# The Founder Talent-Pull Read: Magnet, Ordinary, or Hiring Alone

*Rebuild a startup's first-year roster from public profiles, score it on five dimensions, and reach a Magnet, Ordinary, or Hiring Alone verdict a second partner would match.*

- Canonical URL: https://www.refolk.ai/guides/founder-talent-pull-read
- Pillar: Investing and deal sourcing
- Format: Framework
- Published: 2026-10-19
- Last reviewed: 2026-10-19
- Reading time: 23 min
- Keywords: how to evaluate a founder's ability to recruit, founder talent magnetism due diligence, assessing early hires before investing, startup early employee retention red flags, diligence on a founding team's first hires

## Key takeaways

- Early joiners are the non-founder employees hired in a startup's first year, and research using worker deaths as a natural experiment finds that losing one shrinks the firm for at least ten years, while losing a later joiner causes only a small, temporary dip.
- Carta's benchmark puts cumulative departures at 18.0% within 12 months of joining, so on a roster of five early joiners about one departure is normal and two or more is a warning.
- Acquired workers leave in their first year at 33% against 12% for regular hires, a 2.75x gap, so a team that came with an acquisition has to be removed from both the retention score and the pull score.
- In Refolk's index, 16% to 24% of sampled 'Founding Engineer' profiles turned out to belong to founders or CTOs, so every founding title has to be checked against the company page before it counts as a joiner.
- A roster made up entirely of former colleagues predicts higher survival odds and slower early growth, so it counts as low survival risk rather than as evidence that the founder is a magnet.
- Refolk's index holds 5.5x more US founding-engineer profiles than founding-sales profiles, so a slow-to-fill founding GTM role is partly a supply problem and needs its own baseline.

This is the read I run before backing a seed or Series A founder. It answers one question: do strong people actually follow this founder, and do they stay? It is for early-stage investors, platform and talent partners, and angels who need that answer from public evidence instead of from the pitch. You get a five-dimension scoring model, baselines to score against, and a verdict of Magnet, Ordinary, or Hiring Alone that a second partner scoring the same roster should reach independently.

## What does the founder talent-pull read measure?

The read measures whether people with something to lose chose to join this founder in the company's first year, and whether they stayed. The input is the first-year roster, rebuilt from public profiles. The output is a score on five dimensions and one verdict.

### Why the first year

Choi, Goldschlag, Haltiwanger and Kim (NBER Working Paper 28417, later published in the Review of Economics and Statistics) define early joiners as non-founder employees in a startup's first year. They used the premature death of workers as a natural experiment. Losing an early joiner had a large negative effect on firm size that lasted at least ten years. Losing a later joiner caused only a small and temporary decline. Compared with losing a founder, losing an early joiner had a smaller effect on whether the firm died but a similar effect on how large it grew.

So the first-year roster is where much of a company's capacity to deliver its plan gets built. If you are funding a hiring plan, this cohort is the best evidence of whether the founder can carry it out.

### Why not the pitch

Investor checklists name the trait but do not measure it. The Angel Investors Network diligence checklist says "Evaluate the team's ability to recruit" and gives no method. Hustle Fund's founder diligence guide, which includes Azella Perryman's questions, starts with founder interviews and back-channel references. Those steps are useful. But they give you the founder's account, filtered through references the founder chose. I run this read first, from public records, and use interviews and references afterwards to test what it found.

### Terms used throughout

- **Early joiner:** a non-founder employee who started within 12 months of the founding date.
- **Clean roster:** the early joiners left after removing cofounders, acquired teams, contractors, advisors, interns and unresolved stealth profiles.
- **Followership share:** the share of the clean roster who worked at the same employer as a founder before joining.
- **Stranger:** an early joiner with no shared employer and no shared school with any founder.
- **Give-up:** what a joiner left behind to join: seniority, the size and stability of their employer, and their location.
- **Open-role drag:** how long the roles this round pays for have been open, compared with medians for that function.

The read applies to companies with at least three non-founder hires. Carta reports that the median seed team is four employees, so three non-founder hires is close to the smallest roster that can support a score.

## The five dimensions and how to score each

Each dimension scores 0, 1 or 2, so the total runs from 0 to 10. The bands are deliberately coarse, because coarse bands are what let two partners agree (see the verdict section). For each dimension, here is what it proves and what it looks like when it misleads.

| Dimension | 2 points | 1 point | 0 points |
|---|---|---|---|
| Followership share | Some, not all, are prior colleagues | All are prior colleagues | No prior colleagues |
| What joiners gave up | A High give-up within hires 1 to 5 | Best give-up is Medium | All give-ups Low |
| Stranger pull | A stranger with Medium or High give-up | Strangers, all Low give-up | No strangers |
| Early-joiner retention | None gone within 12 months | One gone | Two or more gone |
| Open-role drag | No plan-critical role past median | Past median, not re-posted | Past median and re-posted |

### 1. Followership share

**What it proves:** people who saw the founder at work, with nothing to gain from being polite, chose to follow. That is the hardest form of reference to fake.

**How it lies:** a roster made up entirely of former colleagues looks like maximum pull, but it is a different signal. Sarada and Tocoian (ILR Review) measured prior professional connections among founding employees in Brazilian matched employer-employee data. Previously connected founding employees had higher survival odds but slower early growth. Ties close information gaps. Hires from outside the circle bring, in the authors' words, "a more varied resource set". So a 100% share scores 1, not 2. It means lower survival risk, not a magnet.

> **Rule:** Count only ties formed at a shared employer
>
> A shared school goes in the stranger column, not the followership column. Followership means a joiner saw the founder work. A classmate or relative with no prior jobs proves nothing about that.

Alongside the share, compute the shared-employer concentration from a related study:

**HHI = Σj Nj(Nj − 1) / [N(N − 1)]**

Here N is the number of clean joiners and Nj is the number who came from previous employer j. The result is the chance that two joiners picked at random came from the same previous employer. A value near 0 means the roster came from many places. A value near 1 means one employer supplied nearly everyone. That can be a genuine lift-out of a trusted team, or it can be an acquisition you have not caught yet (see the failure modes). HHI is not scored. It is a flag that tells you where to look.

### 2. What joiners gave up

**What it proves:** a joiner who took a real cost to join made a costly, believable bet on this founder.

The cost is measurable in aggregate. Sorenson, Dahl, Canales and Burton (Organization Science) found that employees hired by startups earn roughly 17% less over the following ten years. In Danish registry data, hires at small startups earned at least 5% less at the time of hiring than comparable peers. Those who joined as one of the first five employees took the largest long-term pay penalties, and by about employee 40 the penalty had disappeared.

Band each person:

- **High:** left a senior title or a large, established employer, and joined as one of hires 1 to 5.
- **Medium:** left a stable role at a larger employer but joined after hire 5, or left a senior role at a peer startup.
- **Low:** came from study, a gap, freelance work, or a startup of similar size and stage.

Relocating to join can raise a person by one band, and never by more than one.

**How it lies:** titles inflate, and a move to a 12-person company looks like a sacrifice even when it is the joiner's best available offer. How often public titles overstate or understate seniority is not established publicly, so rate the employer and the scope of the role as well as the title.

### 3. Stranger pull

**What it proves:** the founder can persuade people who had no prior reason to trust them. This is the skill the round is paying for, because the next ten hires will mostly be strangers.

**How it lies:** relatives and classmates look like strangers if you only check employers, which is why the definition excludes shared schools. Agency placements can also look like organic pull. A stranger who joined from a recruiter at a Low give-up proves little, which is why that case scores 1.

> A team of old colleagues proves trust. A stranger who took a pay cut proves pull.

### 4. Early-joiner retention

**What it proves:** the reality of the job matched what the founder sold. People who left within a year found something they had not been told about.

Count clean joiners who left within 12 months of their own start date. Carta's benchmark for startup employees with equity shows 18.0% cumulative departures at one year. With a roster of five that works out to roughly one person, which is why one departure scores 1 rather than 0. Carta also notes that startups with 25 employees have typically already had at least one employee leave.

**How it lies:** Carta splits that 18.0% into 11.8% voluntary and 5.4% involuntary departures. A founder who quickly removes a bad hire is not showing weak pull, but public profiles rarely show which kind of departure it was. Treat that as a question for references, not a reason to change the score. Departures from an acquired team are the bigger distortion. Remove them entirely.

### 5. Open-role drag

**What it proves:** whether the hiring the plan depends on is happening now, at the speed the plan assumes.

Score only plan-critical roles, meaning the ones the round's use of funds names. Age each one from its first appearance in archived copies of the careers page. Compare that age with the function's time-to-first-fill median in the baselines below.

**How it lies:** "always hiring" posts look like roles that never get filled. Founding GTM roles draw from a far thinner pool (see the baselines), so a slow founding sales hire is partly a market problem rather than a founder problem.

#### Followership against stranger pull

Horizontal axis runs from Few or weak strangers joined to Strangers with real give-up joined. Vertical axis runs from No prior colleagues followed to Prior colleagues followed.

| Quadrant | What it means |
| --- | --- |
| Closed circle | Good survival odds; test whether the next hires can come from outside the network |
| Magnet candidate | Confirm retention and open-role drag before calling it |
| Hiring alone | The hiring plan relies on a skill the founder has not yet shown |
| Pitch pull | Ask why former colleagues did not follow; weight retention heavily |

*The two tie dimensions separate a founder who is trusted inside a circle from one who can recruit outside it.*

## Baselines to score against

Score against three sets of baselines. Departure rates set the retention band. Founding-title supply tells you how to read titles and GTM drag. Hiring-speed medians set the open-role band. Every one of them has been revised over time, so treat them as values to re-check, not constants.

### First-year retention baselines

| Population | 12-month departure | Source |
|---|---|---|
| Carta startup employees | 18.0% | Carta attrition benchmark |
| Regular hires (Census data) | 12% | Kim |
| Acquired workers (Census data) | 33% | Kim |
| Acquired vs regular | 2.75x | Derived |

Carta's figures cover more than a million employees with equity who started since 2016. Cumulative departures reach 6.9% at 6 months, 37.9% at 2 years and 50.7% at 3 years. Kim's Census study covers 4,000 high-tech startup acquisitions and 350,000 employees. Carta's published median startup tenure has stayed around two years across its samples (2.0 years in an earlier cut of 185,000+ employees, 2.2 years in a later one). Check the current figure on Carta's attrition page before you cite it. A split of these rates by funding stage from a neutral primary source is not established publicly.

### Founding-title supply in Refolk's index

| Title, market | Current profiles | Exact title in sample | Ratio |
|---|---|---|---|
| Founding Engineer, US | 4,559 | 21/25 (84%) | Baseline |
| Founding Engineer, UK | 683 | 19/25 (76%) | US is 6.7x UK |
| Founding AE/Sales, US | 826 | 22/25 (88%) | Engineering is 5.5x GTM |

In Refolk's index of professional profiles, no company appeared more than twice in any 25-profile sample. These titles are spread across many companies, not concentrated in a few, so a "Founding" title tells you about the label more than about any one employer's culture.

**5.5x - US founding-engineer profiles for every founding-sales profile**

4,559 against 826 current titles in Refolk's index, so a founding GTM role needs its own drag baseline.

### Open-role baselines

| Measure | Business | Technical |
|---|---|---|
| Time to hire | 30 days | 40 days |
| Time to first fill | 56 or 60 days | 75 or 76 days |
| First to last interview | 14.4 days | 17.9 days |

These are medians from Ashby's State of Startup Hiring report, as given in secondary write-ups. The write-ups disagree on time to first fill, which is why two values appear. I use the lower figure (56 business, 75 technical) and write it on the sheet, so that both scorers use the same line. The same write-ups give first-fill medians by seniority: 52 days junior, 63 mid and 71 senior. One hiring benchmark puts senior roles at roughly 37% longer to fill than junior ones. For a senior plan-critical hire you may use the 71-day senior median instead, as long as you record that you did. Figures specific to seed and Series A are not established publicly.

## How to run the read, step by step

The procedure takes roughly five to six hours, and nearly all of that is rebuilding and cleaning the roster. Run it before the founder interview so the interview can test the evidence, not create it.

#### The talent-pull read

1. **Fix the founding date** - Take incorporation or the first hire's start date, whichever is earlier, and add twelve months. About 15 minutes; done is a dated month-12 cutoff.
2. **Rebuild the first-year roster** - Find every non-founder whose profile shows a start date before the cutoff, each with a source link. One to two hours; done is at least three dated people.
3. **Exclude non-joiners** - Remove cofounders, acquired teams, contractors, advisors, interns and unresolved stealth placeholders, logging a reason for each. About 30 minutes; done is a clean roster.
4. **Tag prior ties** - Mark shared employers with any founder, and note shared schools separately. About an hour; done is a followership share, a stranger count and the HHI.
5. **Score what joiners gave up** - Compare previous title, employer size and location, and record each hire number. About an hour; done is a High, Medium or Low band per person.
6. **Check early-joiner retention** - Count clean joiners gone within 12 months of their own start date and compare with the 18.0% baseline. About 20 minutes.
7. **Measure open-role drag** - Snapshot the careers page, find each plan-critical posting's first archived appearance, and age it against the function median. About 30 minutes.
8. **Get a blind second score** - A second partner scores the same clean roster without seeing your scores. Done is matching verdicts.

### Rebuilding the roster without the founder's help

Step 2 is where most reads fail, because the obvious method is to ask the founder for an org chart. Don't. Search public LinkedIn and Crunchbase records for the company name, then for each founder's previous employers to catch people who followed them. Then search for "stealth" and placeholder employers with start dates around the founding month. Record the start month and the link for each person. A name with no dated source does not go on the roster.

This is the part that eats the hours: one query per founder, per previous employer, per spelling of the company name. With [Refolk](/) you describe the cohort in plain English, such as people who joined a given company in its first year and previously worked with its CEO at the same employer, and get the matching profiles back in one pass. Then do the cleaning by hand.

For give-up scoring, it helps to see what a High give-up looks like across the market before you band the person in front of you:

I ran this search: `Founding account executives in New York who left an enterprise sales role at Salesforce or Datadog for a seed-stage company` - [see the full result list](https://www.refolk.ai/s/eyf5bafsqw).

*Returns founding sales hires with their previous enterprise employer and title, a reference set for what a High give-up GTM joiner looks like.*

### The scoring sheet

Copy the sheet below once per company. Give the second scorer a blank copy with the clean roster filled in and the scores left empty.

**Talent-pull scoring sheet**

```
Company:
Founders:
Founding date (earlier of incorporation or first hire):
Month-12 cutoff:
First-fill medians used (business / technical / senior):

ROSTER - clean early joiners only
Name | Start month | Source link | Hire no. | Previous employer | Previous title | Employer size | Shared employer with founder (Y/N, which) | Shared school (Y/N) | Relocated (Y/N) | Give-up (High/Medium/Low) | Left within 12 months (Y/N, month)

EXCLUDED - with reason
Name | Reason (cofounder / acquired team / contractor / advisor / intern / unresolved stealth) | Evidence link

SCORES (0, 1 or 2)
Followership share: __   (prior colleagues __ of __)
Shared-employer HHI: __   (flag only, not scored)
What joiners gave up: __
Stranger pull: __   (strangers __ of __)
Early-joiner retention: __   (gone __ of __; baseline 18.0%)
Open-role drag: __   (oldest plan-critical role __ days; median used __)
TOTAL: __ / 10

VERDICT: Magnet / Ordinary / Hiring Alone / Insufficient evidence
Gates checked: stranger pull = __ ; retention = __
Questions this raises for references:
```

*One roster row per clean early joiner; fill the scores last and hand a blank copy to the second scorer.*

## Turning five scores into a verdict

Add the five scores, apply two gates, and read off one of three verdicts. The gates exist because a high total can hide the two failures that matter most for staffing a plan: nobody from outside the circle joined, or the people who joined are already gone.

| Verdict | Rule | What it means for the round |
|---|---|---|
| Magnet | Total 8 to 10, stranger pull 2, retention 1 or 2 | The hiring plan is credible as written |
| Ordinary | Anything not Magnet or Hiring Alone | Plan is fundable; budget for slower hiring |
| Hiring Alone | Total 0 to 3, or stranger pull 0 and retention 0 | Staffing the plan is the main risk; diligence it as one |

**Short rosters.** If cleaning leaves fewer than three joiners, look at why. If the people removed are contractors or an acquired team doing the work early joiners would normally do, the founder is filling seats without attracting people, and Hiring Alone is the honest verdict. If the roster is short because profiles are missing or in stealth, record "Insufficient evidence", move to references, and run the read again once profiles update.

### Why the bands are coarse

Fine-grained judgements do not survive two raters. In one occupation-coding study, non-expert raters reached a Krippendorff's alpha (a measure of agreement between raters where 1 is perfect and 0 is chance) of just 0.093 on fine-grained judgements, rising to 0.52 on obvious cases. Three-level bands push the call toward the obvious end of that range. That is why each dimension has three levels and not a 1-to-10 scale.

> **Rule:** The verdict needs a blind second score
>
> A second partner scores the same clean roster without seeing your sheet. If the verdicts differ, or any dimension differs by two points, go back to the evidence for that dimension; never average the two scores.

A one-point disagreement on a single dimension is normal and usually sits on the give-up band. Settle it by agreeing on that person's employer size and hire number, not by splitting the difference.

## How the talent-pull read goes wrong

Almost every false positive comes from counting someone as a joiner who did not choose to join, or from counting a tie or a sacrifice that was not real. Run the exclusion pass before any scoring, and check each of the failure modes below against the clean roster.

#### The exclusion pass

1. **Raw roster** - Every non-founder profile with a start date before month 12
2. **Strip cofounders** - Drop anyone listed as founder or CTO on the company page or cap table
3. **Strip bought teams** - Drop cohorts that share a start date and an employer tied to an acquisition
4. **Strip non-employees** - Drop contractors, advisors, interns and freelance or agency lines
5. **Resolve stealth** - Date placeholder employers from archived snapshots, or drop them
6. **Clean roster** - The early joiners every dimension is scored on

*Every dimension is scored on the clean roster, so each person removed here is a false positive avoided.*

### 1. Followership inflated by family and friends

**False positive:** a high prior-tie share made up of classmates or relatives with no previous jobs. **Check:** count only ties formed at a shared employer, which is how Sarada and Tocoian define them. Classmates move to the stranger column, where a Low give-up keeps them from inflating that score too.

### 2. Title contamination

**False positive:** cofounders and CTOs with "Founding Engineer" or "Founding AE" titles counted as early joiners. That inflates followership (they obviously worked with the founder), give-up and retention all at once.

**16-24% - Sampled "Founding Engineer" profiles that carried CTO or founder titles**

In Refolk's index, 4 of 25 in the US sample and 6 of 25 in the UK sample.

**Check:** confirm the person does not appear as a cofounder on the company page or on any cap table you have seen.

### 3. Acquired teams counted as joiners

**False positive:** a cohort that arrives in one month from a single employer looks like a founder pulling a whole team. An HHI close to 1 is the usual sign. **Check:** look for news of an acquisition and for identical start dates.

> **Watch out:** Remove acquired cohorts from both scores
>
> Acquired workers never chose the employer, which is why 33% of them leave in the first year against 12% of regular hires. In the Dialpad case, more than 70% of roughly 40 employees left Yahoo after the acquisition. Leave them in and you will overstate pull and understate retention at the same time.

### 4. Contractors counted as employees

**False positive:** a profile that lists a freelance practice or an agency alongside the startup. **Check:** remove anyone whose relationship is consulting, advisory or agency-based. Carta's tenure benchmark excludes consultants and advisors, so including them makes your retention figure incomparable with the 18.0% baseline.

### 5. Give-up overstated

**False positive:** a move to a 12-person company scored as a sacrifice when the person joined late enough that the pay penalty had mostly faded. The penalty is largest for hires 1 to 5 and disappears by about employee 40. **Check:** record the hire number for every joiner, and keep High for hires 1 to 5 only.

### 6. Stealth and placeholder employers

**False positive or false negative:** a joiner hidden behind "Stealth" or a placeholder employer. One UK sample in Refolk's index showed "TBD" as the company. **Check:** archived snapshots of the profile and later edits, which often reveal the real employer and start month once the company launches. Refolk searches can target these directly, for example engineers in a given region whose profile lists a stealth or "TBD" employer. That is the quickest way to find joiners the company-name search missed.

### 7. Open-role drag misread

**False positive:** evergreen "always hiring" posts read as roles that have gone unfilled for months. **Check:** find the posting's first appearance in archived copies and confirm it names a specific plan-critical role. Then age it against the 75-day technical or 56-day business first-fill median. Leave generic talent-pool posts out entirely.

## What this read cannot establish

The read rests on public profiles and published benchmarks, and both have gaps the scoring sheet should state openly. Where a gap applies to the company in front of you, write it on the sheet and say what you would check.

- **Profile coverage of seed hires.** How many first-year hires at seed companies have a public profile is not established publicly. The nearest figure is one profile aggregator's own claim to cover about 76% of the labour force in Hong Kong and 83% in Singapore, which says nothing about seed startups. Check locally by comparing the roster count with the headcount the founder states. A large gap means missing people, not a small team.
- **Start-date drift.** Self-reported start dates can drift, and administrative records lag too. Carta notes that its hire records are typically logged after the board approves share grants. For anyone near the month-12 cutoff, check archived profile snapshots before you put them on either side of it.
- **Title accuracy.** How often public titles overstate or understate seniority is not established. That is why the give-up band uses employer size and hire number alongside the title.
- **Effect sizes.** This guide uses the direction of the Choi et al. and Sarada and Tocoian findings, not their coefficients. If your memo needs magnitudes, take the point estimates from the papers' own tables.
- **Stage-specific baselines.** Neither the departure rates nor the hiring-speed medians have a published split by seed versus Series A from a neutral primary source. No published method exists for measuring time to fill from re-posted listings, so the archived-snapshot approach here is a working method, not a benchmark.
- **Order of steps.** No published source runs these steps in this order. Investor checklists start with interviews and references. Starting from public evidence is my choice, and it is the reason the verdict can be checked by a second partner.

## Before the verdict goes into the memo

The read is done when another partner could check every score against a linked source and reach the same verdict. Run this checklist, then set a date to run the read again.

#### Talent-pull read sign-off

- [ ] The founding date and the month-12 cutoff are written on the sheet, with the source for each.
- [ ] Every clean joiner has a start month and a working source link.
- [ ] Every exclusion has a reason and an evidence link, including any cofounder with a "Founding" title.
- [ ] No cohort that shares a start date and a previous employer remains unless it has been checked against acquisition news.
- [ ] Followership counts only shared-employer ties; shared schools sit in the stranger column.
- [ ] Every give-up band records the joiner's hire number, and no joiner after hire 5 is banded High.
- [ ] Retention excludes contractors, advisors and acquired workers before it is compared with 18.0%.
- [ ] Every aged posting is plan-critical, dated from its first archived appearance, and compared with a median that is written on the sheet.
- [ ] A second partner scored the clean roster blind and the verdicts match.
- [ ] Gaps in the evidence (coverage, stealth, start-date drift) are written down, along with the reference question each one raises.

### Keeping the read current

The benchmarks behind the bands get revised. Before each use, re-check Carta's attrition page for the 12-month figure and median tenure, and re-check the latest Ashby startup hiring report for time-to-first-fill medians. Update the sheet's header if either has moved. Founding-title counts in Refolk's index shift as profiles update, so re-run the founding-title queries before relying on the 5.5x engineering-to-GTM ratio.

For portfolio companies, run the read again before the next round. The first-year cohort's 12-month retention will have resolved by then, open-role drag will show whether the last round's hiring plan happened, and a founder whose read moved from Ordinary to Magnet has given you evidence no pitch could.

## Frequently asked questions

### How do you evaluate a founder's ability to recruit before investing?

Rebuild who joined in the company's first year from public profiles, then score that roster rather than the founder's account of it. I score five things: how many joiners had worked with a founder before, what they gave up to join, whether strangers joined, how many have already left, and how long current roles have sat open. The total maps to Magnet, Ordinary, or Hiring Alone. Founder interviews and references come afterwards, to test what the roster shows.

### What is a normal early employee departure rate at a startup?

Carta's benchmark for startup employees with equity shows 6.9% gone by six months and 18.0% by one year, of which 11.8% left voluntarily and 5.4% did not. On a roster of five early joiners that works out to about one departure. Census research on acquisitions found that 12% of regular hires left in their first year, against 33% of acquired workers, so remove acquired teams before you compare.

### Is it a good sign if a founder's first hires all worked with them before?

It is a good sign for survival but not proof of pull. Research on Brazilian matched employer data found that founding employees with prior professional connections had higher survival odds but slower early growth. A roster that is entirely former colleagues shows trust inside a closed circle. It does not show that the founder can persuade strangers, and the hires this round pays for will mostly be strangers. So I score it below a mixed roster.

### How long is too long for a startup job posting to stay open?

Use the time-to-first-fill median for the function as the line. Secondary write-ups of Ashby's startup hiring report put it at 56 or 60 days for business roles and 75 or 76 days for technical roles, with senior roles slower than junior ones. A plan-critical role that is older than the median and has been re-posted is the clearest drag signal. Evergreen 'always hiring' posts should be left out.

### Does a 'Founding Engineer' title mean the person was an early employee?

Not reliably. In Refolk's index, 4 of 25 sampled US 'Founding Engineer' profiles and 6 of 25 UK profiles belonged to people who also carried CTO or founder titles. Before you count anyone as a joiner, check them against the company page and any cap table you have seen. A cofounder counted as a joiner inflates every dimension of the read.

### What if the startup is in stealth and profiles show no employer?

Date the placeholder employer from archived copies of the profile and from later edits, which often reveal the company name once it launches. If you cannot connect a 'stealth' or 'TBD' line to this company and a start month, leave that person off the roster. If the roster falls below three clean joiners as a result, record insufficient evidence rather than a verdict, and run the read again once profiles update.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/guides/founder-talent-pull-read*