Faith in Algorithms

Tech is powerful, but only when applied with in-depth knowledge of the sector in which it’s used.

Two-sided markets

Two-sided markets are platforms that bring together two parties - buyers and sellers - to help them meet their respective needs efficiently. eBay, for example, is designed to enable individuals to sell items to those that wished to buy them.

Two-sided markets work best when the needs of both parties are well-understood and can be clearly defined. Airbnb, for example, asks its buyers - people seeking accommodations - to define their accommodation needs. How many bedrooms? Price range? Neighbourhood? The sellers in this case - property owners - define their offerings in similar and familiar terms.

two_sided_marketplace.001.jpeg

A Lack of Precision Where Education and Labour Meet

The needs of buyers and sellers cannot always be as easily defined, particularly in the area of education and employment.

A fast-growing start-up in Canada has generated a great deal of attention by applying the two-sided market model to the field of recruitment. They bring together graduates seeking employment (sellers) and employers seeking graduate (buyers). The company aims to displace the traditional student job fairs, in which a limited number of employers visit campuses to conduct preliminary evaluations of graduates.

The platform captures the relevant student information - the program of study, work experience, and other relevant information. Employers, likewise, share job requirements. Algorithms are then applied to match the two parties.

The quality of the matches made within the platform is wholly dependent on the quality and breadth of the information provided by the buyer and seller. Without good input, the output will be flawed. However, students, first of all, don’t have access to the kinds of detailed information they need. Transcripts from alma mater and college-age work experience provide little insight into the individual’s actual skills and knowledge - and even less for soft skills and general sensibilities - such as the kinds of work environments in which they are most comfortable, productive or both.

The information needed by employers is no better. The typical job description tells us very little about who will succeed in the role. They’re designed primarily to describe a baseline of skills and experience which most HR professionals admit doesn’t align with the make-up of those who have succeeded in the role in the past. Moreover, if we add the variable of generational differences - that are particularly acute at this point in history - we frequently see job applicants with higher levels of education than the hiring manager.

The point here is that some types of information can not be easily matched through algorithms; certain aspects of education and employment included.

Measuring Immeasurables

That said, the inability to produce excellent job matches - which is fundamental to this business - may not stand in the way of the venture’s success. This is because the very same factors that make aligning graduates with ideal jobs so difficult, also makes it very hard to evaluate the quality of the matches. Hiring managers may not find that the actual job applicants generated by this platform any more or less accurate than other methods used to identify ideal job candidates. It’s possible - though difficult to measure - that the belief in the power of algorithms (or more generally, the “magic” of technology) may be what determines how happy hiring managers are with the platform. If they think the technology is capable of making these matches accurately, then they may interpret it positively.

It’s important to remember that hiring managers don’t hire employees frequently. Hiring managers responsible for a typical department in a mid-sized corporation, for example, may not hire staff more than once per year - and not for the same position. So, there’s not a great deal of feedback coming to them that, over time, helps them develop a precise sense of whether the most recent crop of job applicants are any more suitable than the last time they sought to find someone for the same position. If it does seem different, they’re as likely to chalk this up to luck or changes in the quality of applicants available or some other factor. In the end, due to the need to fill a position, they’ll choose the best of those that came by whatever process they’ve used, whether driven by algorithms or recommendations by a friend.

Technically, it’s possible for this platform or others like it to add extensive testing services that help define the graduate's skills and translate it into employer job specifications. Many companies offer these kinds of tools. I can’t speak to the value of their value, but it is clear that employers are looking to improve their accuracy when making new hires. If this service is found to be reliable, it will increasingly overlap with what has traditionally been the domain of colleges and universities: the evaluation and reporting of the workforce. It’s a space worth watching.

The needs of buyers and sellers cannot always be as easily defined, particularly in the area of education and employment.

A fast-growing, high-profile start-up in Canada has generated a great deal of attention by applying the two-sided market model to the field of recruitment. They bring together graduates seeking employment (sellers) and employers seeking graduate (buyers). The company aims to displace the traditional student job fairs, in which a limited number of employers visit campuses to conduct preliminary evaluations of graduates.

The platform captures the relevant student information - the program of study, work experience, and other relevant information. Employers, likewise, share job requirements. Algorithms are then applied to match the two parties.

The quality of the matches made within the platform is wholly dependent on the quality and breadth of the information provided by the buyer and seller. Without good input, the output will be flawed. However, students, first of all, don’t have access to the kinds of detailed information they need. Transcripts from alma mater and college-age work experience provide little insight into the individual’s actual skills and knowledge - and even less for soft skills and general sensibilities - such as the kinds of work environments in which they are most comfortable, productive or both.

The information needed by employers is no better. The typical job description tells us very little about who will succeed in the role. They’re designed primarily to describe a baseline of skills and experience which most HR professionals admit doesn’t align with the make-up of those who have succeeded in the role in the past. Moreover, if we add the variable of generational differences - that are particularly acute at this point in history - we frequently see job applicants with higher levels of education than the hiring manager.

The point here is that some types of information can not be easily matched through algorithms; certain aspects of education and employment included.

measuring Immeasurables

That said, the inability to produce excellent job matches - which is fundamental to this business - may not stand in the way of the venture’s success. This is because the very same factors that make aligning graduates with ideal jobs so difficult, also makes it very hard to evaluate the quality of the matches. Hiring managers may not find that the actual job applicants generated by this platform any more or less accurate than other methods used to identify ideal job candidates. It’s possible - though difficult to measure - that the belief in the power of algorithms (or more generally, the “magic” of technology) may be what determines how happy hiring managers are with the platform. If they think the technology is capable of making these matches accurately, then they may interpret it positively.

It’s important to remember that hiring managers don’t hire employees frequently. Hiring managers responsible for a typical department in a mid-sized corporation, for example, may not hire staff more than once per year - and not for the same position. So, there’s not a great deal of feedback coming to them that, over time, helps them develop a precise sense of whether the most recent crop of job applicants are any more suitable than the last time they sought to find someone for the same position. If it does seem different, they’re as likely to chalk this up to luck or changes in the quality of applicants available or some other factor. In the end, due to the need to fill a position, they’ll choose the best of those that came by whatever process they’ve used, whether driven by algorithms or recommendations by a friend.

Technically, it’s possible for this platform or others like it to add extensive testing services that help define the graduate's skills and translate it into employer job specifications. Many companies offer these kinds of tools. I can’t speak to the value of their value, but it is clear that employers are looking to improve their accuracy when making new hires. If this service is found to be reliable, it will increasingly overlap with what has traditionally been the domain of colleges and universities: the evaluation and reporting of the workforce. It’s a space worth watching.

The needs of buyers and sellers cannot always be as easily defined, particularly in the area of education and employment.

A fast-growing, high-profile start-up in Canada has generated a great deal of attention by applying the two-sided market model to the field of recruitment. They bring together graduates seeking employment (sellers) and employers seeking graduate (buyers). The company aims to displace the traditional student job fairs, in which a limited number of employers visit campuses to conduct preliminary evaluations of graduates.

The platform captures the relevant student information - the program of study, work experience, and other relevant information. Employers, likewise, share job requirements. Algorithms are then applied to match the two parties.

The quality of the matches made within the platform is wholly dependent on the quality and breadth of the information provided by the buyer and seller. Without good input, the output will be flawed. However, students, first of all, don’t have access to the kinds of detailed information they need. Transcripts from alma mater and college-age work experience provide little insight into the individual’s actual skills and knowledge - and even less for soft skills and general sensibilities - such as the kinds of work environments in which they are most comfortable, productive or both.

The information needed by employers is no better. The typical job description tells us very little about who will succeed in the role. They’re designed primarily to describe a baseline of skills and experience which most HR professionals admit doesn’t align with the make-up of those who have succeeded in the role in the past. Moreover, if we add the variable of generational differences - that are particularly acute at this point in history - we frequently see job applicants with higher levels of education than the hiring manager.

The point here is that some types of information can not be easily matched through algorithms; certain aspects of education and employment included.

Measuring Immeasurables

That said, the inability to produce excellent job matches - which is fundamental to this business - may not stand in the way of the venture’s success. This is because the very same factors that make aligning graduates with ideal jobs so difficult, also makes it very hard to evaluate the quality of the matches. Hiring managers may not find that the actual job applicants generated by this platform any more or less accurate than other methods used to identify ideal job candidates. It’s possible - though difficult to measure - that the belief in the power of algorithms (or more generally, the “magic” of technology) may be what determines how happy hiring managers are with the platform. If they think the technology is capable of making these matches accurately, then they may interpret it positively.

It’s important to remember that hiring managers don’t hire employees frequently. Hiring managers responsible for a typical department in a mid-sized corporation, for example, may not hire staff more than once per year - and not for the same position. So, there’s not a great deal of feedback coming to them that, over time, helps them develop a precise sense of whether the most recent crop of job applicants are any more suitable than the last time they sought to find someone for the same position. If it does seem different, they’re as likely to chalk this up to luck or changes in the quality of applicants available or some other factor. In the end, due to the need to fill a position, they’ll choose the best of those that came by whatever process they’ve used, whether driven by algorithms or recommendations by a friend.

Technically, it’s possible for this platform or others like it to add extensive testing services that help define the graduate's skills and translate it into employer job specifications. Many companies offer these kinds of tools. I can’t speak to the value of their value, but it is clear that employers are looking to improve their accuracy when making new hires. If this service is found to be reliable, it will increasingly overlap with what has traditionally been the domain of colleges and universities: the evaluation and reporting of the workforce. It’s a space worth watching.