I recently read the book Algorithms To Live By by Brian Christian and Tom Griffiths. You may currently be looking to hire the right staff in hospitality. There is some interesting advice on hiring that right person in the book. I’d like to share with you here.
Regardless of what position you’re hiring for. Trying to hire the right staff can be a stressful experience. It can be worth considering experimenting with algorithms formulated by computer scientists as detailed in the book.
As you will know algorithms are a finite range of steps that solve problems and take decisions. They’re the backbone of digital products. The device you’re using right now utilises algorithms. It runs them to decide which programs to utilise, when to terminate them, how it is sorting incoming information and how to communicate with other devices.
There are three computer algorithms that we can apply to hiring to efficiently solve the following three problems.
- Among a range of qualified candidates. Which should I hire when looking to hire the right staff?
- How should I go about onboarding after the hire?
- When to move on and engage someone new?
First of all, problem number one. You require the selection algorithm.
There are two big ways to fail in selecting an applicant for a role. Decide too early and miss out on better applicants. Wait too long, and miss the opportunity to hire the best applicant if that applicant applied early.
Let’s imagine you have a potential pool of 20 qualified applicants for a role you’re trying to fill. You don’t want to rush and hire the first candidate you interview. Additionally you also don’t want to waste days (or even weeks) interviewing all 20. Thankfully, there’s a third option. A scientifically proven algorithm dubbed the “37% rule”.
Hire the right person using the 37% rule
The 37% rule says that you should evaluate 37% of your applicants with no intention of hiring them.
Meaning, reject the first 37% unconditionally. After having evaluated the first 37% pick the next candidate you who applies who’s better than all previous options that applied.
So in the hypothetical scenario of deciding between 20 qualified applicants, you interview the first seven applicants without intention of making them an offer. Then continue interviewing the remaining 13.
Stop interviewing the moment you find a person who is better than any of those seven you initially interviewed. The benefit of this process is it gives you a chance to evaluate your preference. It additionally provides a handy benchmark which you can base your ultimate decision upon.
Several years ago a University student utilised the 37% rule to choose his future wife. He reasoned that he could be dating whilst between the ages of 18-40. By relying on the 37% rule, when he reached 26.1 years old he’d propose to the first woman better than all the others he’s dated.
When he encountered that woman after becoming 26 he said “without doubt, she reached the qualifications for the algorithm. I asked her to marry me… and she turned me down”.
Naturally the 37% rule isn’t 100% going to get you the results you desire, as the student realised. But it gives you the highest probability of achieving a great outcome.
The 37% rule is only applicable if you can’t revisit previous candidates
It’s important to point out, the 37% rule is only applicable if you can’t revisit previous candidates. It’s usually the situation when dating but not usually when hiring. For example, if you feel there’s a minimum of a 50% chance past candidates are still open to an offer. Or did they go for another role while you were deliberating on other applicants? Did they lose interest in the offer over time as often happens when the hiring process is extended?
If they’re still interested the 37% rule becomes the 61% rule. That is, if you’re seeking the best possible applicant, leave making an offer until you’ve interviewed a minimum 12 candidates. Then if you don’t find a better applicant among the remaining eight candidats, return and choose the best applicant you interviewed of all 20.
The On-boarding Algorithm
The next algorithm worth discussing is what we’ll dub the on-boarding algorithm. When you take new hires on, it’s difficult to know how much workload to give them. You need to maximize their skillset. However you don’t want to provide too many tasks so they make errors, don’t reach standards or cause issues you later need to address.
For years computer boffins came up against a similar problem. Computers would connect to a network of computers. A computer would get overwhelmed with requests and not deliver information to the computer requesting that info.
To resolve this issue computer scientists developed a brilliant solution known as transmission control protocol “TCP”.
If a new computer joins a TCP Network, it receives just one packet of info from the sender. It lets the sender know it’s received the packet by sending its acknowledgment packet. With every acknowledgement the sender doubles up the amount of packets it sends.
When the computer gets overwhelmed with too many packets and fails to send an acknowledge packet back to the sender, the algorithm “additive increase multiplicative decrease” known as “aimd“ steps in.
The aimd algorithm decreases the number of packets being sent by 50%. This large cut creates space on the network, allowing the receiving computer to catch up. After this reduction the sender will be allowed to add one extra packet to each future message. The developers state “aimd is like someone saying a bit more, a bit more, a bit more. Okay, way too much, cut right back. Okay a bit more, another little bit more.”
Use a similar technique to ensure the successful on-boarding of a new team member
Say you hire a new member of staff and you give them a small responsibility. When they return that work in time and on schedule, then double up workload. Keep doubling up and when you receive work that’s not up to scratch immediately cut the workload by 50%.
Utilising such exponential growth you can rapidly find a colleague’s capacity. And by subsequently increasing workload one unit at a time, you provide them an opportunity to build back up. Go step by step, when they achieve the level where they failed they’ll have more confidence and will have much more chance of success.
The Switching Algorithm
The next algorithm worth elaborating on is what we’ll call the switching algorithm. If a team member consistently meets expectations, you won’t need to consider replacing the hire. If they’re just reaching expectations some of the time you should probably explore further options such as the Gittins index.
In the 70s a young mathematician, John Gittins was asked to optimize unilever’s pharmaceutical Investments. Gittins had to decipher if Unilever should continue investing in a proven drug or switch to a new drug being trialed.
Based upon multiple versions and months of testing Gittens developed a table of values to help Unilever decide if they should switch to a new drug they thought could be at least 90% as effective and profitable as a current proven drug.
Unilever utilised this table to determine the success and failure record of two medications by determining how regularly those drugs successfully treated an ailment. They then utilised the Gittins table to decipher the correct Gittins index. The medication with the highest Gittins index was regarded the best avenue to pursue.
Gittins indexes are considered a mathematically dependable method to determine if one should switch or proceed as you are. You can utilise the Gittins index to decipher if you should continue investing in a venture or a recent hire.
Once you’ve predicted the success and failure ratio of your present option and an option with 90% of the expected value of that present option, you should utilise the Gittins index table to decide which choice has the greatest value and opt for that option.
Naturally you need not consult the Gittins index table for every decision.You only need to keep in mind these two key points when looking to hire the right staff.
Firstly, an untested option with a success / fail record of 0 and 0 has a Gittins index of 0.7. That is, if an individual isn’t meeting standards a minimum of 70% of the time (their success rate is lower than 70%), you should think about pivoting to a new hire if you think a new candidate has similar potential.
Secondly. If you go for a new option and their early success to failure ratio looks inferior to your present option it still could be a better choice. If a new hire has four successes and five fails (they reached your expectations four times and didn’t do so five times). They will have a higher Gittins index than someone experienced with a thousand successes and a thousand fails. The untested newbie is more valuable (in the early stages anyway), than the veteran of apparently equal ability because we know less about them. Exploration intrinsically has value, as trying new things raises chances of finding the best.
Take the away the stress when aiming to hire the right staff in hospitality
Why not try taking the stress out of decisions by trusting mathematically validated algorithms.
CONCLUSION
Hiring: The next time you’re when looking to hire the right staff remember the 37% rule. Determine your sample size. Then utilise 37% of your sample size to determine a benchmark. Then choose the next best option that applies.
Onboarding: Keep in mind the transmission control protocol and the aimd algorithm. Begin small keep doubling up, then reduce by 50% when you encounter failure then slowly build back up bit by bit.
Pivoting to a new hire: Remember this finding of the Gittins index. When you’ve got two similar options, go with the newer less tested option for greater potential reward.
That was the essence that I got from Algorithms To Live By, written by Tom Griffiths and Brian Christian. This book provides a new perspective to evaluating decisions.
If you like these tips, please share it and as always I’d love to see your thoughts and comments below.
Until soon, wishing you a productive week.
Nathan
You may check out Algorithms to Live By in the link below
https://www.amazon.com/Algorithms-Live-Computer-Science-Decisions-ebook/dp/B015CKNWJI