Warm intros and double opt-ins perpetuate inequity (and what we can do about it)
The tech and VC worlds, like the world at large, are facing a reckoning over their history and practice of exclusionary bias. And while bias is incredibly complex and we have a long way to go, there are plenty of systemic changes that aren’t complex at all, like changing who and how we talk with people for the first time. The insistence on warm and double opt-in intros actively reduces the diversity of conversations people have and while the privileged have the right to continue those practices, it is empirically true that they ultimately result in systemic inequity.
Before we talk about why, and what to do about it, it is important to first define terms. A warm intro (versus cold outreach, when you simply approach someone you do not know without preamble) is one in which a third party introduces you to the person you want to talk to. So if Daniel wants to talk to Matt, he gets Sam (who knows both Daniel and Matt) to make an introduction that includes both of them. A double opt-in intro is one step further: before Sam introduces Daniel to Matt, Matt requires that Sam ask him if he wants the intro, enabling him to privately say no rather than having to bear the social burden of saying no to Daniel directly (the double comes from both Daniel and Matt opt-ing in, since in theory Sam could intro with neither of their permission, although this rarely happens in practice).
The science is very simple. At every additional decision point, the opportunity for bias increases. If anyone can email Matt freely, the only bias that exists is his willingness to write back; one opportunity for bias. If Matt only accepts warm intros, Daniel must have enough social capital to get access to one of Matt’s gatekeepers and then Matt has to be willing to write back; two opportunities for bias. If Matt only accepts double opt-in intros, Daniel must have enough social capital to get access to one of Matt’s gatekeepers, the gatekeeper must have enough social capital to get Matt to accept the intro, then Matt has to be willing to write back; three opportunities for bias.
This has pernicious secondary effects as well. For example, if Sam knows that Matt is unlikely to talk to Daniel for reasons of bias, he is less likely to offer the intro in the first place in order to preserve his social credibility with Matt, thus creating a world that reinforces Matt’s biases; the net effect is that the only people who get introed to Matt are, for example, white males, reinforcing Matt’s believe that the overwhelming majority of people worth talking to are white males. It is the very definition of a self-fulfilling prophecy.
To argue that these systems do not create bias is to argue one of two things: either that that bias does not exist in the first place or that one of those steps is free from bias. It is certain that bias does in fact exist: in VC, female-only founding teams got less than 3% of total funding in 2020. Ditto for Black and Latinx founding teams. The only alternative explanations to bias for those results is that either female/Black/Latinx people are not creating startups and trying for funding (demonstrably false) or that they are less fundable (which, unless you’re arguing that entrepreneurship is genetic, has to be a result of systemic bias).
It is also certain that gatekeeping systems create systemic bias. Clues as simple as a name or email domain can be enough to change opportunity outcomes; I chose Matt, Sam, and Daniel deliberately as stereotypically white male but what if the intro request had come from Jazmin Jones? And while an individual could argue that their intro system exists as an outlier to the general trend, unless they are specifically tracking diversity metrics that demonstrate that to be true, the assumption should be that the general trend holds. Because in the face of absolutely overwhelming evidence of systemic bias, arguments that hurdles don’t create bias are specious at best and actively sexist/racist/classist at worst, unless specifically disproven with equally overwhelming data on the exception.
So what do we do about it? In a perfect world, the powerful would open the floodgates and accept all comers, but that is both unlikely to happen and impractical. But because bias is supported by systems, so can anti-bias action. Combatting bias doesn’t have to be an all-or-nothing proposition; even if we can’t eliminate it, we can reduce it using the same sorts of systems that support it.
Intros are an easy place to start. We don’t have to stop accepting warm intros, but we can stop relying on them exclusively. Eliminating the need for opt-in makes the intros themselves incrementally less bias, with one fewer opportunity. And creating systems that circumvent intros, like publishing an email address publicly and committing time to reading through it in the order received or creating an opportunity form (along with transparency around process and a willingness to iterate) through Google Forms or Typeform are also key.
Even better, devote time to an open calendaring system. Personally, I use recurring Google Calendar appointment slots and then publicly distributing a bit.ly link (bit.ly/MattWallaertMeet) that allows people to claim time for themselves, which over five thousand people have done. I can vary the length of the slots, hours I devote per week, and timing of those hours on the backend with minimal disruption in order to control the amount of time I’m spending; if people are forced to book too far out, I can also change the visibility of the bit.ly link (by, for example, including or not including it in my Twitter bio) until I can clear the backlog. I even have a stock sentence (“bit.ly/MattWallaertMeet will let you schedule a call.”) setup as a text replacement for a shortcode on my phone and computer to make it easy to distribute.
The slots themselves lay out rules for engagement, like sending preread in advance and not attempting a LinkedIn connection without speaking first. A Google Meet is automatically created, although they also have the option to call me directly if they prefer. The calendar invite also includes a link to a Google Form to leave anonymous feedback, in case someone has a negative experience.
Open calendaring doesn’t mean I am available to people 24/7 but rather allows me to spend a specific quantity of time on activities that directly reduce bias. One of the critiques that people like Fred Wilson have leveled against open systems is that the demand is simply too great for the supply, but that’s not actually a critique: bias is about how we allocate supply, not about the overall demand. People far more important than me are devoting time to grooming their personal networks; I’m simply advocating they groom differently.
Open systems are still imperfect. Even knowing about a published email or calendar link is a form of bias, in the same way knowing what equity is and any other kind of knowledge. But they are orders of magnitude better than a double opt-in system, most specifically because they combat the most important bias: our own.
Side Note: The entire paragraph on how bias works can be applied to employment phone screens as well. To be clear, I’m not talking informationals, in which a recruiter or hiring manager explains the job in greater detail or clarifies details like salary or the need to relocate; I’m talking screens, which necessarily “catch” some people in them when recruiters are allowed to not pass them along based on the call. Every additional round of vetting introduces increased bias, which is best combatted through panel hiring and other known anti-bias systems. Individual recruiters can address their own biases and do great, anti-racist and anti-sexist work. But as with the statistics on VC funding, if relying on individuals to address their own biases worked, we wouldn’t have the employment demographics we do now. Systemic answers consistently win and we should not let clinging to our own power (like the feeling of being able to decline a candidate) hold us back from making key changes.
