PCAOB Auditor & Engagement Change Monitor

An auditor resignation is one of the oldest distress tells there is. So is an unexplained engagement-partner change at a bank that just restated, or a lead partner quietly serving a seventh consecutive year on an issuer audit. This actor turns PCAOB's daily Form AP archive into a clean, machine-readable feed of auditor switches, engagement-partner rotations and tenure, first-year audits, component-audit disclosure, amendment lineage and per-firm mandate wins and losses — for every SEC issuer, employee benefit plan and investment company that files.

Data source: assets.pcaobus.org/firm-filings/FirmFilings.zip — the official PCAOB Form AP extract, refreshed daily, ~13 MB compressed / ~93 MB uncompressed, 62 columns, 155,000+ filings covering the full history back to February 2017.

Who this is for

Audit committees and governance analysts. You need to know the moment an issuer's audit firm changes, whether the outgoing firm was replaced by a smaller one, and whether your lead engagement partner is approaching the SEC's five-year rotation limit. mode: "changes" gives you the switch feed with prior_audit_firm, prior_partner and prior_fiscal_period_end on the same row; partner_consecutive_periods and partner_rotation_status give you the rotation clock.

Short sellers and forensic accountants. A Big Four to micro-cap auditor downgrade, a partner rotation immediately before a restatement, a Form AP that amends the audit report itself rather than a typo, or an audit where a large share of the work was performed by a firm in another country — those are all first-class fields here, not something you reconstruct by hand. amended_sections, other_participating_firm_countries and participant_range sit on the row.

Audit-firm competitive intelligence and business development. mode: "firm-rollup" gives you mandate wins, mandate losses, net change, market share, distinct partners fielded, and the split between operating issuers, benefit plans and investment companies — per firm, filterable to your target segment.

SOX and independence compliance teams. SEC Regulation S-X 2-01(c)(6) caps the lead engagement partner on an issuer audit at five consecutive years. partner_rotation_status reports within_five_periods, fifth_consecutive_period or over_five_consecutive_periods — and unknown_left_censored when PCAOB's own history is too short to tell, rather than guessing.

Investor relations and benchmarking. Compare your auditor, partner tenure and audit structure against peers in the same segment. bankIssuersOnly narrows the file to depositories.

Bank credit and counterparty risk. bankIssuersOnly: true reduces the file to banks, thrifts and bank holding companies. An auditor change at a mid-size bank is worth a phone call.

What makes this different from reading Form AP yourself

• Identity is not naive. PCAOB reuses one Issuer ID for a company and its benefit plans (Issuer ID 32577 covers "PulteGroup, Inc.", "PulteGroup, Inc. 401(k) Plan" and two more name variants — 16.0% of Issuer IDs carry more than one entity name). One company can also hold several Issuer IDs (Zillow Group is 596939 and 596940 under CIK 0001617640 — 16.5% of CIKs do this). Keying on Issuer ID alone fabricates a switch between a 10-K auditor and a 401(k) auditor, and hides real EY-to-Deloitte switches. This actor keys CIK first, then entity type, then plan slug, and publishes the key it used in issuer_entity_key so you can audit the join.
• A negative is only asserted when it was actually checked. is_auditor_switch: false means there was a predecessor and the firm did not change. Where there is no predecessor the field is null, not false, and prior_source says why. The same contract governs is_new_since_last_run.
• The source is asserted live on every run. Row-count floor, a freshness window on the newest filing date, row-width uniformity, firm and issuer breadth, and the Latest Form AP Filing flag vocabulary. A stale or truncated archive that still parses perfectly fails the run instead of quietly returning a plausible answer. Nothing is emitted and nothing is billed when it does.
• The Issuer Ticker column is a trap. Its header literally contains a newline (Issuer Ticker \n(Not Available)), so a naive split yields 21 columns instead of 62 and silently corrupts everything after it — and the column is empty in 0 of 155,189 rows. It is parsed correctly and deliberately not emitted. Join issuers on CIK or name.

Modes

The delta baseline

Cross-run state lives in a named key-value store (pcaob-auditor-engagement-state); Apify's default store is created fresh for every run and would never persist anything. The baseline is keyed by the filter scope that decides which rows enter it (latestOnly, bankIssuersOnly, issuerNameContains), so a bankIssuersOnly run and a national run keep separate baselines and never overwrite each other. delta_scope on every row records which baseline was used.

If the baseline could not be loaded for any reason, is_new_since_last_run is null on every row and baseline_status names the reason. It is never false.

Example input

{
  "mode": "changes",
  "bankIssuersOnly": true,
  "latestOnly": true,
  "sinceFilingDate": "2026-01-01",
  "maxResults": 200
}

Every audit-firm change, partner rotation and first-year audit at a US bank, thrift or bank holding company, filed since the start of 2026, newest first.

A few more:

{ "mode": "new-filings", "maxResults": 500 }

{ "mode": "firm-rollup", "latestOnly": true, "maxResults": 100 }

{ "mode": "engagements", "issuerNameContains": "bancorp", "maxResults": 1000 }

{ "mode": "changes", "firmNameContains": "Deloitte", "sinceFilingDate": "2025-01-01" }

{ "mode": "fund-series", "firmNameContains": "Cohen", "maxResults": 2000 }

firmNameContains matches either side of a switch, so you see mandates won and lost.

Output fields

Identity

The engagement

Change detection

Cross-run delta

Partner and firm tenure

Audit scope and component audits

Amendment lineage

Fund series (Investment Company filings)

Who signed, and provenance

firm-rollup mode

audit_firm, firm_id, firm_country, engagements, distinct_issuers, distinct_partners, mandate_wins, mandate_losses, net_mandate_change, partner_rotations, first_year_audits, bank_issuer_engagements, issuer_engagements, benefit_plan_engagements, investment_company_engagements, amended_filings, divided_responsibility_engagements, engagements_with_other_participants, engagement_share_pct, latest_filing_date, new_filings_since_last_run, plus the provenance fields.

Use as an MCP tool

This actor is available to AI agents through mcp.apify.com. Every input and output field carries a description, so an agent can select and chain it without extra prompting. Useful agent phrasings:

• "Has this company changed auditors, and to whom?" — mode: "changes", issuerNameContains
• "What Form APs were filed since yesterday?" — mode: "new-filings"
• "Is the lead partner past the five-year rotation limit?" — mode: "engagements", then read partner_rotation_status
• "Which audit firm is winning the most bank mandates?" — mode: "firm-rollup", bankIssuersOnly: true
• "Was any of this audit performed outside the US?" — read other_participating_firm_countries

Pricing

$12 per 1,000 results (pay-per-event, Result event only — one event per dataset row). Graduated discounts apply on paid Apify plans. There are no hidden charge events, and a failed run emits nothing and bills nothing.

Cost control: maxResults is a hard cap applied before rows are built, so a capped run is cheap as well as small. It defaults to 200 — the same as the prefill — so a bare {} from an API or MCP caller returns a small, cheap sample rather than a sweep nobody asked for. Raise it deliberately. A national changes run over the full history finds ~31,000 change events; a scheduled new-filings run typically returns tens of rows a day.

FAQ

How fresh is the data? PCAOB refreshes FirmFilings.zip daily. Measured 2026-07-28: 155,189 rows, newest filing dated 2026-07-27. The actor fails the run when the newest filing date falls more than 21 days behind, so a stale cached archive can never be served to you as a quiet market.

How fast is it? The 93 MB CSV is decoded in 4 MiB slices straight into a streaming parser, filtered as it goes, and pushed in batches — the full file is never held as one string. A bank-only run finishes in about 6 seconds; a full national run in about 20. Comfortably inside Apify's timeout.

Why did my first new-filings run fail? By design. That mode answers "what is new since last time", and with no baseline the honest answer is "not knowable yet" — a different claim from "nothing is new". The run records the baseline, fails without billing you, and the next run at the same scope returns the real delta. Use mode: "changes" if you want results immediately.

Does an audit-firm rename show up as an auditor switch? No. Comparison is on firm_id, which PCAOB keeps stable across renames. Crowe Horwath LLP becoming Crowe LLP is not a switch; the old name surfaces in firm_other_name.

Why is a company's 401(k) plan a separate row from the company? Because it is a separate audit, often by a different firm. PCAOB gives both the same Issuer ID; this actor separates them by entity type and plan slug, so a 401(k) auditor is never mistaken for a change of the 10-K auditor.

Can I diff against my own prior data instead of the stored baseline? Yes — pass priorEngagements as an array of

{ issuer_cik, issuer_id, issuer_name, audit_report_type, firm_id, audit_firm, engagement_partner, fiscal_period_end }

What does latestOnly: false do? Keeps superseded amendment history so you can see how a filing changed over time. Amendments of the same engagement still dedupe to one row per issuer x firm x fiscal period, preferring the current filing.

Why is there no ticker field? PCAOB publishes the column but has never populated it — empty in 0 of 155,189 rows. Rather than emit a permanently blank column, this actor omits it. Join on issuer_cik.

Related actors

• RIA Registration Delta Monitor — the same delta pattern over SEC investment-adviser registrations. A natural pair if you cover financial-services counterparties.
• FDIC Bank M&A, Charter & Structure-Change Monitor — bank mergers, charter changes and failures. Combine with bankIssuersOnly: true here: an auditor change plus a pending structure change at the same bank is a far stronger signal than either alone.
• SBOM Vulnerability Rollup — if you are assembling an issuer-level risk file, the software-supply-chain layer of the same picture.

Data is republished from the Public Company Accounting Oversight Board's public Form AP archive. This actor does not add, correct or interpret PCAOB's data beyond the derived fields documented above; partner_rotation_status and is_bank_issuer in particular are derived heuristics, not determinations by PCAOB or the SEC.