Sally Gibson runs Dawleys, a contact center and fulfillment business in Ross-on-Wye, Herefordshire. When she joined Debra and me on Corey-osity Unleashed, she told us about a tip from one of her best performers. This person knew someone working in a coffee shop. Good with people. Worth a look.
Dawleys hired the coffee shop worker. The new hire excelled straight away.
Sally's lesson: A players attract A players. When people see a culture of growth and respect, they want to work there, and they bring their friends.
I agree with her. I also think most managers hear a story like this and keep the wrong half.
The Half Everyone Remembers
Referrals work. The research here is old and solid.
In 2000, sociologists Roberto Fernandez, Emilio Castilla and Paul Moore studied hiring at a customer service phone center. They put dollar figures on screening, hiring and training. Their verdict: a firm which invests in its employees' social connections earns a measurable return.
Castilla went further in a 2005 study of call center performance. Referred hires started out more productive than everyone else. His effect was modest... about 3.5% more calls per hour.
Then came the big one. Stephen Burks, Bo Cowgill, Mitchell Hoffman and Michael Housman pulled personnel data from nine large firms across call centers, trucking and high tech. Their working paper, later published in the Quarterly Journal of Economics, found:
- Referred workers were 10 to 30% less likely to quit.
- Referred truck drivers had about 7% lower weekly accident risk.
- Referred high tech workers produced more patents.
- Overall, referred workers earned the firm roughly 25% more profit per head.
So far, Sally wins. Pay the referral bonus. Go home.
The Half Nobody Talks About
Read the abstract of the Burks paper and one line jumps out. People refer others like themselves. Same age, same background, same behavior. The authors give the example in plain terms: unsafe workers refer other unsafe workers.
Truck drivers with accidents on their record referred drivers who went on to have accidents. Productive workers referred productive workers.
None of this should surprise you. People spend time with people like them. The careful driver knows other careful drivers. The person who leaves early on a Friday knows plenty of other people who leave early on a Friday. A referral doesn't filter for quality. It copies whatever the referrer already is.
Then look at Table 16, an accounting exercise on the trucking firm's data:
| Who brought them in | Profit per worker |
|---|---|
| Referred by an above-median performer | $4,190 |
| Not referred at all | $1,756 |
| Referred by a below-median performer | $1,063 |
Read the bottom row twice. A referral from one of your weaker people earned the firm less than a stranger off a job board.
The authors built those dollar figures on cost assumptions (fuel, price per mile, training), so treat the exact numbers as illustration. The ranking is the point. A players attract A players. C players attract C players. And C players have friends too.
The authors draw the obvious conclusion: firms gain most by rewarding referrals from their best workers. Most referral schemes I've seen pay the same flat bonus to everybody.
Every Referral Is a Bet on the Referrer
The same table holds a second finding I find more interesting.
Referrals whose referrer stayed at the firm earned $2,756 each. Referrals whose referrer eventually quit earned $1,798. Castilla saw the same pattern in his call center. As Joshua Pinkston summarizes it, the performance gains from referrals depended on the referrer staying with the firm.
So when you hire a referral, you aren't hiring one person. You're hiring a relationship. The person who vouched for them becomes their translator, their unofficial mentor and their reason to push through a bad first month. Lose the referrer and the referral loses their anchor.
Here's what this means in practice. When a strong performer hands in their notice, check who they brought in. Those people are now a flight risk. Your exit interview needs a second conversation, with the people they referred.
Your Referral Pipeline Is a Mirror
Burks and his co-authors found something else. Even after controlling for where people live, employees referred people of similar age, ethnicity, gender, marital status and smoking status.
Smoking status. Your team copies itself right down to the cigarette break.
I've argued before for hiring people who won't fit, and I've called out "culture fit" as a polite way of saying no weirdos. A referral scheme on autopilot is culture fit at scale. If your team is all one type, referrals will make it more so.
Sally's story cuts the other way, which is why I like it. Her best performer didn't refer someone from another contact center. They referred a barista. No keyword match on a CV. The referrer spotted something a screening filter would miss.
The research backs her up here too. Burks found referred and non-referred workers had similar cognitive and non-cognitive skills. The referral advantage came from better fit for the specific job, not from referrals being smarter in general. Your best people know the job from the inside. They know what "good at this" looks like, and it often isn't on a CV. I made the same point about engineers without a CS degree.
What I'd Do on Monday
- Build your own Table 16. Pull two years of referral hires. Next to each name, write the referrer's performance and whether the referrer still works for you. You'll see the Burks pattern in your own data, or you'll learn something new.
- Ask your best people by name. Don't wait for them to volunteer. Ask: "Who's the best person you've met who doesn't work here?" Pay a bigger bonus for referrals from your top performers if your scheme allows it.
- Ask for the unusual ones. "Do you know anyone in the industry?" gets you a copy of your current team. "Who have you met outside work who'd be good at this job?" gets you Sally's barista.
- Track the pairs. Link referrer and referral in your HR system. When one leaves, talk to the other within a week.
- Keep a second door open. Make sure people with nobody inside to vouch for them still have a fair route in. Otherwise your network becomes your ceiling.
Sally already pays attention to whether her people would recommend Dawleys. I wrote about how she fixed her eNPS question so the answer meant something. A referral is the same question answered with a name instead of a number.
Culture Compounds Both Ways
Sally says culture compounds. She's right, and the math runs in both directions. A team of people who feel respected brings in more people like them. A team of people who coast brings in their friends who coast.
Your referral bonus isn't a recruiting cost. It's a vote on who your culture copies next. Look at who's casting the votes.
Who on your team would you clone tomorrow? And when did you last ask them who they know?