Warm Introductions: How They Actually Get Made

An introduction gets made when the person in the middle works out that passing you on costs them less than it returns. That calculation is the whole mechanic. Everything you do before the ask either lowers that cost or raises that return.

Three things set the price. The path decides whether the connector can reach the investor at all. The price decides whether they will spend the reputation. The pattern decides which version of you appears in the first line of their message.

Most founders work only on the path. Intros die on the other two.

Why these four models

This piece runs the Wire Model: score the features of the decision, route to a small ensemble of formal models, then force the ensemble to produce dated actions. The scores that mattered:

• Exposure topology (0.9). Who reaches whom decides the outcome before quality enters the room.
• Strategic actors with asymmetric information (0.7). Founder, connector and investor hold different facts and different downside.
• Cognitive and affective distortion (0.75). The connector recalls you through affinity. The investor hears you through a frame someone else chose.
• Regime-break risk (0.5). Machine-written outreach is collapsing the cost of a plausible cold email, which quietly breaks the filter every investor relies on.
• Deep uncertainty (0.6). You cannot observe a fund’s dry powder, its cycle position, or its LP call last week.

That routes to four models: network centrality (the path), signaling (the price), behavioral (the pattern), and mechanism design (the workaround). They span three outcome types, equilibrium, complex and noise, so their errors point in different directions and partly cancel.

The framework: an intro is a loan against the connector’s reputation

1. The path: access is a property of the graph you sit in

Start with the base rate. In a survey of 885 institutional venture capitalists at 681 firms, roughly 30% of deals came through the investor’s professional network, 20% through referrals from other investors, 8% from portfolio companies, close to 30% from proactive self-generation, and about 10% from inbound sent by company management.¹ The same study found the average firm screens around 200 companies a year and closes four.¹

Read that as a funnel and you learn nothing. Read it as a graph and the shape is obvious. Roughly six in ten opportunities arrive through someone the investor already trusts, so your pitch is competing for a slot that network position allocated before anyone read a deck.

Position is measurable and it pays. Better-networked venture firms post significantly better fund performance, and their portfolio companies are significantly more likely to survive to the next round and to an exit.² Centrality is an asset on the investor’s side of the table. Treat it as one on yours.

We also know which ties carry opportunity. A five-year experiment across 20 million LinkedIn members, 2 billion new ties and 600,000 job moves found an inverted-U: weaker ties move more opportunity than strong ties, up to a point, after which extra weakness stops helping.³ Granovetter’s 1973 argument survived its own causal test.⁴ The translation is blunt. Your co-founder and your best friend cannot introduce you. The person who has met you twice, in a setting where you did something visible, is the highest-yield node you have.

Exposure has a causal result behind it too. Random variation in how many venture capitalists were assigned as judges to panels at a business school new venture competition raised participants’ later odds of founding a venture-backed startup, and raised them far more for men than for women. The mechanism was follow-up: men were more likely to contact the judges afterwards.⁵ Exposure creates the edge. Someone still has to walk across it.

In Nairobi, Lagos, Kampala or Cairo the highest-yield node is rarely a fund partner. It is the founder one round ahead of you who already took that fund’s money. It is the distributor whose LPO you fulfil. It is the bank relationship manager who onboarded your merchant float and can read your settlement volumes unprompted.

Network centrality, the access lens

Assumes: opportunity flows along existing edges, and edge weight is roughly observable.

Fits because: exposure topology scored 0.9; six in ten deals arrive through a trusted node.

Breaks when: the graph is rewired fast, by a new fund, a new accelerator cohort, or a regulator opening a licence window.

Counteracts: the belief that a better deck fixes an access problem.

May reinforce: fatalism, and the idea that outsiders should stop trying.

2. The price: an intro is worth what it cost to send

Spence’s signaling result is the cleanest lens here. A signal separates types only when it is expensive, and expensive in a way that costs the weak sender more than the strong one.⁶ The connector is posting collateral. If you waste the investor’s hour, the connector pays for it on the next call, not you.

A connector who sends forty intros a month has an intro worth nothing, and everyone in the chain knows it. A connector who sends four a year gets read the same day. When you ask for an intro, you are asking someone to spend a scarce, priced asset. Ask accordingly.

Endorsement is worth real money. Privately held ventures with prominent alliance partners and prominent equity investors reached IPO faster and at higher valuations than otherwise comparable ventures without them.⁷ The market prices who vouches for you.

Referrals also carry genuine information, and we know what kind. Across nine large firms in three industries, referred applicants were more likely to be hired and to accept, were 10% to 30% less likely to quit, and performed substantially better on rare high-impact outcomes, despite looking similar on paper to everyone else.⁸ What the referral predicted was fit.

Here is the counterweight, and it changes what you write. A randomized field experiment sent nearly 17,000 emails to 4,500 active early-stage investors, varying which startup facts each investor saw. The average investor responded strongly to information about the founding team, and did not respond to information about traction or about the existing lead investor.⁹

The intro buys you the read. Nothing inside it buys you the meeting. The meeting is bought by three sentences on who is building this and why they can. Write those sentences yourself, put them at the top of the forwardable paragraph, and stop hiding them on slide nine.

Signaling, the price lens

Assumes: the connector bears a real, remembered cost for a bad forward.

Fits because: strategic actors with asymmetric information scored 0.7.

Breaks when: the connector is paid per intro, or is building a portfolio brand where volume is the strategy. Then the signal is decoration.

Counteracts: treating an intro as a favour that costs nobody anything.

May reinforce: credential worship, and over-paying for a famous name whose taste is untested.

3. The pattern: the connector describes you from memory

Memory is biased in a measured direction. Venture investors who share an undergraduate school are about 34% more likely to co-invest, and investors from the same ethnic minority group are about 39% more likely. Those same high-affinity pairings then perform worse: roughly 17% lower success where they shared an employer, 19% where they shared a school, and 20% where they shared ethnicity, with the damage concentrated in early-stage deals and traced to post-investment groupthink.¹⁰

Affinity picks who gets forwarded. It does not pick who wins.

The frame arrives with you. Across seven years of investor question-and-answer sessions at a New York pitch conference, investors asked men promotion-focused questions about upside and asked women prevention-focused questions about risk. Founders answered inside the frame they were handed, and funding outcomes diverged accordingly.¹¹ The first line of an intro sets that frame before you open your mouth.

Africa runs this pattern in the open. In East Africa across 2015 and 2016, more than 90% of disclosed startup investment went to companies with at least one European or North American founder, in a market where most early-stage investors were themselves expatriates. A Nairobi founder in that reporting named the mechanism plainly: the looks-like-me-sounds-like-me investing method.¹² Affinity plus centrality, doing exactly what the models predict.

Behavioral, the distortion lens

Assumes: people deviate from rational screening in stable, predictable directions.

Fits because: cognitive and affective distortion scored 0.75.

Breaks when: the screen is blind, scored and written down. Structure removes most of the effect.

Counteracts: the assumption that a good business gets found on merit.

May reinforce: attributing every rejection to bias, which stops you from fixing the proof.

4. The workaround: build the certificate when you cannot buy the path

Mechanism design asks the useful question. If you cannot own the graph, which rules can you enter that produce the same signal?

Across 87 new venture competitions, using a regression discontinuity around the judging cutoff, winning a round raised the chance of securing external financing by about 35%. The effect was strongest for ventures just above the cutoff that received no cash prize at all, which isolates certification as the active ingredient, and judges’ ranks predicted later venture success.¹³

A judged, scored, public process manufactures the signal an introduction carries, without requiring the connection. That escape hatch is open from anywhere.

So is the harder version: a certificate that travels alone. A signed LPO from a listed distributor. Ninety days of mobile-money settlement statements pulled from the operator rather than your spreadsheet. An agent-network retention curve with agent IDs attached. A regulator’s sandbox letter. Each one can be verified by a stranger in under ten minutes, which is the only test that matters.

The graph is tightening, so this matters more each year. African tech ventures raised US$4.1 billion across 570 deals in 2025, with Kenya, South Africa, Egypt and Nigeria taking 72% of the capital, while the number of participating investors fell 7% year on year.¹⁴ Fewer nodes, more concentration. If nobody can reach the investor for you, stop shopping for a path and start manufacturing a certificate.

Mechanism design, the rules lens

Assumes: you can enter a rule set whose output third parties already trust.

Fits because: deep uncertainty scored 0.6, and certification is the documented substitute for access.

Breaks when: the competition or accelerator has no reputation, so its certificate signals nothing.

Counteracts: passivity while waiting for a warm path to appear.

May reinforce: demo-day theatre, and optimising for judges instead of customers.

GEER: the levers, cheapest first

Four channels carry the exposure: access (graph distance), certification (who vouches), legibility (what the connector can say in one line), reciprocity (what they get back). Pull the cheap, reversible levers first.

1. Write the intro for them. One forwardable paragraph, under 90 words, no attachment. Team first, proof second, ask third. Hits legibility. Costs an hour.
2. Name the person. “Intros to investors” is unanswerable. “Would you forward this to Amina at that fund” is a yes or a no. Hits access.
3. Pay in information. Send the connector something they can use in their own work before you ask. Hits reciprocity.
4. Use double opt-in. Ask the connector to check with the investor first. It lowers their cost, which is the entire constraint. Hits price.
5. Climb the certification chain. Judged competition, named customer, signed LPO, regulator letter. Hits certification. Costs weeks.
6. Move your position. Ship where investors already look: a cohort, a sector working group, a public build log with real numbers. Hits access. Costs months.

No-lever flag: if you have no path and no verifiable proof, no intro technique closes the gap. You are in a certification-required state, and that costs time. Say so to yourself in writing and re-plan the quarter.

RADAR: the portfolio, dated

DO NOW, by T+3 days. These are reversible and dominant across every scenario.

1. List twelve named investors, with the specific partner, the cheque size and the thesis line that makes you fit.
2. For each, find two second-degree paths. Portfolio founders first, ex-colleagues second, customers third.
3. Write the 90-word forwardable paragraph. Team, proof, ask. Have one person who has never heard your pitch read it and tell you what you do.
4. Send three double-opt-in requests. Three, not thirty. Volume destroys the signal you are borrowing.

HEDGE, by T+14. Cheap insurance against the path never opening.

1. Enter one judged process with published criteria and named judges.
2. Produce one certificate a stranger can verify without you in the room.
3. Get two customers to agree, in writing, to be named as references.

DEFER AND TRIGGER. These are irreversible, so wait, but pre-commit the trigger now.

1. Defer: spending your three strongest relationships, announcing the raise publicly, engaging a placement agent.
2. Trigger to spend: two of your first six intros convert to a second meeting within 14 days. That confirms the paragraph works. Spend the top three relationships in the following week.
3. Counter-trigger: zero of six convert by T+28. The failure is in the paragraph or the proof. Stop asking, go back to the hedge list, rebuild the certificate.

If you are the one being asked. DO NOW: publish your intro rule in one line, including whether you read cold submissions. HEDGE: run one scored cold channel each quarter, written rubric, named reader. DEFER: changing your sourcing model until you have two quarters of both channels measured on source to first meeting, first meeting to term sheet, and 24-month survival.

CHAIN: what usually happens next

Match the reference class on structure. Referral hiring, third-party endorsement and weak-tie job search all run the same machinery as investor intros, and they show the same base rate: referrals carry real information about fit, and they reproduce the referrer’s own population.^{8, 7, 3}

Second-order consequence: intro-dominated sourcing narrows the founder pool to the connectors’ resemblance set, which is the 90% East Africa figure showing up as arithmetic rather than as malice.¹² Third-order: the resulting high-affinity syndicates underperform by 17% to 20%.¹⁰ The system pays for its own comfort, on a lag long enough that nobody connects the invoice to the purchase.

Subtract the counterfactual before you over-credit the intro. Part of what looks like referral value is selection. The connector forwarded the founder they already believed in.

Matrix-break flag. The cost of writing a plausible cold email has fallen close to zero, which destroys the information content of a well-written approach. That has been the quiet filter for a decade. Short run, the price of a genuine intro rises, because it is the only cheap filter left. Medium run, investors move to scored, structured, verifiable channels, because nothing else still separates. Build for the medium run.

What this ensemble cannot see

Four things, and they are large.

Fund timing. Dry powder, position in the fund cycle, an LP conversation last Tuesday. Invisible from outside, frequently the entire answer, and permanently mistaken by founders for a verdict on the business.

The private sentence. What the connector actually says about you when the forward is not in writing. That is where the decision happens and no dataset contains it.

Whether the connector’s taste is any good. Pre-seed picking sits close to the luck end of the luck-to-skill continuum, so one person’s judgment carries very little information about your company and a great deal about their own network. Treat one endorsement as one draw.

Survivorship. Every study cited here counts intros that got made. The intros nobody sent leave no record, and they are the majority.

So run the ensemble on the three things you control and price the rest as noise. One list of twelve names. One paragraph a stranger can forward without editing it. One certificate that travels without you. Build those by T+14, then ask.

Sources and notes

1. Gompers, P., Gornall, W., Kaplan, S. N., and Strebulaev, I. A. “How Do Venture Capitalists Make Decisions?” Journal of Financial Economics 135(1), 2020, 169-190. Survey of 885 VCs at 681 firms. Working paper: NBER w22587. Deal-source shares and the 200-screened, four-closed figure are from the authors’ deal-sourcing section, summarised at Harvard Law School Forum on Corporate Governance.
2. Hochberg, Y. V., Ljungqvist, A., and Lu, Y. “Whom You Know Matters: Venture Capital Networks and Investment Performance.” Journal of Finance 62(1), 2007, 251-301. Full text.
3. Rajkumar, K., Saint-Jacques, G., Bojinov, I., Brynjolfsson, E., and Aral, S. “A Causal Test of the Strength of Weak Ties.” Science 377(6612), 2022, 1304-1310. Record and abstract.
4. Granovetter, M. “The Strength of Weak Ties.” American Journal of Sociology 78(6), 1973, 1360-1380. Full text.
5. Howell, S. T., and Nanda, R. “Networking Frictions in Venture Capital, and the Gender Gap in Entrepreneurship.” NBER Working Paper 26449, 2019. Abstract. Random variation in the number of VC judges on competition panels, 2000 to 2015.
6. Spence, M. “Job Market Signaling.” Quarterly Journal of Economics 87(3), 1973, 355-374. Full text.
7. Stuart, T. E., Hoang, H., and Hybels, R. C. “Interorganizational Endorsements and the Performance of Entrepreneurial Ventures.” Administrative Science Quarterly 44(2), 1999, 315-349. Record.
8. Burks, S. V., Cowgill, B., Hoffman, M., and Housman, M. “The Value of Hiring through Employee Referrals.” Quarterly Journal of Economics 130(2), 2015, 805-839. Full text.
9. Bernstein, S., Korteweg, A., and Laws, K. “Attracting Early-Stage Investors: Evidence from a Randomized Field Experiment.” Journal of Finance 72(2), 2017, 509-538. Roughly 17,000 emails to 4,500 investors on AngelList. Record.
10. Gompers, P., Mukharlyamov, V., and Xuan, Y. “The Cost of Friendship.” Journal of Financial Economics 119(3), 2016, 626-644. Working paper. The 34.4% and 39.2% co-investment figures and the 17% to 20% success-rate reductions are as reported in HBS Working Knowledge.
11. Kanze, D., Huang, L., Conley, M. A., and Higgins, E. T. “We Ask Men to Win and Women Not to Lose: Closing the Gender Gap in Startup Funding.” Academy of Management Journal 61(2), 2018, 586-614. Field study of investor question-and-answer sessions at a New York startup pitch conference, 2010 to 2016. Summary and record.
12. Village Capital, “Breaking the Pattern” (2017), reporting that more than 90% of disclosed East African startup investment in 2015 and 2016 went to companies with at least one European or North American founder. Finding and the founder quotation as reported by Rest of World, 2021.
13. Howell, S. T. “Reducing Information Frictions in Venture Capital: The Role of New Venture Competitions.” Journal of Financial Economics, 2020. Regression discontinuity across 87 competitions. NBER w23874.
14. Partech, 2025 Africa Tech Venture Capital Report. US$4.1B total funding, 570 deals, 72% of capital to Kenya, South Africa, Egypt and Nigeria, investor participation down 7% year on year. Report page.