The Employer Walkaway-Cost Reference: Vesting, Clawbacks, Payout Dates
For any employer on a slate, find the public record behind each forfeiture mechanism and produce a dated forfeiture curve, cheapest-exit window, and buyout range.
Key takeaways
- Amazon's back-loaded 5/15/40/40 schedule leaves 80% of a new-hire grant unvested at month 24, against 30% on Google's reported 38/32/20/10 curve, so the same tenure carries very different walkaway costs.
- A filed equity award form proves that a vesting mechanism exists, not which split a given candidate is on, and some forms literally read "Vesting Schedule: [TBD]".
- Private-markets data put 81% of terminated options on a 90-day exercise window while employee exercise rates fell from 46% to 23%, so walkaway cost includes vested options a candidate cannot afford to exercise.
- Under California AB 692, a sign-on clawback in an agreement from Jan 1, 2026 is void unless it meets every statutory condition, and a retention bonus granted after hire may carry no repayment obligation at all.
- NYC lets employers and headhunters ask about the value and structure of unvested equity and request documentation, but bars searching for information intended to uncover a specific applicant's pay.
- Refolk's index holds 182,844 US Senior Software Engineer profiles but 33,334 at Staff level, so senior slates can wait for cheap exit windows while Staff slates more often have to pay the buyout.
This reference is for in-house recruiters, sourcers, talent leaders, and founders who need to know, before the first call, what a candidate at a given employer gives up by leaving and when that cost drops. For each forfeiture mechanism it names the public record that documents it, what that record proves, and how it misleads. It then takes you to a dated forfeiture curve, a cheapest-exit window, and a buyout range, with all the research kept at the employer level so nothing touches an individual's pay history.
What an employer walkaway cost is
An employer walkaway cost is everything a person forfeits by resigning from that employer on a given date: unvested equity, vested options they cannot afford to exercise, cash they must repay, and bonuses they lose by not being on payroll when the bonus pays. The mechanisms and dates belong to the employer. Only the amounts belong to the candidate.
The whole reference is built on that split. You research the employer's vesting shape, cliff, cadence, clawback terms, and payout rules from public documents before the call. You get the candidate's amounts from the candidate, in the conversation, and only where the law allows the question.
Three terms recur:
- Forfeiture curve: a dated line showing what a person at this employer would lose by resigning on each day across the period you are hiring in.
- Cheapest-exit window: the days where that curve bottoms out, usually just after a large vest or bonus payout, adjusted for clawback step-downs.
- Buyout: the sign-on cash or equity the hiring company offers to cover forfeited value, also called a make-whole.
The layers of a walkaway cost
- Unvested equityNew-hire grant plus every refresh grant, each on its own schedule
- Vested options at riskOptions the person cannot afford to exercise inside the post-termination window
- Cash clawbacksSign-on and retention money repayable before a set date
- Unpaid annual bonusForfeited if the person is not employed on the payout date
- Benefit vestingESPP and retirement match, not researched here and kept as an open row
Walkaway-cost mechanisms: where each is documented and how it misleads
Each mechanism has a best public record, and each record proves less than it appears to. Use the table to find the record, then read the subsection for the trap.
| Mechanism | Best public record | Proves | Misleads when |
|---|---|---|---|
| Vesting shape and cliff | New-hire RSU form, schedule reports | A shape exists for some hires | Form reads [TBD]; hire year differs |
| Vest cadence | Exhibit wording | When value lands | Exhibit is an executive letter |
| Refresh grants | Schedule reports | A second curve exists | Only the new-hire grant is modelled |
| Option exercise window | Published policy, holman list | Deadline after leaving | Set in a layoff year; ISO converts |
| Sign-on and retention clawback | Sign-on agreement exhibits | Period and step-down | Gross wording; void under AB 692 |
| Bonus payout date | Plan text; proxy CD&A (unchecked) | Employed-on-date condition | Actual date later than typical |
Vesting shape and cliff
Vesting shape is the share of a grant that vests in each year. Front-loaded schedules vest most of the grant early; back-loaded schedules vest most of it late. A cliff is a period at the start of a grant before which nothing vests.
| Employer | Y1 / Y2 / Y3 / Y4 | Cliff | Vested by end of Y2 |
|---|---|---|---|
| Amazon | 5 / 15 / 40 / 40 | Yes | 20% |
| Google (reported) | 38 / 32 / 20 / 10 | None | 70% |
| DoorDash | 40 / 30 / 20 / 10 | Unverified | 70% |
| Cisco (2022 letter) | 25, then 6.25% a quarter | 1 year | 50% |
| GE HealthCare | 33 / 33 / 34 (3 years) | Annual | 66% |
The last column is derived from the split. Where the sources disagree or run thin:
- Google: crowdsourced compensation reports give 38/32/20/10, monthly, with no cliff, and have also seen 33/33/22/12 and 25/25/25/25. Another source still describes roughly 25% a year with a one-year cliff, which is probably out of date. The public dating of the shift traces to a 2022 Forbes article citing Business Insider.
- Asana: its filed RSU form tells drafters to enter a cliff of 12, 13, or 14 months from hire, with prorated cliff vesting if the person is terminated without cause.
- DoorDash: the split is reported; the cliff is not verified.
- Every employer above: no exact effective dates have been published for any of these schedules.
Proves: that the employer has used this shape for some new hires. Lies when: the candidate was hired under an earlier schedule, or the form you found leaves the split blank.
A filed award form proves the mechanism exists. It rarely proves the split your candidate is on.
Vest cadence
Cadence is how often vested shares land after the cliff: monthly, quarterly, or annually. It decides whether the forfeiture curve slopes or steps.
- Monthly (Google, as reported): the curve slopes smoothly, so there is no single cheap day and timing buys little.
- Quarterly (Cisco's 2022 new-hire letter: 25% after the one-year anniversary, then 6.25% a quarter): four small step-downs a year.
- Annual (GE HealthCare's 33/33/34 form, filed in two consecutive years' first-quarter 10-Qs): one large step-down a year, so the day after the anniversary vest is worth waiting for.
The sources behind this reference give Amazon's annual split but not its in-year cadence after the cliff. Confirm it before you place dates on an Amazon curve.
Proves: when value lands. Lies when: the exhibit is a named-executive offer letter, like Cisco's or AMD's. Those state real schedules and clawbacks, but only for executives.
Refresh grants
A refresh grant is an additional equity award given to an existing employee, usually on its own schedule. Google refresh grants reportedly vest 25% a year rather than on the front-loaded new-hire curve. A tenured employee therefore carries several overlapping curves, and the unvested balance on flat refresh grants can outweigh what is left of the new-hire grant.
Proves: that a second curve exists. Lies when: you model only the new-hire grant for a tenured employee.
Option exercise window
The post-termination exercise window is how long a departing employee has to buy their vested options before they lapse. It matters most at private employers, where exercising means paying cash for shares the person cannot yet sell.
| Measure | Value | Source |
|---|---|---|
| Terminated options with a 90-day window, Q4 2023 | 81% | Private-markets report |
| Options with a window over 90 days, peak 2023 quarters | 20% | Same report |
| Long-run share with an extended window | About 10% | Industry summary of same data |
| Older figure: windows of 90 days or less | 91.4% | Undated, around 2019 |
Two mechanics shape the cost:
- Tax conversion. An incentive stock option (ISO) is an option with favourable tax treatment; a non-qualified stock option (NSO) has none. Under IRC 422(a)(2), ISOs exercised after day 90 are taxed as NSOs, so the bargain element becomes ordinary income at exercise instead of an AMT preference item. After a permanent-disability termination the ISO window is one year, under IRC 422(c)(6).
- Affordability. The employee exercise rate declined for eight straight quarters, from 46% to 23%. Most departing option-holders lose vested options as well, so the walkaway cost includes vested-but-unaffordable options, not just unvested ones.
Extended windows exist. Pinterest extended its window to 7 years in 2015 for employees with two or more years of tenure, and the public holman/extended-exercise-windows list on GitHub tracks companies that publish one. But the share of extended windows doubled to about 20% in quarters when layoffs were heavy, against a long-run norm of about 10%. A window recorded in a downturn may not hold later.
Proves: the deadline. Lies when: the window is "extended" but ISOs still convert at day 91, or the policy dates from a layoff year.
Sign-on and retention clawbacks
A clawback is a contract term requiring repayment of a bonus if the person leaves before a set date. Filed clauses show periods of 12, 18, and 24 months. AMD's 2019 sign-on agreement is the clearest public example of a step-down: $500,000 paid over 24 months, 100% repayable before month 13, then reduced by 8.33% for each month after month 12. Another filed clause requires repayment of "the gross amount of the signing bonus", with a reduction-in-force carve-out.
California AB 692, codified at B&P §16608 and Labor Code §926, applies to agreements from Jan 1, 2026. A sign-on repayment term survives only if all of these hold:
- It is in a standalone agreement.
- The worker received written notice of the right to consult an attorney.
- The worker had at least five business days before signing.
- Repayment is triggered only by voluntary resignation or termination for misconduct.
- The retention date is no more than two years from the date the sign-on bonus was paid.
- No interest accrues, and the worker has an option to defer.
A retention bonus given after hire "may not include a repayment obligation, regardless of the specific terms". So for California employees on agreements from that date, post-hire retention bonuses add no walkaway cost. Whether AB 692 requires proration was not confirmed in the sources I reviewed; read the agreement and ask counsel before you assume either way.
Proves: the repayment period and step-down. Lies when: the clause repays gross while the candidate received net, or a California clause misses one AB 692 condition and is void.
Bonus employed-on-payout-date rules
An employed-on-payout-date rule pays the annual bonus only to people still employed on the day it is paid. Under a calendar-year plan paid in March, a resignation in February can forfeit a full year of qualifying performance. California courts enforce clear continued-employment conditions.
Where employers disclose payout timing publicly is not established. The likeliest place is the Compensation Discussion and Analysis (CD&A), the section of a public company's proxy statement that explains executive pay, and I have not checked it. A commonly quoted window of Feb 15 to Mar 31 comes only from glossary sources, so treat it as a guess, not a date.
Proves: that a payout-date condition exists. Lies when: the plan says "typically late March" and the actual payment lands later.
ESPP and retirement match vesting
An employee stock purchase plan (ESPP) lets staff buy employer stock through payroll; a retirement match is an employer contribution that may vest with service. I have not researched either for this reference. Keep both as open rows on every employer entry and get them from the candidate rather than guessing.
Best time to recruit someone before vesting
The cheapest time to recruit someone is the day after their next large vest or bonus payout, adjusted for any clawback step-down. How much that timing is worth depends on the curve's shape: a lot on back-loaded or annual-cadence schedules, very little on monthly front-loaded ones.
Clean is not the same as cheap
Amazon at month 24 is the standard example, and it is usually misread. At the end of year two, 20% of an Amazon new-hire grant has vested, against 70% on Google's reported curve. Amazon fills the early gap with sign-on bonuses in the first two years, so by month 24 that sign-on cash has run its course and there is no sign-on clawback left to price. Confirm that against the candidate's own sign-on terms in the conversation.
That makes month 24 clean on cash. It does not make it cheap on equity. 80% of the grant is still unvested, and the next vest delivers 40% at year three. A candidate who leaves just after the year-three vest forfeits 40% of the new-hire grant instead of 80%. Calling at month 24 or waiting a year is a buyout question: can the hiring company cover 80% of the grant, or will the candidate wait?
Google at month 24 is the opposite case. 30% remains, and monthly vesting means each month of waiting buys a sliver. There is no window worth waiting for, so call now.
When to call, by what is still at stake
Sizing the buyout range
The buyout range is two numbers for the same candidate. The high end is what they forfeit if they start on your preferred date. The low end is what they forfeit if they start on the cheapest-exit date you can realistically wait for. Each end adds the same four components:
- Unvested equity on every grant, new-hire and refresh, valued at the current share price for public employers, with a stated assumption for private ones.
- Vested options the candidate is unlikely to exercise within the window.
- Annual bonus forfeited by leaving before the actual payout date.
- Clawback owed on that date: the gross amount if the clause says gross, and zero if the clause is void.
Before the call you hold the employer's shape and dates, not the candidate's grant size. The range is a formula with the candidate's amounts left blank, and the conversation fills them in.
Finding people at the right point on the curve
Where someone sits on the curve depends on when they joined, so a slate built on employer alone is not enough; it needs start dates. Refolk takes that as a plain-English request and returns people with employer and tenure, so you can place each one on the curve before you write the first message.
Pool depth decides how often you can afford to wait for a cheap window.
| Segment (Refolk's index) | Profiles | Ratio |
|---|---|---|
| Senior SWE, US | 182,844 | 6.9x the UK pool |
| Senior SWE, UK | 26,408 | Baseline |
| Staff SWE, US | 33,334 | 5.5 Senior per Staff |
With a pool that deep at Senior level, you can favour people already inside a cheap window and still fill a slate. At Staff level there is one profile for every 5.5 Senior, so you will more often pay the buyout than wait. The UK Senior pool is 6.9 times smaller than the US one, and there timing becomes a luxury.
Can a recruiter ask about unvested equity?
In several jurisdictions with salary-history bans, yes: questions about forfeited equity and deferred pay are expressly allowed even though questions about current or past pay are not. What some of those places forbid is searching public records to uncover a specific person's pay, which is why this reference keeps all research at the employer level.
| Jurisdiction | Forfeiture questions | Records-search limit |
|---|---|---|
| New York City | Value, structure, and documentation allowed | No searching to uncover an applicant's pay |
| Illinois | Letter of aggregate amount, if candidate raises it | Not established here |
| Kansas City | Carve-out names unvested equity; 6+ employees | Not established here |
| Cincinnati | Same carve-out, §804-03(b) | Asking includes searching public records |
| Toledo | Carve-out unverified | Inquiry includes searching public records |
| Cleveland | Forfeiture questions allowed | Not established here |
The detail that matters in practice:
- NYC: employers may ask about the value and structure of deferred compensation or unvested equity and request documentation to verify the applicant's representations. Headhunters are covered.
- Illinois: if the applicant raises it unprompted, the employer may ask for a letter verifying the aggregate amount, from the current employer or the entity that administers the funds, at the applicant's choice.
- Kansas City: the ordinance took effect Oct 31, 2019. Cleveland: its ban took effect Oct 27, 2025.
Forfeiture is the one compensation figure a recruiter may verify in these places, and the statute sets the route.
Where the question is permitted, ask it plainly and offer the documentation route rather than demanding it.
Before we talk numbers, I want to make sure a move doesn't cost you anything you've already earned. If you left on the start date we're discussing, what would you leave behind? For example: - equity that hasn't vested yet, including any refresh grants - vested options you'd need to exercise within a deadline - an annual bonus that pays after that date - a sign-on or retention payment you'd have to repay If it's useful, you can share documentation of those amounts and I'll use it to shape the offer. If a different start date would cut what you forfeit, tell me and I'll work that into the plan.
Use only where the role's jurisdiction permits forfeiture questions; never ask what they earn now or earned before.
How to build a dated forfeiture curve: the procedure
The procedure runs from slate to verified buyout range in eight steps: about two to three hours of desk work for a new public employer, then the conversation. Sourcers own steps 1 to 4, the recruiter owns 5, 6, and 8, and the talent lead owns 7.
Slate to dated forfeiture curve
- List the employers on the slateWrite down every source employer and mark each as public or private (about 15 minutes). Done when each employer's status is recorded.
- Pull the primary filed documentsUse SEC EDGAR full-text search for new-hire RSU forms, offer-letter exhibits, and sign-on agreements (30 to 60 minutes per public employer). Done when you hold a dated exhibit URL for each mechanism found.
- Cross-check against current schedule reportsCompare filings with crowdsourced compensation reports, since exhibits often leave the schedule blank (about 15 minutes). Done when every split has a source and a date and conflicts are noted.
- Check the exercise window at private employersCheck the public holman/extended-exercise-windows list and any published policy (about 10 minutes). If nothing is published, assume 90 days.
- Map bonus payout and clawback datesPut every cliff, vest, payout, and clawback step-down date on one calendar per employer (about 20 minutes). Done when each date carries its source.
- Build the forfeiture curve and cheapest-exit windowPlot forfeiture by resignation date and mark the cheapest exit as the day after the next large vest or payout, adjusted for clawback step-downs (about 30 minutes). Price the buyout range at your preferred date and the cheapest-exit date.
- Check the role's jurisdiction before the callConfirm which salary-history rules apply and whether forfeiture questions are allowed (about 10 minutes). Never search records for an individual's pay.
- Verify forfeiture in the conversationAsk what the candidate would forfeit and, where allowed as in NYC and Illinois, request documentation. Done when the candidate's amounts fill your employer-level curve.
Practitioners disagree on whether the jurisdiction check belongs before step 2. My view: because every step before 7 is employer-level research, it does not trigger the Toledo or Cincinnati public-records rules, which apply when the purpose is an individual's pay. If your team cannot guarantee the research stays employer-level, move step 7 to the front.
Step 1 goes faster when the slate already carries start dates. A tenure-qualified search in Refolk gives you employer and joining date together, which is exactly what steps 5 and 6 need.
- Employer: - Public or private: - New-hire vesting split (Y1/Y2/Y3/Y4), source, source date: - Cliff length, source: - Vest cadence after cliff (monthly, quarterly, annual), source: - Hire years this split applies to (or "unknown"): - Refresh grant schedule, source: - Post-termination exercise window, source (or "assumed 90 days"): - ISO-to-NSO conversion noted (yes/no): - Sign-on clawback period and step-down, source: - Repayment basis (gross or net): - California agreement from Jan 1, 2026? AB 692 conditions all met (yes/no/unknown): - Bonus plan year and actual payout date, source: - ESPP and retirement match vesting: open, ask candidate - Next large vest date: - Next bonus payout date: - Next clawback step-down date: - Cheapest-exit window: - Buyout range formula: unvested equity + unaffordable vested options + forfeited bonus + clawback owed, at preferred date and at cheapest-exit date - Role jurisdiction and forfeiture-question status:
One entry per employer; every row carries a source and the date you checked it, and blanks stay blank.
How walkaway-cost research goes wrong
Most errors come from treating a document as more current or more specific than it is. These are the failure modes I see, with what each looks like and the check that catches it.
- Stale schedule. Google has reportedly used at least five different splits, so a candidate hired in an earlier year may be on a different curve from the one you modelled. Check the candidate's hire year against the date of your source; if they do not match, mark the split unconfirmed.
- Blank exhibit. EDGAR forms show [TBD] where the schedule should be, and the cliff can be 12 to 14 months. Treat a form as proof of the mechanism, not the split, and take the split from a dated schedule report.
- Refresher mismatch. Google refresh grants reportedly vest 25% a year rather than on the front-loaded curve. If you model only the new-hire grant, a tenured employee's walkaway cost is wrong, usually understated.
- Gross versus net. A clause requiring repayment of the gross amount, when the candidate received net pay, means the real cost is higher than the cash they remember receiving. Read the repayment wording before you price it.
- Non-compliant California clawback. A California sign-on clawback signed from Jan 1, 2026 that misses any AB 692 condition is void. Pricing it into a buyout overpays.
- Treating an extended window as extra tax-free time. An extended window still converts ISOs to NSOs at day 91, so the tax cost arrives even though the deadline moved.
- Public-records search on a person. In Toledo and Cincinnati, searching records to find an individual's pay counts as a prohibited inquiry. If a search names the candidate and is about pay, stop.
- Payout-date slippage. Plans often say "typically late March", and the actual payment date is what counts. A start date set to the typical date can fall one day short and cost a full year's bonus.
Keeping each employer entry current
An employer entry is only as good as the date on its sources, so re-check it whenever a new source could change a row rather than on a fixed calendar. The mechanisms drift at different speeds, and each has a trigger worth watching.
- Vesting splits: re-check when a new award form or offer-letter exhibit is filed, and whenever a candidate's hire year falls outside the period your source covers. GE HealthCare's form appearing in two consecutive years' filings is the kind of repeat that lets you mark a split as stable.
- Exercise windows: re-check after a layoff cycle at the employer, because extended windows tend to appear when layoffs are heavy and may not survive them.
- Clawbacks: re-check when stay-or-pay law changes in a state you hire into. AB 692 shows how one statute can void a whole class of terms from a single date.
- Bonus payout dates: record the actual payment date each cycle, and replace any "typical" date with the real one.
- Open rows: ESPP and retirement match vesting stay open until a candidate or a published plan fills them.
When an entry changes, re-run the tenure search in Refolk so the slate reflects who now sits inside the cheapest-exit window.
Before the first call
- Every vesting split on the slate has a named source and a source date.
- Each candidate's hire year falls inside the period the split's source covers, or the split is marked unconfirmed.
- Refresh grants are modelled separately for anyone past their first year.
- Private employers have a documented exercise window or a recorded 90-day assumption.
- Every clawback has been read for gross or net wording and checked against AB 692 where it applies.
- Bonus payout dates are actual dates, not typical ones.
- The cheapest-exit window and the buyout range at both dates are written down.
- The role's jurisdiction has been checked and forfeiture questions are confirmed as allowed.
- No search in the file names a candidate in connection with pay.
- ESPP and retirement match vesting are listed as open questions for the conversation.
Questions practitioners ask
Can a recruiter ask a candidate about unvested equity?
Often yes, even where salary-history questions are banned. NYC lets employers, including headhunters, ask about the value and structure of unvested equity or deferred pay and request documentation. Kansas City and Cincinnati carve out unvested equity, Cleveland allows forfeiture questions, and Illinois allows a verification letter if the candidate raises it unprompted. What you may not do in places like NYC, Cincinnati, and Toledo is search records to uncover a specific person's pay.
How do I calculate an unvested equity buyout for a candidate?
Add four things on the planned start date: unvested equity across the new-hire grant and every refresh grant, vested options the candidate likely cannot afford to exercise in the window, any bonus lost by leaving before the payout date, and any clawback owed. Price it twice, once for your preferred start date and once for the cheapest-exit date, to get a range. Get grant sizes from the candidate, never from records about them.
Should a sign-on bonus cover forfeited RSUs in full?
Cover what the candidate actually loses on the start date you agree, not the headline grant. On a back-loaded schedule like Amazon's 5/15/40/40, starting just after the year-three vest halves the forfeited share of the new-hire grant, from 80% to 40%. Agree the date before the number. In California, check that any clawback on the new sign-on meets every AB 692 condition for agreements from Jan 1, 2026.
Does a candidate lose their bonus if they resign before the payout date?
Often yes, if the plan requires employment on the payout date. Under a calendar-year plan paid in March, resigning in February can forfeit a full year of qualifying performance, and California courts enforce clear continued-employment conditions. Set the start date after the actual payment date, not the typical one, because a plan that says payment is typically in late March pays on whatever day it actually pays.
What is the best time to recruit someone before vesting?
Just after their next large vest or bonus payout, adjusted for clawback step-downs. Timing matters most on back-loaded or annual-cadence schedules: an Amazon new-hire grant on 5/15/40/40 has 80% unvested at month 24 but 40% after the year-three vest. On Google's reported monthly, front-loaded curve there is no single cheap day, so waiting buys very little.
What happens to a candidate's stock options when they leave?
They usually have 90 days to exercise vested options or lose them; private-markets data put 81% of terminated options on a 90-day window. After day 90, ISOs are taxed as NSOs. Employee exercise rates fell to 23%, so most departing option-holders lose vested options as well. Count vested-but-unaffordable options in the walkaway cost, and check whether the employer publishes an extended window.
Try it on the search you came here for
Stop building boolean strings. Just describe the person.
Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.
01Describe them
One plain sentence. Role, city, stack, stage, whatever matters to you.
02I read the web live
GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.
03You read the shortlist
Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.
- Staff backend engineers in NYC who shipped Rust in production
- Series A fintechs in SF under 50 people, growing headcount this year
- Maintainers of fast-growing Rust web frameworks on GitHub
- No boolean, no filters, no seat to buy. One box.
- Read at search time, so a profile updated yesterday counts today.
- Every step visible as it runs, every name with its reason.
500 free credits on sign-up. No card, no demo call. See real searches.