The Founder Talent-Magnet Score: Grading Recruiting Pull From Public Hires

You are diligencing a founder and you need to know one thing the pitch deck will not tell you: can this person pull A-players into a risky, low-cash company against the pull of safe, high-paying jobs. This guide is for early-stage investors, platform and talent partners, and angels who want to convert that gut-feel signal into a defensible, source-cited score before a term sheet. It gives you the dimensions that matter, how to score each from public evidence, and how to tell genuine magnetism apart from hires the cash or the brand bought.

Why recruiting pull needs its own score

Recruiting pull is the founder's forward-looking ability to attract talent, and it is almost always scored as a sub-line rather than a headline, which is why it goes under-examined. Named frameworks bury it. Venture Briefing scores companies on five dimensions, each worth 0 to 20 for a max of 100, and lists "talent magnetism (can they recruit exceptional engineers and go-to-market leaders?)" as one sub-dimension inside the team score. A named GP at Hyde Park Venture Partners puts it plainly: a big piece of the evidence is your ability to recruit talent, so if you can demonstrate that, perfect. Allied Venture Partners lists attracting talent alongside planning, execution, and leadership as core competencies.

The mechanism to watch: because no one isolates it, a founder can pass team diligence on domain expertise alone while having zero demonstrated pull. The current team's quality tells you what the founder has already done. Recruiting pull tells you whether they can staff the next eighteen months. Those are different questions, and only the second predicts whether the company can grow past its founding cluster.

A founder guide frames the test as a filter: can you attract A+ talent to leave safe, high-paying jobs to join your risky venture, and your first 5 to 10 hires are a powerful signal. That is the window this framework works inside. No source publishes a numeric weight isolated for recruiting pull, so treat any composite you build as your own house standard, not an industry constant.

The four dimensions and what each one proves

The score has three scored dimensions plus a scarcity modifier. Each proves something specific, and each has a way it lies.

• Follow-on loyalty. People who left a job or followed the founder across ventures. Proves the founder has pulled talent before and can do it again. It lies when two ex-colleagues at one startup are coincidence, not loyalty.
• Caliber versus affordability. Senior hires who joined for less than the market would pay them. Proves conviction you can price. It lies when the package was actually at or above market and the money explains the move.
• Verification strength. How many independent public sources confirm each hire joined and stayed. Proves the team is real, not a wish list. It lies when a marquee name is listed with no corroborating commit, team-page, or announcement.
• Scarcity modifier. How thin the relevant talent pool is. Pulling a scarce-skill hire in a thin market is harder to buy, so the same hire is a stronger signal. In Refolk's index, US founding engineers outnumber UK ones by more than seven to one, and fewer than one in five US founding engineers carries a machine learning skill.

Verification sits at the bottom because everything above it collapses without it. Score loyalty on a hire whose join you cannot verify and you have scored a rumor.

Which public sources reveal who joined, and how much to trust each

Four public sources reveal who joined, when, and from where, and they are not equally trustworthy. Rank them by tamper-resistance, not by convenience.

The gap between top and bottom of that table is the whole game. GitHub commit authorship is recorded permanently and cannot be faked without leaving a trace. A professional profile, by contrast, has no accuracy check even when the account is verified, anyone can edit their own positions and degrees, and there is no verification mechanism for most details. Most employers do not even use profiles as their source for background data; the formal employment application is the record. Workplace verification exists but is not available to all companies and generally confirms only current employer by company email.

GitHub is the strongest join record for technical hires, but it has documented blind spots. The contributors graph shows only the top 100 contributors, does not count merge or empty commits, and is only available for repos with fewer than 10,000 commits. Commits authored with an email not linked to a GitHub account will not appear in the graph at all. So when you check a technical hire, read the raw commit log by author email, not just the graph, or you will conclude a real engineer "isn't really there."

How to score it: the procedure

Run these seven steps in order. Steps 3 through 5 have no fixed sequence across practitioners, so if you prefer to benchmark affordability before verifying, that is defensible; the dossier's sources do not converge on one order.

Budget roughly a day of analyst time and two to three hours of investor time for a first company. Most of the analyst hours go to steps 2 and 3, pulling and cross-checking sources, because that is where the score earns its defensibility.

The affordability test: pricing conviction instead of inferring it

The cleanest magnetism test is the affordability gap, and unlike loyalty it is computable. A senior hire who took a documented below-market package is absorbing quantifiable risk, which is conviction you can price rather than infer. The startup compensation tradeoff is documented: lower salaries are offset by equity at roughly 0.1% to 0.25% additional equity per $10,000 below market. So if a hire who would command a market salary takes a seed-stage cash number, you can estimate the equity that discount is worth and check whether the founder actually granted it.

Use the early-hire equity decay curve as your affordability yardstick. These are the points a real grant should land near.

Payroll data anchors the cash side too: seed-stage founders pay themselves roughly $132,000 to $149,000, and first employees take equity in the 0.5% to 4% range with a median around 1.49%. A senior operator joining a company whose founder pays themselves $140,000 cannot be on a $250,000 base. If they joined anyway, the money did not do the pulling. That is the signal.

Reading loyalty: who followed, and did it happen twice

Follow-on loyalty is the strongest reputation signal, but a single shared employer proves almost nothing; repetition is what proves pull. The qualitative anchor is the PayPal case, where within four years all 220 PayPal employees had left, and the alumni then hired from each other's teams and cross-invested, seeding YouTube, LinkedIn, Yelp, SpaceX, Palantir, and others. The lesson is not the fame of the outcomes. It is the pattern: people who were pulled once became a reusable talent pool the founders drew on again.

So loyalty compounds through reinvestment, not just re-hiring. A founder who pulled people once and then pulled some of the same people into a second venture is giving you a disproportionately predictive signal. A one-off overlap at a single company is weak; a follow across more than one prior company, or into a second startup, is strong.

No published percentage tells you what share "should" have followed. Treat it as a computed ratio: headcount who overlapped with a prior employer of the founder, divided by total early headcount. Report the ratio, but weight the repetition more than the raw number.

A single shared employer is a coincidence. A follow into a second venture is a reputation you can bank.

This is the step where finding the follow-on hires is most tedious by hand, because you are reconstructing prior-employer histories one profile at a time and matching them against the founder's own path. A plain-English search collapses that.

Refolk works across the public GitHub graph, public LinkedIn and Crunchbase records, and its own index, so you can ask for the overlap directly rather than cross-referencing employer histories manually. When you have the follow-on set, run each name back through the verification step; a followed hire still needs a second source.

How scarcity changes the score

The same hire is a stronger magnetism signal when the talent pool is thin, because a scarce-skill hire in a small market is far harder to buy than the same hire in a deep one. This is a modifier, not a fourth dimension: it scales the loyalty and caliber scores rather than adding to them. Two figures from Refolk's index set the scale.

A founder pulling a founding engineer in London is fishing in a pool roughly one-seventh the size of the US pool. Now narrow by skill.

Only 18.4% of US founding engineers carry a machine learning skill. Pulling an ML-capable founding engineer, especially outside the US, is a harder recruiting act than pulling a generalist in San Francisco, where in Refolk's index the deepest founding-engineer pool sits, followed by New York. Apply the modifier by nudging a scarce-market or scarce-skill pull up a band, and treating an easy-market generalist pull as the baseline it is.

How this score goes wrong

This is the part that survives partner review, so give it weight. Eight failure modes turn a good-looking hire into a false positive. Each has a check.

Two of these deserve extra attention. The brand trap: an accelerator brand like Y Combinator is a credibility signal that gets you in the door faster but is not sufficient on its own, and it opens doors independently of the founder. Isolate person-pull by asking whether the hire predates the brand association. If someone joined before the company got into an accelerator or landed a marquee investor, that hire is person-pull. If they joined after, the brand may have done the work.

The churn trap: startup annual churn runs about 57%, roughly triple the US average. A strong hire who already left is a stale signal. "Stayed" matters as much as "joined." Verify current status with recent commit activity or a current-employer field before you credit a hire at all.

Announcement lag is the quiet bias underneath all of this. Press-driven sourcing surfaces teams only after they are no longer early, so it skews toward already-validated hires and hides the risky, formative ones. Cross-check with behavioral traces - commits, registrations, contributor entries - not just launch posts, so you see the team as it actually formed.

Turning the score into a memo

Write the score as three dimension bands plus the scarcity modifier and a composite, each line carrying its source. Keep the bands coarse; false precision is worse than an honest range.

FOUNDER RECRUITING PULL

Team window: first [N] hires. Verified [X] / partial [Y] / unverified [Z].

1. Follow-on loyalty          [Low / Med / High]
   - Overlap ratio: [count] of [total] early hires shared a prior employer with the founder.
   - Repeated across ventures? [Yes / No]. Source: [profiles + second source].

2. Caliber vs affordability   [Low / Med / High]
   - Senior hires plausibly above founder-affordable comp: [names].
   - Tagged conviction-signal / cash-plausible. Benchmark source: pear.vc / kruzeconsulting.

3. Verification strength       [Low / Med / High]
   - Sources per hire, GitHub commit corroboration where technical.

Scarcity modifier: [pool size / skill rarity]. Applied: [+ / 0].

COMPOSITE: [band] with rationale in two sentences.
Open risks: [single-source names, stale hires, brand-attributable joins].

Before you call the score done, run the checklist. It exists to stop the two errors that get past partner review: scoring an unverified name and crediting pull that the cash or the brand actually bought.

Keeping the read current

A recruiting-pull score has a short shelf life because the team it grades keeps moving. Re-run verification, not the whole framework, on a cadence tied to the deal. The two things that decay fastest are current status, given ~57% annual churn, and the follow-on set, which grows as more of the founder's network joins after a raise. Recheck recent commit activity and current-employer fields before any follow-on decision, and re-pull the overlap set if the company has hired since your last read. The dimensions and benchmarks in this guide are structural and change slowly; the names attached to them do not. Grade the pull, then watch whether the people who were pulled stay.