The Strength of Weak Ties

Late last spring, after seven and a half years in various staffing leadership roles at Google, I moved back to the Bay Area from New York to lead Talent at Square. While letting people in San Francisco know I was coming back, I learned that one of my closest friend’s new roommate worked at my future company as an engineer. It was a bit serendipitous to think that in this apartment I had spent so many years hanging out in and never once met someone who was even good at math, there now lived an iOS developer who worked at my new employer. What were the chances? It turns out – surprisingly high.

In 1973, an American social scientist named Mark Granovetter published a landmark paper called the Strength of Weak Ties. Like anything discovered that ultimately becomes important, its submission to a preeminent research journal was rejected. Eventually published by a competing publication, the Strength of Weak Ties went on to become one of the most cited papers in social science. It also lays out a fundamental inefficiency in how many Silicon Valley tech companies are trying to recruit engineers. Granovetter is now a professor at Stanford studying, among other things, how connections in venture capital in Silicon Valley have impacted the region’s exceptional success nurturing innovative companies over time. This stuff is worth checking out.

The basic tenets of the Strength of Weak Ties research draws from a survey of almost 300 active candidates who had recently found a job, a data set which is well aligned to learn more about how to find engineers for a growing tech company. Even 40 years later, the manner in which growing companies identify and attract talent has not changed: via referrals from current employees. Granovetter’s focus was on exploring the depth of relationship between the candidate and the person who referred them into the company. He broke the referrers down into three categories, which were basically: someone the candidate knew, someone the candidate kind of knew and someone the candidate didn’t really know but was not a total stranger. He defined these groups by how often the referrer and candidate communicated prior to the job search (once a week, less than once a week but more than once a year, and once a year). Provocatively, 56% of the jobs referred to the newly employed came from the second category (someone the candidate kind of knew), 27% came from the third (someone they did not really know but was not a stranger) and just 17% came from the first (those considered close connections). If you’re looking for a job or if you’re looking for engineers and someone tells you the secret is to get out there and “network” with people you don’t know very well, this is what they’re talking about.

In today’s world, a common approach taken by many tech companies to identify and recruit engineers is to get everyone from a certain team into a room for a couple hours to source from their personal network. People pull up their LinkedIn and Facebook accounts and send out inquiries or hand over email addresses to a recruiting team to follow up with. The basic idea is to “map the talent” that current employees are connected to so that they can be targeted for potential hire. Online social networks have illuminated the 2nd and 3rd connections from any individual interested in who they might know in a helpful fashion (or perhaps unhelpful when I consider the 100’s of weak ties clogging up my Facebook Wall with pictures of children I’ve never met.) This is not a terrible way to identify new engineers that have a connection to the enterprise. On the downside, after one or two sessions, the recruiters or hiring managers are told “I’ve referred everyone I know”. From there the challenge is for the recruiting team to cultivate relationships with this talent map in the hopes there’ll be a match between the skills of one of the candidates and their timing in wanting to pursue a new opportunity.

This approach of sourcing the network of the company’s employees, meaning which connection is a qualified candidate to go recruit has a flaw: it does not fully take advantage of the network effect to attract new candidates. A more dynamic approach is to activate the networks, which means to engage all network connections to help identify a match that meets the specific needs of the organization. Put another way, instead of hoping to find that one great engineer who is waiting to be turned on by a customizable template email you sent to 100s of people, it’s better to consistently tell everyone in your network that you are looking for a certain type of engineer and to keep sending this message into your network all of the time. If the request is short and specific (you know, like something you could say to someone if you were in an elevator for half a minute or so), this request effectively reverberates across first, second and third level network, scaling across an exponential number of people seeking its target. You cannot anticipate which one of those nodes is tied to the right person and you shouldn’t try. You just need to keep sending out the signal relentlessly and sending it out across as many nodes as possible. The goal then for recruiting teams shouldn’t just be to sit down and get a list of leads, it should be to focus the company’s energy into sending specific needs based requests across each employee’s network — constantly.

The specificity part (“We need a senior data scientist, proficient in machine learning, who is interested in solving problems in financial services” vs. “we need engineers”) is important, as it helps to distinguish the signal from the noise – especially in a market like today’s when everyone is hiring. When this packet of need-based information crosses a social network and lands on a relevant target (“Hey! I’m a senior data scientist, proficient in machine learning and I love problems in the financial services sector! I didn’t know my friend’s friend worked there!”), it’s comes racing back across the network in the form of an interested candidate like sonar.


Now read this

The At Least 2 Pizzas Rule

Jeff Bezos is credited with a long standing management principle out of Amazon called the “The Two Pizza Rule”. According to myth, it stemmed from a desire to keep teams small, nimble and totally focused on one challenge. Two pizza teams... Continue →