses complete calendar years through 2025 and complete rolling months through June 2026. ^{Source 2 · Source 15}

Fixed exhibit versions

Exh. A · HQ64881-PAGA Notice filed 2004-2026.xlsx
PAGA Notices - all time · B1: “PAGA Notices Filed 2004-2026 (7/30/26 at 11:25am)”886a7b57e8637f8d613c4c3eb446eb60cc97f4454bbdba5ff13ea8234ba958fd

Exh. B · HQ64881-RLLVs-2016-2026.xlsx
RLLVS · B1: “Report of Labor Law Violation Submission Date (9/6/2016 to 7/29/2026)”bf5e97d1dee9435b121506bcbbfc6b91b28a560a8ea8ca8d7f783f04681d8871

Exh. C · HQ64881-Notice of Investigations.xlsx
NOIs · B2: “PAGA cases with Notice of Intent to Investigate (NOI)-HQ64881”df5b975880dd476b92437404c9bc5473536a26c50ddc7c92aa9a4e5d4678b38b

Exh. D · HQ64881-PAGA Issues+Citation.xlsx
PAGA Issues+Citation · B2:B3: “PAGA Issues+Citation- HQ64881”; “PAGA Notice Filing 2016-2026”c826240fc8214a87e43bf1706dbdf7761735bd3f13744fec14a5846f22ab6c45

^{Source 16}

Full source notes

(Lab. Code, § 2699.3, subds. (a)(1)(A), (c)(1)(A) [“shall commence only after”; “specific provisions ... alleged”; “facts and theories”] [required notice content and sequencing before specified civil actions]; Cal. Dept. of Industrial Relations, Private Attorneys General Act (PAGA)—Filing (Sept. 2025) [as of Aug. 2, 2026] [“New PAGA Claim Notice”; “Court Complaint”] [separate portal submission types].)↥
(Exh. A, PAGA Notices - all time, cells B3:F3 [“LWDA Case Number”; “PAGA Notice Filing Date”], B4:K84096 [84,093 rows; 81,783 distinct nonblank identifier strings; 2,310 excess rows are exact duplicates across substantive fields; 84,079 rows have parseable filing dates and F4:F17 are the 14 blanks], and B84097:D84097 [“Total”; “Count”; 84,093] [static, not formula-driven].)↥
(Exh. A, PAGA Notices - all time, cells B3:F3 [“LWDA Case Number”; “PAGA Notice Filing Date”], B15076:F51445 and B51446:F78203 [calculation: count each distinct nonblank LWDA string once; group parsed filing dates by month and year; 2017–2022 = 36,065/72 = 500.9 monthly; 2023–2025 = 26,632/36 = 739.8, up 47.7%; omit all 1,237 June 2024 records and that month = 25,395/35 = 725.6, up 44.9%; 2022–2025 totals = 5,811, 7,900, 9,437, and 9,295; stated percentage changes follow from those counts].)↥
(Exh. A, PAGA Notices - all time, cells B3:F3 [“LWDA Case Number”; “PAGA Notice Filing Date”], B64618:F73620 and B73621:F83328 [calculation: distinct nonblank LWDA strings; July 2024–June 2025 = 8,962; July 2025–June 2026 = 9,597; 635/8,962 = 7.1%; 10 of 12 months higher; half-year differences +333 and +302; 9,597 is the maximum complete rolling twelve-month window beginning January 2017].)↥
(Office of Governor Gavin Newsom, Governor Newsom & Legislative Leaders Announce Agreement on PAGA Reform (June 18, 2024) [as of Aug. 2, 2026] [“announced an agreement”]; Lab. Code, § 2699, subd. (v)(1)–(2), added by Stats. 2024, ch. 44, § 1, AB 2288 [“on or after June 19, 2024”; “filed before June 19, 2024”] [application rule and notice-based exception]; Stats. 2024, ch. 44 [“Approved by Governor July 01, 2024”].)↥
(Exh. A, PAGA Notices - all time, cells B3:K3 [“LWDA Case Number”; “PAGA Notice Filing Date”; “Law Firm for PAGA Plaintiff”; “Employer”] [no motive or processing-lag field], B63823:F64389, B59229:F68455, B54832:F54936, and B73211:F73321 [calculation: distinct nonblank LWDA strings; June 19–21, 2024 = 117, 246, and 204, totaling 567; June total 1,237; matched Wednesday–Friday counts 105 and 111 and 567/((105+111)/2) = 5.25; seven-day counts 206/681/161; equal pre-window→post-window counts are 740→1,135, 1,522→1,760, 2,361→2,453, and 4,572→4,627, or +53.4%, +15.6%, +3.9%, and +1.2% over 30/60/90/180 days].)↥
(Exh. A, PAGA Notices - all time, cells B3:F3 [“LWDA Case Number”; “PAGA Notice Filing Date”], B63823:F64389 and B51446:F78203 [arithmetic sensitivity: 9,437 − 567 = 8,870; (8,870 − 7,900)/7,900 = 12.3%; 567/(9,437 − 7,900) = 36.9%; deletion is not a causal estimate].)↥
(Exh. B, RLLVS, cells B3:L3 [“LWDA Case Number”; “PAGA Notice Filing Date”; “Report of Labor Law Violation Create Date”; “Amended PAGA RLLV”], B4:L73097 [73,094 rows; 68,708 distinct identifiers; 3,893 identifiers with multiple rows], and M3:AO73097 [“Classified as Independent Contractor” through “Retaliation for Protected StatusActivity”] [29 Boolean fields; calculation: select maximum create date per identifier and later row for ties; cohort on notice year; 2017/2025 means 5.9686/8.9182 and medians 6/9; 0–3 counts 1,060/4,654 and 471/9,287; 8–11 counts 1,246/4,654 and 5,604/9,287; 12-plus counts 164/4,654 and 1,100/9,287; 2025’s share is higher at every threshold 1–28; raw-row 0–3/8–11/12-plus shares change 22.76%→5.27%/26.76%→59.90%/3.50%→11.86%; ever-selected shares change 22.45%→3.37%/26.92%→61.61%/3.55%→12.21%; create-year grouping preserves direction; first-to-latest cells change 17,377 true-to-false and 3,898 false-to-true, which are not classified as withdrawals].)↥
(Exh. B, RLLVS, cells P3:S73097 [“Minimum Wage”; “Overtime”; “Not paid for all hours worked”; “Not Paid Wages Due on Termination”], V3:V73097 [“PMNT Reimbursement of Empl Expenses”], Y3:Y73097 [“Meal and Rest Breaks”], and AE3:AE73097 [“Improper or Incomplete Wage Statements”] [calculation under note 8’s latest-row rule: 2025 counts and rates, in the stated field order, are 7,816/9,287 = 84.16%, 8,332/9,287 = 89.72%, 8,017/9,287 = 86.33%, 8,012/9,287 = 86.27%, 7,710/9,287 = 83.02%, 8,571/9,287 = 92.29%, and 8,584/9,287 = 92.43%; selecting all seven produces the annual series shown; raw-row endpoints 687/4,709→5,958/9,609; ever-selected endpoints 687/4,654→5,986/9,287; create-year endpoints 674/4,633→5,812/9,510].)↥
(Exh. B, RLLVS, cells N3:R73097, V3:V73097, Z3:Z73097, and AF3:AF73097 [“Nonexempt Classified as Exempt”; “Minimum Wage”; “Overtime”; “Not paid for all hours worked”; “PMNT Reimbursement of Empl Expenses”; “Sick Leave”; “Other Notice or Posting or Recordkeeping”] [calculation: expense reimbursement 1,453/4,654 = 31.22% in 2017 and 7,710/9,287 = 83.02% in 2025; its 0.517988 increase divided by the 2.9496-category mean increase = 17.56%; “contribution” is arithmetic, not causal].)↥
(Exh. A, PAGA Notices - all time, cells B3:G3 [“LWDA Case Number”; “PAGA Notice Filing Date”; “Law Firm for PAGA Plaintiff”], G4163 [“In pro per”] [field is not limited to law-firm entities], B26132:G32574, and B68824:G78203 [calculation: distinct nonblank LWDA strings; outer-trimmed, case- and punctuation-sensitive labels; 2019 nonblank n = 6,408, unique = 1,129, top ten = 1,150, top 25 = 2,105/6,408 = 32.8%; 2025 nonblank n = 9,294, unique = 777, top ten = 2,767, top 25 = 4,403/9,294 = 47.4%, largest label = 680 and 680/9,294 = 7.3%; total notice change (9,295−6,421)/6,421 = 44.8%; light-normalization top-ten shares = 19.4% and 32.0%; labels are not verified entities].)↥
(Exh. B, RLLVS, cells B3:K73097 and M3:AO73097 [“Law Firm for PAGA Plaintiff”; 29 structured allegation fields] [calculation: latest-row rule; trim and collapse whitespace while preserving case and punctuation; require at least ten records in both 2019 and 2025; 42 labels cover 1,687/6,357 and 2,593/9,287; 34 of 42 mean-bundle increases and 38 of 42 expense-rate increases; preserving internal double spaces changes the bundle result to 35 of 42].)↥
(Exh. A, PAGA Notices - all time, cells B3:H3 [“LWDA Case Number”; “PAGA Notice Filing Date”; “Employer”] and B68824:H78203 [calculation: 9,295 distinct 2025 LWDA strings; 9,267 nonblank employer strings; 7,949 normalized strings (Unicode NFKD, casefold, replace ampersands with “ and ”, replace each run outside ASCII a–z/0–9 with a space, collapse whitespace and trim); 2,208 records fall within strings appearing at least twice; 2,208/9,295 = 23.75%; text groups are not verified employers].)↥
(Exh. C, NOIs, cells B2:B3 [“PAGA cases with Notice of Intent to Investigate (NOI)-HQ64881”; “PAGA Notice Filing 2016-2016”] [second line conflicts with dates through 2026], B5:AX203 [198 rows; 177 distinct LWDA strings; 80 blank BOFE cells]; Exh. D, PAGA Issues+Citation, cells B5:Q9 [“Status of Citation”; “Under Appeal”; “Canceled - Invalid”] [same case/issue has different status records and rows 8–9 are exact duplicates], B6:O728 [723 rows; 104 BOFE keys; 71 shared with Exhibit C; many-to-many join], and P5:Q729 [“Total Amount of Civil Penalties Owed to State”; “Balance of Civil Penalties Due to State”] [case-level amounts repeat; footer is static and has no stated as-of date].)↥
(Cal. Dept. of Industrial Relations, Private Attorneys General Act (PAGA)—Filing (Sept. 2025) [as of Aug. 2, 2026] [“Past records are also subject to being destroyed”] [retention warning]; Exh. A, cells B1 [“PAGA Notices Filed 2004-2026 (7/30/26 at 11:25am)”] and F3:F17 [14 blank dates]; Exh. B, cell B1 [“Submission Date (9/6/2016 to 7/29/2026)”] [2026 is incomplete].)↥
(Exhibit manifest and structural audit: Exh. A, HQ64881-PAGA Notice filed 2004-2026.xlsx, worksheet PAGA Notices - all time, cell B1 [“PAGA Notices Filed 2004-2026 (7/30/26 at 11:25am)”], SHA-256 886a7b57e8637f8d613c4c3eb446eb60cc97f4454bbdba5ff13ea8234ba958fd; Exh. B, HQ64881-RLLVs-2016-2026.xlsx, worksheet RLLVS, cell B1 [“Report of Labor Law Violation Submission Date (9/6/2016 to 7/29/2026)”], SHA-256 bf5e97d1dee9435b121506bcbbfc6b91b28a560a8ea8ca8d7f783f04681d8871; Exh. C, HQ64881-Notice of Investigations.xlsx, worksheet NOIs, cell B2 [“PAGA cases with Notice of Intent to Investigate (NOI)-HQ64881”], SHA-256 df5b975880dd476b92437404c9bc5473536a26c50ddc7c92aa9a4e5d4678b38b; Exh. D, HQ64881-PAGA Issues+Citation.xlsx, worksheet PAGA Issues+Citation, cells B2:B3 [“PAGA Issues+Citation- HQ64881”; “PAGA Notice Filing 2016-2026”], SHA-256 c826240fc8214a87e43bf1706dbdf7761735bd3f13744fec14a5846f22ab6c45. Each hash was calculated from the complete XLSX byte stream. The audit traversed all 4,395,313 positions inside the four used worksheet rectangles and inventoried every OOXML package member; each workbook has one visible sheet and no hidden rows or columns, formulas, comments, hyperlinks, external links, macros, drawings, or pivot tables. XLSX files have no stable page numbers and none of the sources is a message transcript, so worksheet/cell locators substitute for inapplicable page/message-turn locators.)↥

Bottom line

The strongest conclusion is structural, not causal.

DIR’s records show a higher period of dated notice records evident by 2023, a dramatic but narrow June 2024 concentration, broader and increasingly standardized allegation packages, and increased concentration among recurring representative-field labels. They do not establish why those changes occurred. ^{Source 1 · Source 3 · Source 4 · Source 6 · Source 8 · Source 9 · Source 11 · Source 14}