We've written before about what investors want to see in your people data – the shape of it, and why messy data costs leverage.

This is the operational version. The actual request list. What each item is really testing. What it costs when the answer is wrong. And which of it your HRIS can produce on demand versus which of it turns into three weeks of somebody's evenings.

None of this is exotic. That's the point. It's the same list every time, and companies at 50–200 employees are surprised by it every time.

The request list

A buyer's employment counsel will structure their request roughly this way. Assume three years of history on every category unless noted.

Census and org. Employee census as of the most recent month-end. Org chart. Headcount by function, level, and location. Open requisitions against the hiring plan, including signed offers not yet started.

Compensation. Base, bonus target, commission plan, equity, and other compensation for every employee. Commission plan documents. Bonus plan documents. Compensation change history. Any promised-but-undocumented arrangements.

Agreements. Every employment agreement, offer letter, severance agreement, change-of-control agreement, retention agreement, arbitration agreement, non-competition and non-solicitation covenant, and – the one that gets people – every confidential information and invention assignment agreement, for employees and contractors both.

Equity. Cap table. Full grant ledger reconciled to the cap table and to your plan administrator. Board consents approving every grant. Every 409A valuation covering every grant date. Exercise records and 83(b) elections. Any phantom equity, SARs, or profit interests.

Contractors. A separate schedule from the census: name or entity, whether individual or entity, start date, scope, rate, 1099 totals by year, whether a written agreement exists, whether that agreement assigns IP, whether engaged through an agency or EOR, and whether the person has ever been an employee.

Benefits and retirement. Plan documents, trust agreements, and every amendment. Summary plan descriptions. Three years of Form 5500s with schedules and any required audit reports. Nondiscrimination and coverage testing results. Section 105(h) and 125 testing. Three years of claims experience and renewal quotes. COBRA documentation and procedures. Any correction filings you've made with the IRS or DOL – the voluntary correction programs.

Wage and hour. Timekeeping records. Exemption analysis for every exempt classification. Regular-rate calculations. Meal and rest break records for California. Wage statement formats.

Immigration. I-9s for every employee. E-Verify status. Sponsored visas. Any ICE or USCIS history.

Policies. Handbook, all standalone policies, acknowledgment records, required postings.

Claims and litigation. Every charge, claim, investigation, and agency matter with date, type, venue, status, exposure, and insurance coverage.

Terminations. Termination log split by voluntary, discharge, layoff, and reduction in force. Severance paid. Unemployment claims. Any WARN-triggering events.

Multi-state. Every state where an employee works or lives, with foreign qualification, withholding account, unemployment insurance account, state disability and paid family leave enrollment, and workers' compensation coverage.

Deal-dependent additions: EEO-1 filings and affirmative action programs for federal contractors and larger filers, OSHA logs for field or industrial workforces, collective bargaining agreements and withdrawal liability letters where there's a union, pay equity analysis (increasingly requested), and – newly – an AI tooling inventory.

There is no such thing as "the census"

The most common founder mistake in this whole process is assuming the census file you send your benefits broker is the census the deal team wants. It isn't. There are three, and they have different fields, different as-of dates, and different consumers.

The deal census backs a representation in the purchase agreement, which is why it's requested as of the most recent month-end and then refreshed at signing and again before closing. The field set, taken from how these schedules are actually defined in executed purchase agreements: name or employee ID, hire date, employing legal entity, title, full-time or part-time, leave status and type, work location by country and state, annualized base, bonus potential, commission potential, other compensation, unpaid or accrued commissions and bonuses, PTO accrual balance, severance potential, exempt or non-exempt status, visa status, and benefits eligibility. Buyers routinely add FTE fraction, department or cost center, manager, pay type and frequency, last increase date, and performance rating.

The benefits and actuarial census is demographic: date of birth, gender, ZIP code, coverage tier elected, dependent dates of birth, salary for life and disability volume, hire and eligibility dates, COBRA participants with qualifying event dates. It has to tie to the carrier's current enrollment file.

The QoE census is the one that breaks companies. It's requested as of each historical period end across the review window – typically monthly or quarterly for the trailing 24 to 36 months – because the analyst is building a headcount roll-forward. That is a point-in-time query, not a current snapshot, and most HRIS reporting is current-state only.

Open requisitions are not census rows. They sit on their own schedule: req ID, title, department, location, target start date, budgeted base and variable comp, whether it's in the budget, and offer status. Signed offers not yet started are a committed cost, and they're the item buyers most often pull into run-rate compensation.

The reconciliation that decides how the rest of diligence goes

Three numbers have to tie:

  1. HRIS headcount as of a date
  2. The payroll register for the pay periods spanning that date
  3. General ledger compensation and benefits expense for the period, by cost center

If they don't tie, nothing catastrophic happens immediately. What happens is worse in a slow way: the QoE team stops treating management's numbers as reliable, expands scope, and gets more conservative on every judgment call in front of them. That flows straight into the cash-to-accrual conversions and run-rate adjustments below, and into how much benefit of the doubt you get for the rest of the process.

A census that doesn't reconcile to payroll also shows up a second time, in a different room, as a potential 401(k) operational failure – because census-to-payroll reconciliation is a standard benefit plan audit procedure testing eligibility, compensation definition, and deferrals.

Where people data moves the price

Quality of earnings adjustments. The people-cost adjustments that show up in nearly every lower-middle-market QoE:

  • Owner and founder compensation normalization – adjusting owner pay to market replacement cost. An owner-CEO paid $500,000 in a role that markets at $250,000 produces a $250,000 annual add-back. This one increases adjusted EBITDA and is the most-negotiated people item in the deal.
  • Owner-discretionary payroll – family members on payroll who don't work in the business, and personal expenses run through comp.
  • Cash-to-accrual conversion of payroll, benefits, 401(k) match, PTO, bonuses, and commissions. This is where a company that has never accrued a bonus takes a one-time hit.
  • Pro-forma run-rate compensation – repricing the P&L to current headcount and current rates. Mid-year raises, partial-year new hires, and signed-not-started offers all get annualized. This generally reduces adjusted EBITDA.
  • Unfilled roles. If the buyer concludes a role must be filled to run the business, you get charged for it whether or not you were running without it. A lean org with an open VP Finance seat is not free.
  • Transaction-triggered compensation – change-in-control bonuses, retention and stay awards, transaction bonuses. Typically excluded from adjusted EBITDA and handled as debt-like items or transaction expenses.
  • Phantom equity, SARs, and management puts – identified with vesting schedules, valuation dates, exercise prices, withholding, and employer-side payroll taxes.

Working capital versus debt-like. Accrued payroll, PTO, and bonuses are a live fight in most deals. Buyers often prefer to pull them out of working capital and treat them as debt-like, reducing the price dollar-for-dollar at closing. Advisors on the sell side generally argue for keeping them in working capital so they're captured in the post-closing true-up. The economics at closing are similar; the difference shows up later.

Two things to know regardless. First, payroll accrual accounts in most companies at this size are reviewed periodically rather than maintained – they are stale, and diligence will find it. Second, a legitimate diligence adjustment can consist of proposing an accrual where none existed before, which creates a liability that was never on your balance sheet. Unaccrued PTO in a state where accrued vacation is a wage payable at termination is the classic version.

The representations and warranties problem. Employment reps in a US private-target deal conventionally cover: the census is complete and accurate; compliance with employment laws including wage-hour, classification, and immigration; no pending claims or investigations; no union activity; no WARN events in the preceding 90 days; at-will employment except as scheduled; all employees and contractors who contributed IP have executed invention assignment agreements; and no payment becomes due solely by reason of the transaction.

Here is the mechanic that matters, and it's the least-understood thing in this article:

Wage-and-hour and worker-misclassification exposure is among the most commonly excluded items in a representations and warranties insurance policy – Hinshaw & Culbertson lists employee/contractor misclassification and wage-and-hour issues alongside projections, asbestos and environmental, cyber, pension underfunding, and receivables collectability as standard exclusions. The same analysis describes the underwriting mechanic: insurers lean on secondary diligence of the buyer's primary diligence rather than conducting their own review.

Which means the quality of the buyer's employment diligence memo directly determines whether wage-hour risk is covered or carved out – and a carve-out converts straight back into seller indemnity.

Thin employment diligence doesn't make the risk go away. It moves the risk onto you.

The market context is that claim severity is rising. Aon reported in June 2026 that median North American R&W claim payments exceeded $8.2 million in 2025, up from $5.5 million in 2024, with compliance-with-laws the most frequent breach category at more than 20% of notifications – and 51% of claims now filed more than twelve months after closing. Rising severity is exactly what drives underwriters to tighten exclusions.

That timing statistic has a practical consequence: a twelve-month escrow doesn't cover the tail. Neither does it cover the statute of limitations on the underlying claims – FLSA runs two years, three for willful violations, and California meal and rest break claims run three.

Nine findings that cost money

For each: the failure, the artifact that proves you're fine, and the cost when you aren't.

1. Missing invention assignment agreements

Proves it's fine: a 100% complete matrix of every current and former employee and every contractor who touched product, mapped to a countersigned confidential information and invention assignment agreement executed on or before their start date.

Costs: this is a chain-of-title defect in your primary asset. Remediation means getting signatures retroactively, which requires fresh consideration in many states and is often impossible for people who've left. The practical outcome is a special indemnity, an escrow, or a covenant to obtain a stated percentage of signatures before closing.

At 15–250 employees this is the single most common material finding we see, and it's almost always a contractor gap rather than an employee gap.

2. Contractor misclassification

Proves it's fine: written agreements for every 1099 relationship, a documented classification analysis under the applicable federal and state tests including California's ABC test, no employee-to-contractor conversions without a defensible break, and 1099 totals reconciled to the AP ledger.

Costs: back overtime, back payroll taxes, retroactive benefit plan eligibility (which can itself become a 401(k) coverage failure), penalties, and class exposure. And because it's now a known issue, R&W insurance won't cover it. We wrote about the underlying problem in worker classification compliance.

3. California meal and rest breaks

Proves it's fine: timekeeping records showing meal period punches, an exemption analysis for every exempt classification based on duties rather than salary alone, and regular-rate calculations that include nondiscretionary bonuses and commissions.

Costs: under Labor Code § 226.7, failure to provide a required rest period obligates one additional hour of pay at the regular rate for each workday, capped at one hour per day regardless of how many were missed – with a parallel premium for meal periods. Three-year lookback. A single misconfigured timekeeping rule across a 200-person California hourly workforce compounds into seven figures, plus derivative wage statement claims.

4. Multi-state registration

Proves it's fine: a registration matrix by state showing foreign qualification, withholding account, unemployment insurance account, state disability and paid family leave enrollment, workers' compensation, and required postings – for every state where an employee works or lives.

Costs: back withholding, back unemployment contributions, penalties, interest, and in a stock deal, direct successor liability. This is the most common finding in remote-first companies built after 2020, and it's the one where the tax and employment workstreams collide.

5. I-9s – and this changed in March 2026

Proves it's fine: a complete I-9 for every employee hired after November 6, 1986, stored separately from personnel files (best practice and an inspection-exposure control, not a statutory requirement), with retention applied (three years from hire or one year from termination, whichever is later), reverification tracked, and E-Verify status documented.

Costs: per the DHS penalty adjustment effective January 2, 2025, a paperwork violation runs $288 to $2,861 per Form I-9. Knowingly hiring an unauthorized worker starts at $716 and escalates on repeat offenses.

What changed: on March 16, 2026, ICE reclassified more than ten categories of I-9 error from "technical" to "substantive," eliminating the traditional ten-business-day cure period for them. The reclassified list includes missing date of birth, missing USCIS number, incomplete document data in Section 2 even where the employer retained copies of the documents, missing employment start date, and remote verification without active E-Verify enrollment. Illustrative math from practitioners: an employer with 200 deficient forms now faces roughly $57,600 to $572,200 in exposure. Violations remain live until corrected, and the five-year limitations clock only starts running after remediation – so proactive fixing starts the clock while post-inspection fixing gets minimal mitigation.

The practical consequence for anyone thinking about a transaction: "we'll clean up the I-9s at close" stopped working in March 2026. It's a pre-LOI activity now.

6. 401(k) operational failures

Proves it's fine: a deposit-timing log showing, for every pay date, when deferrals were withheld and when they hit the trust. Filed 5500s with acceptance confirmations. The independent auditor's report if you crossed the large-plan threshold. Any correction program filings.

Costs: late deferral deposits are prohibited transactions. The rule is "as soon as the employer can reasonably segregate" – the IRS is explicit that the 15th business day of the following month is an outer limit, not a safe harbor. Plans with fewer than 100 participants get a 7-business-day safe harbor. The excise tax is 15% of the amount involved per year, plus an additional 100% if not corrected.

Missed deferral opportunities from eligibility errors generally require a corrective contribution of 50% of the missed deferral plus 100% of the missed match, adjusted for earnings, with reduced tiers available for prompt correction and written participant notice within 45 days.

Failure to file a Form 5500 runs $2,739 per day with no cap – the DOL held ERISA penalties at 2025 levels for 2026 rather than applying an inflation adjustment. And a large plan filed without the required audit is treated as not filed, which puts that daily clock in play retroactively. Buyer's counsel checks Schedule H line 4a first, every time.

7. COBRA

Proves it's fine: the administrator agreement, a qualifying-event log tied to the termination log with dates, proof of timely election notices, and the initial notice given at plan entry.

Costs: an excise tax of $100 per day per qualified beneficiary during noncompliance, capped at $200 per day where multiple beneficiaries share a qualifying event. Failures not corrected before an IRS examination notice carry minimum taxes of $2,500 per qualified beneficiary, rising to $15,000 where the failure is more than de minimis. It's self-reported on Form 8928, and a buyer will ask whether you've ever filed one.

8. Unwritten compensation practices

Proves it's fine: signed commission plans that define when commission is earned versus when it's paid, written bonus plan documents stating discretion and any employment-on-payment-date condition, and documented approvals for merit and promotion increases.

Costs: two problems at once. A QoE timing problem – the analyst can't determine when revenue-linked comp is earned – and a claims problem. Undocumented "discretionary" bonuses have a way of becoming accrued liabilities. Verbal arrangements appear explicitly on employment diligence checklists for exactly this reason.

9. PEO exit

This is the most underrated item on the list for companies in this range, and it deserves its own paragraph.

If you're on a PEO, plan on materially more diligence time. Plan documents, insurance policies, and claims experience are harder to extract from a PEO than from a company that sponsors its own plans. The benefit-plan representations in the purchase agreement have to be rewritten to address controlled-group plans plus more limited reps for PEO-sponsored plans. Most PEO agreements require 30 to 60 days' notice to terminate, sometimes longer – which can force a buyer to keep the PEO in place after closing in a stock deal.

The hardest part is exiting the PEO's multiple-employer 401(k), which typically requires adopting a mirror plan and then terminating it before closing. And for mid-year transactions, an uncertified PEO may require resetting accumulated FICA and SUTA wage bases for affected employees – a real cash cost landing entirely in the transaction year.

Get the deductible and out-of-pocket accumulator reports too, or your people restart their deductibles mid-year on day one of new ownership. That's a Day-1 experience problem you will hear about.

What your HRIS has to be able to produce

Mapping the request list to system capability. The items marked ▲ are where companies under 250 people most often discover the system can't do it.

What diligence needs Why Reality check
Point-in-time census ▲ Every historical period end in the QoE window; carve-out transfer lists The hardest ask. Most HRIS reporting is current-state only.
Effective-dated field history ▲ Reconstructing comp, title, location, FTE, and status at each period end Requires storing when a change took effect, not just when it was entered.
Comp change history with reasons Separating merit from promotion from market adjustment from retention in run-rate normalization Reason codes are usually optional and therefore usually blank.
Org and reporting history ▲ Span of control, integration planning Requires effective-dated manager relationships, not a current org chart.
Attrition with reason codes and regrettable flags Turnover analysis is a standard diligence item The taxonomy has to be consistent across the whole lookback.
Headcount reconciliation to payroll The QoE roll-forward Hardest where payroll sits in a different system.
Document completeness reporting ▲ Signed offer letters, invention assignments, handbook acknowledgments, I-9s You need a who is missing what report, not just storage.
Multi-state registration status Nexus and successor liability Almost never in the HRIS. Usually a spreadsheet nobody owns.
Accrual balances by policy and state, as of a date Working capital and PTO liability
Open reqs tied to the hiring plan The unfilled-role adjustment Requires the ATS and the budget to be the same object.
Audit trails Who changed what, when, with approval status Underwriter scrutiny.

Two platform notes, both worth verifying in your own tenant before you rely on them:

Rippling shipped Object History in its Data Cloud in June 2026, which directly addresses the point-in-time problem. It exposes VALUEASOF(field, date) and DATEOFCHANGE(field) as functions usable in reports, dashboards, and transformations, captures who made each change and its approval status, and applies the same permission model to historical data as to current data. Rippling's stated limitation as of launch is that it covers employee record changes and custom fields – whether effective-dated org and manager hierarchy is fully included is worth confirming before you promise a buyer a historical org chart.

HiBob publishes claims of custom reports "for any time period," audit-ready reporting, and secure archives of historical data. We could not verify an explicit point-in-time or effective-dated query capability equivalent to the above from public documentation. Don't assume parity in either direction – ask, and test it against a date eighteen months back.

Two things that are new in 2026

Non-competes are back to being a state-law question. The FTC's Non-Compete Clause Rule was set aside by a federal court; the Commission voted in September 2025 to dismiss its appeals and accede to vacatur; and 16 CFR Part 910 was formally removed from the Code of Federal Regulations effective February 12, 2026. A lot of content still describes the rule as pending. It isn't.

This did not simplify diligence. It made it jurisdictional. Enforceability now depends on where each covered employee works, which means the analysis runs state by state against your census.

AI in HR is entering diligence through the compliance-with-laws door. Littler's employer survey published in May 2026 found 68% of employers now have formal AI use policies, up from 38% a year earlier; 55% have formal review and approval processes for AI tools; 54% restrict what information can be entered into AI systems; and 79% report concern about AI-related litigation risk.

Those first three numbers are exactly the artifacts a buyer can now credibly ask for. And where AI is used in hiring or employment decisions, the hook is the compliance-with-laws representation – which is already the most frequently breached category in R&W claims. We haven't yet seen an AI line item on a published people-diligence request list, so treat this as a direction rather than a documented standard. But if you're deploying AI in hiring in California or Illinois, the policy work is now part of your diligence readiness, not a separate project.

What replaces the fire drill

The failure mode is always the same: a data cleanup project six weeks before the process kicks off, run under pressure by the person who least has time for it, reconciling records and hoping the gaps don't surface in the wrong meeting.

What actually works is boring and continuous:

  • Monthly. Headcount reconciled to payroll and to the GL. New hires' document set confirmed complete – offer letter, invention assignment, I-9, handbook acknowledgment – before the first payroll run, not after.
  • At every change event. Compensation updated with a reason code. Title, manager, department, and location updated with an effective date. Terminations logged with a reason and a regrettable flag.
  • Quarterly. Contractor list reviewed against the classification test. Multi-state registration matrix refreshed against where people actually live. Accrual balances tied to the GL.
  • Annually. 5500s filed and confirmed. Benefit plan testing reviewed. I-9 self-audit – and given March 2026, this one moved up the priority list considerably.

When diligence comes, you're not preparing. You're pulling.

The pre-LOI pull list

Twelve artifacts to have ready before anyone sends a request list. If any of these takes more than a day to produce, that's the work:

  1. Current-month census, full deal field set, reconciled to payroll
  2. Historical headcount roll-forward, monthly, trailing 24 months, tied to the GL
  3. Org chart with effective-dated manager history
  4. Invention-assignment completeness matrix – every employee and contractor who touched product
  5. Equity ledger reconciled to the cap table, with board consents and 409A coverage for every grant date
  6. Contractor schedule with classification rationale per relationship
  7. Commission and bonus plan documents, signed
  8. Benefits binder: plan docs, three years of 5500s with acceptance confirmations, testing results, claims experience
  9. 401(k) deposit-timing log
  10. I-9 self-audit results – post-March 2026, this one is pre-LOI work, not pre-close
  11. Multi-state registration matrix against where people actually live
  12. Claims and charge log with insurance coverage noted

People Street's take

Diligence doesn't find problems. It finds problems and it finds out how long you took to answer, and the second finding does more damage than the first.

A missing invention assignment agreement is a fixable issue with a known remediation path. A missing invention assignment agreement that took eleven days to discover because nobody could tell which employees had one tells a buyer something much more expensive: that the people function has been run on memory rather than on a system, and that whatever else is in there hasn't been found yet.

The companies that come through this well aren't the ones that scrambled hardest in the last two weeks. They're the ones whose HRIS was configured to answer these questions from the start – effective-dated, reconciled monthly, with a document completeness report somebody actually looks at.

If you're twelve to eighteen months from a process and you can't produce a point-in-time census as of last September, that's the gap worth closing now. Book a 20-minute call and we'll tell you what a buyer is going to find.

Related: What investors want to see in your people data · What a clean HRIS looks like · You're growing fast. Your compliance isn't.