The Retention-Cliff Founder Queue: Deal Histories to Reach-Out Dates

This playbook turns 10 to 50 acquirers in your thesis area into a monthly queue of 20 to 40 acquired founders and senior operators. Each person in the queue gets a reach-out-by date and a stated basis for that date. It is for early-stage investors, platform and talent partners at funds, and angels who want to meet second-time founders before they raise, and before other funds find them. Most of the dates will be estimates, so every one carries a label saying how it was worked out.

Why acquired founders belong on a calendar

Acquired founders leave on a schedule set by their own deal contracts. That makes the end of a retention period one of the few founder events you can see coming months ahead. It matters because repeat founders take a large share of early-stage capital and can close quickly once they decide to raise.

The evidence for the opportunity:

• Share of capital. Carta's repeat-founder data found that more than 50% of Seed + Series A capital went to repeat founders in the period it measured.
• Speed. No named dataset splits time to close by founder type. The best available figure is a practitioner estimate: a serial founder running a SAFE party round with dense proof and parallel meetings can close in three to six weeks.
• Price. Forum VC reports that multi-time founders, especially those with past successful exits, often raise at valuations between $15M and $20M+.
• Where they go. Academic work on startup acquisitions as a hiring strategy notes empirical evidence that acquihired employees who leave the acquirer are likely to join a new startup.

Put together: if you first hear about the company when the round is announced, the round was decided weeks earlier in meetings you were not in. The retention clock gives you a reason to be known to the founder well before that, so you are not arriving cold.

The queue is a list of people still inside an acquirer, or recently out of one, dated by when their contractual reasons to stay run out. It is not a stealth-founder detector. By the time someone lists "Stealth" as their employer, the queue has either done its job or missed.

When do founders leave after an acquisition?

Founders leave in two waves: a heavy first-year exit, then a long tail. In the best founder-only study, about 30% of acquihired founders were gone within two years and nearly 50% by year four. Only 27% remained at year five.

How to read the table:

• Kim, "Predictable Exodus". This is the strongest source on timing, but it covers all acquired workers, not just founders. Use it to see that departures peak in year one, not to estimate founder rates.
• "Does acqui-hiring pay off?" (Small Business Economics). This study is founder-only. It is a hand-collected dataset of every acquihire by Facebook and Google from their first acquisition up to 2019. Of those founders, 77% left during the observation window, and average tenure was 3.7 years.
• Bargeron and Agashi. Both measure CEOs at 24 months, so they bracket the founder-CEO case in your queue: 44% and 37% gone.

The practical consequence: a large group of founders stays past the first equity cliff. With 30% gone at two years and a mean tenure of 3.7 years, a queue keyed only to the 12-month cliff misses most of the people it should catch.

No public source gives clean founder-only curves at 12, 18, 24 and 36 months. If you need a value between the published points, interpolate and label it as interpolated.

Figures to keep out of your model

• "52% within two years / 71% within 18 months." These are widely repeated, come from a blog aggregator, and name no sample. They also contradict each other: a cumulative share cannot fall from 71% at 18 months to 52% at 24 months.
• "Only about 10% still there after two years." This claim, from the author of Practical Founders, is about SaaS founders and rests on "hundreds of transactions" with no named sample. It is far harsher than the academic figures. Read it as one practitioner's experience, not a benchmark.

How long do acquihire and product-acquisition retention periods run?

In an acquihire, expect a cash stay bonus paid 6 to 24 months after close and retention equity that vests over three or four years with a one-year cliff. Product acquisitions have no published norm, so any window you assign to one is inferred and must be labelled that way.

Definitions, in one line each:

• A stay bonus is cash paid to an employee who is still employed on a set date after closing.
• A cliff is the date before which no equity vests at all. A one-year cliff means nothing vests until the first anniversary.
• An acquihire is a deal whose main purpose is to hire the team, not to keep running the product.

a16z's model term sheet uses retention equity that vests over a standard 4-year structure. On how much is at stake, SaaStr says acquirers try to put 30% to 40% of the founders' consideration at risk, "if they can get away with it". The more consideration that depends on staying, the harder it is to leave before the last tranche. You will rarely see the actual number, so treat it as context rather than an input.

The public record does not establish two things. There are no published figures on reverse vesting of founder shares in product acquisitions, and there is no published distribution of earn-out lengths. If you have seen term sheets from your own portfolio exits, those are your best local evidence. Note them in the row as "inferred from comparable deal".

Telling an acquihire from a product acquisition

The two retention shapes differ, so the label changes the date. In an acquihire, old equity is cancelled and a fresh grant starts at close, so the windows above apply cleanly. In a product acquisition the founder may be tied to an earn-out or an integration milestone of unknown length.

The three signals, and how each one can mislead:

• Product shut down at close. This shows the buyer paid for the team. It misleads when the product was wound down months later after a failed integration. That is a product acquisition that died, and its retention clock may differ.
• "Team joins" language in the announcement or a founder post. This shows the buyer framed the deal as a hire. It misleads when the wording is just softer PR for a product deal, so check what happened to the product.
• Asset-purchase structure. This shows the buyer did not take the company whole, which is common in acquihires. It misleads because product tuck-ins use it too, so it needs one other signal alongside it.

My standard: call a deal an acquihire only if the product was shut down at close, or if at least two signals agree. Quote the evidence in the row.

Where do close dates come from, and how far can you trust them?

For most startup deals by large acquirers, no SEC filing gives a close date. A Form 8-K under Item 2.01 is required only for acquisitions above 20% significance, or 10% for asset purchases. So dates mostly come from announcements and founders' own posts, and every date needs an evidence tier.

What each Form 8-K item gives you:

• Item 1.01 (material definitive agreement). It requires only "a brief description of the material terms and conditions of the agreement or amendment that are material to the registrant." Filing the agreement itself is encouraged but not required, so retention terms rarely appear.
• Item 2.01 (completion of acquisition). This is the dated close. The SEC's Financial Reporting Manual says it is due within 4 business days of closing, but only above the significance thresholds. For individually insignificant acquisitions, "The registrant has no Item 2.01 Form 8-K reporting obligation."
• Item 8.01 (other events). Registrants may choose to report insignificant acquisitions here. Some do. Most startup tuck-ins never produce a dated close in any filing.

Private acquirers file none of this. Coverage from Item 5.02, proxy statements and non-US registry filings is not publicly established, and neither is the share of private-target deals that any public record covers. If an acquirer on your list is outside the US, check its local registry yourself and record what you find. Do not assume it is complete.

Finding the people: why deal lists beat headline search

Find people from the acquirers' deal lists, not from profile headlines, because most acquired founders do not advertise their exit. Refolk's index holds only 228 US Director/VP profiles whose headline says "founder acquired by", against 9,457 US profiles with a Founder title and "stealth" in the headline.

The two pools are different populations at different points on the clock. The stealth pool is mostly people who have already left: their window is open and every other fund can see them. The pool this playbook targets is still inside the acquirer, and headlines barely show it. Sample titles look like "Co-Founder (Acq. by SAI360)", but most acquired founders simply take the acquirer's title and drop the history.

The stealth pool is still worth knowing, as a check on where people go after leaving:

Two cautions on this table. First, the percentages come from samples of 25, so treat them as a rough direction, not a precise rate. Second, in the US sample roughly three in four "stealth" founders did not list a stealth-named employer. The word was in the headline only, and the US sample also contained an orthopaedics professor and a General Partner.

Working from a deal list back to people is the slow part. An acquired company's name often no longer appears on its founders' profiles, and the senior operators who joined with them are harder still to find. Refolk lets you ask for that cohort in plain English across LinkedIn, GitHub and the open web. It returns named people with their current title and employer, which is the field the dating step depends on.

The procedure: acquirer list to monthly queue

Seven steps take you from an acquirer list to a scored, dated queue. Setup time grows with the list: 1 to 2 hours per acquirer for deal histories, about 10 minutes per deal to classify, and about 15 minutes per company to find the people. After that, the monthly rerun takes about two hours of investor time. The order is my own synthesis; no single published source documents the full sequence.

Step 4 is where Refolk saves the most time: one plain-English request per acquirer and date range replaces a company-by-company search. Check every returned person against the deal row before you date them.

Setting the reach-out-by date

The lead time before each window is my heuristic, not a sourced figure. Kim's data shows departures peak in year one, which argues for starting before the first cliff, not at it. My defaults:

• Acquihire, confirmed close. First touch by month 6 after close, when the shortest stay bonuses can end, well ahead of the 12-month cliff.
• Still there after the cliff. Re-queue ahead of the 24-month mark and again ahead of full vest at 36 to 48 months. The Google/Facebook tail is large enough to justify it.
• Product acquisition or unconfirmed close. Set a relationship touch about 6 months after the announcement, tier "inferred", and widen the window instead of guessing a tighter one.
• Covenant still binding. The retention date still sets when you reach out, but the conversation is about what they could build outside the restricted field, or after the covenant ends.

Scoring the queue

Practitioners disagree on order here: some score acquirers before finding people, which saves time on long lists. If your list runs past 50 acquirers, rank them first by how often their acquired teams leave to found companies, then resolve people only for the top half.

person | role at acquired co | acquired co | acquirer | deal type | deal-type evidence (quote) | announced | closed | close source URL | evidence tier | current title | current employer | next window | may build | may hire from | score | reach out by | owner

SCORE (0 to 12)
Thesis fit:        0 none | 1 adjacent | 2 inside | 3 core
Window proximity:  0 12+ months away | 1 6 to 12 months | 2 under 6 months | 3 open now
Evidence tier:     1 inferred | 2 announcement | 3 filing
Covenant freedom:  0 bound in field | 1 bound, adjacent field open | 2 free
Motion:            0 none | 1 title or employer change since close

Cut: take the top 20 to 40 by score; break ties on earliest reach-out-by date.

Covenants: may this founder build, and whom may they hire?

A founder whose retention has ended may still be barred from their old field. A non-compete tied to a sale of goodwill survives even in California and typically runs 2 to 5 years, and non-solicits of 2 to 3 years limit whom the founder can hire. Where the deal was a goodwill sale, the date that actually binds can be the covenant expiry, not the vest date.

What the sources support:

• Federal rule. The FTC's non-compete rule was struck down by a court. The agency dismissed its appeals and removed the rule from the Code of Federal Regulations. To re-check the current position, look up the rule's status in the Code of Federal Regulations, not in commentary.
• California. Section 16600 broadly voids restraints on a lawful profession, trade or business. Section 16601 allows a covenant in specified sales of goodwill, ownership interests or operating assets. The covenant must name a geographic area, and it lasts only while the buyer runs a similar business there.
• Duration. One practitioner table gives 2 to 5 years as the typical sale-of-business duration, "routinely upheld". Non-solicits typically last 2 to 3 years and often outlast the non-compete.
• Partial sales. Samuelian v. Life Generations Healthcare, 104 Cal. App. 5th 331, concerns a partial sale, and the two sources I have describe its holding differently. Do not rely on either reading without counsel.
• UK, EU and Israel. These were not covered in my research. Ask local counsel before you assume anything.

The deciding question is structure. Was this a sale of goodwill or ownership interests, or was it a hire with cancelled equity? Put the answer and its source in the row. This is a sourcing heuristic, not legal analysis: it tells you which conversation to have, not what the founder is allowed to do.

Being allowed to leave is not the same as being allowed to build.

How this goes wrong: failure modes and false positives

The queue fails in seven predictable ways. Most of them come from treating an estimate as a fact. Each has a cheap check, and the check belongs in the monthly rerun, not just the first build.

1. Bad statistics.
   • What happens: aggregator figures such as "71% by 18 months" and "52% by 24 months" enter the model and contradict each other.
   • Check: a cumulative departure curve can only rise. Drop any figure that has no named sample.
2. Wrong date.
   • What happens: the announcement date is used as the close date. Below the 20% significance line there is no Item 2.01 Form 8-K, so this is the default error.
   • Check: every close date has a source URL that states a closing. Without one, write "unconfirmed" and tier "inferred".
3. Acquihire mislabelled.
   • What happens: a product acquisition that was wound down later looks like an acquihire, and gets acquihire windows it may not have.
   • Check: was the product shut down at close, or months later? Only shutdown at close counts as an acquihire signal.
4. Wrong cliff date.
   • What happens: the 12-month cliff is treated as the typical departure date. People leave before vesting finishes, and Kim found departures are highest in year one.
   • Check: treat the cliff as the cutoff for the first leavers. Sanity-check the queue against the roughly 30% two-year figure: if your model has most founders out by month 13, it is wrong.
5. Keyword noise.
   • What happens: "Stealth" matches people who are not founders. The US sample in Refolk's index included an orthopaedics professor and a General Partner.
   • Check: the current employer is new and dated after the person left the acquirer.
6. Ignored covenants.
   • What happens: a founder who is free to leave in California is still barred from their old field by a §16601 covenant, and your pitch assumes they can compete.
   • Check: record the deal structure (sale of goodwill or hire) before the person reaches the queue.
7. Survivorship bias.
   • What happens: YC's 5.1-year average gap between a founder's appearances is used as the time from leaving an acquirer to starting a new company. It counts only founders who came back to YC.
   • Check: do not use it as a departure-to-company interval. No published dataset measures that interval. Also note how the YC figures are spread: nearly 30% of returns came within two years, while 61 came after a decade or more. An investor who waits for the average misses the fast group.

Keeping the queue current each month

Rerun the queue every month: add new deals, upgrade evidence tiers, refresh each person's current role, and log when people actually left. The log matters most. Public founder-only departure curves are not established, so your own record of predicted versus actual departures becomes the best benchmark you have.

The monthly loop:

1. Add new deals. Re-check each acquirer's press room and its Item 8.01 filings for new closes.
2. Upgrade tiers. When a founder post or filing confirms a close, move the row from inferred to announcement or filing and recompute the windows.
3. Refresh current roles. Re-run the people step for everyone whose window is open or under six months. Rerunning the same Refolk request each month shows title and employer changes without manual checks, profile by profile.
4. Retire rows. Move out anyone who has joined a new startup (record where and when) and anyone who has drifted out of your thesis.
5. Log outcomes. For each departure, record months since close, deal type and evidence tier. After a few cycles, compare your local curve with the table above and adjust your lead times.

Subject: [Acquired company] and the [problem space] work at [Acquirer]

Hi [First name],

I invest in [thesis area] at [Fund]. I have followed what [Acquired company] was building and how the team has carried it into [Acquirer], and I wanted to introduce myself well before you need anything from me.

No ask today. If you ever want a sounding board on [specific problem], or a second opinion on people in the space, I am glad to spend 30 minutes.

[Your name]