I’ve written a bit about the process of job matching previously and commented on its role in job architecture engagements. Here I would like to consolidate some of the guidance I have received over the years, and some of my own learnings about the job matching process. For this discussion, I will focus on the process of job matching as it pertains to the “whole job classification” approach as opposed to the point factor methodology. In layman’s terms, whole job classification means selecting a role and level from a catalogue of available options created by the survey platform as opposed to assigning points to various attributes of the role.
This is going to be a long post so before we begin and to set the mood, I got ChatGPT’s help to write a Shakespearean sonnet on the topic:
Job matching is the task of HR,
To pay employees their just reward
For skills and duties they do bear
And market rates that they can afford.
By analyzing work and its demands,
And understanding the job’s pay range
HR ensures the compensation stands
Fitting for each job, neither too high nor strange.
Salary surveys, a tool most wise
HR uses to compare and rate
The worth of skills and experience lies
And set the salary at an appropriate rate.
So let us praise HR’s task with grace,ChatGPT/William Shakespeare/James Seechurn
For matching jobs and pay with great pace.
Are we all feeling sufficiently Shakespearean? Then let us proceed…
Job matching is part art, part science
The first point I want to make is that there are few absolutes in the process of job matching. Survey job titles, job descriptions, and leveling guides are made in a deliberately vague way since they need to be understood quickly, and accommodate a multitude of different companies across various industries.
The rule of thumb I was taught is to aim for a 70% match in responsibilities between the role, and the survey job description. For the vast majority of roles, this should be feasible but there are always a handful of roles that are unique or at least unusual and may not have an appropriate job description in the survey catalogue. In these scenarios, we often need to make our peace with the closest available match, a broad functional match, or use another comparable role as a proxy for the purposes of market data.
The best advice I can give here is; don’t aim for perfection. Even at the end of a thorough job matching process, it is unlikely that we will have 100% confidence in the matches and that is perfectly fine. The objective of market data is to increase our confidence in decision-making around pay and not to be an absolute recommendation.
Match the role, not the person
I have written about this before but it bears repeating: It is important to distinguish the role from the individual occupying it. Creating a link between market data and individual pay is, very basically, a two-step process, first of all we establish a link between the survey roles and our internal roles, then we establish a link between our internal roles and the individuals that occupy them. The goal of job matching is the first half of this and individual alignment comes later.
If we focus on an individual and not the role, then there is a good chance that we will obscure variances to the market that we want to be aware of, and that we create a data point that doesn’t truly relate to the role in question creating problems further down the line.
I was taught by job match guru, Eddie Tan at Radford, to think of this as matching the chair, and not the person that sits in it. A good test I like to use is to ask the question, “if that person were to leave, what level of skill and experience would you look for to back fill the role?”. This helps decouple the role from the individual.
This is sometimes easier said than done. Particularly in smaller organizations, individuals often come to define their own roles and the distinction blurs together – that is totally fine in my view but we should be aware of those situations.
Related to this point, it is important to match all individuals within the same role, to the same survey job and level.
Leveling is key
Mis-leveling is likely to cause bigger issues than mis-matching the family. To be clear and to help those unfamiliar, the two major questions to answer in whole job classification are “What is the right family?” and “What is the right level?”. While families are broad descriptions of career paths and include everyone from a grad to a principal, the level dictates the seniority within that family.
In reality the chances of pay between two very similar families being materially different, are minimal. If you get to the point where you are struggling to distinguish between two relatively similar matches, chances are the pay will be close so just pick the one that is closest and move on.
With leveling, however, things are much more sensitive and the typical variance is usually around 15-25%. Mis-leveling may also have a knock-on impact and push other levels up or down in the process
Leveling is certainly one of the more difficult aspects to get right, particularly if there is no pre-existing internal definitions that we can use. Leveling charts focus on the career path within a family which is relatively conceptual to begin with. Here are a few tips:
- Familiarize yourself fully with the leveling guides. Read through them, then read them again, then again, and highlight key words and phrases that you can use in conversation. “Autonomous”, “recognized internally as an expert”, “receives daily guidance”, “sets long-term strategy”; these are all examples of words and phrases that can clearly and effectively distinguish the expectations of one level from the next. Conversely “highly accomplished”, “experienced”, “learning”, are vague and easy to challenge.
- Check the distribution within the family. By this I mean, check how many people fall into each level to see if it is reasonable. For example, it is unlikely that a team can function if they are made up predominantly of entry-level roles or if it is purely distinguished/fellow levels. With enough of a sample, we should see a relatively normal distribution around the career-level which is the level at which a role is fully autonomous (referred to as “P3” in many surveys).
- Err on the side of leveling down and look for good reasons to level up. Particularly with roles occupied by high-performers, there is often a push to use higher levels. Start with the lower level first and identify reasons that the role requires more skills and experience – not that the individual in it happens to be a superstar. Use those phrases that you have up your sleeve to stress test the notion that the role is fitting of a higher level.
- At the end of the process, filter your job matching file by level and check that you have been consistent – Are all the Director roles an equivalent level? How about your P3s and your Managers. This check may indicate inconsistency in some areas.
Blend roles with caution
I am generally not a fan of blending datapoints together unless there is a very good reason for it. To my previous point, if you come down to two similar families and can’t decide, it is better to pick one rather than blend the two together – in this scenario, blending will not help you arrive at a more reliable datapoint – Granularity does not always mean accuracy and just creates an additional step in the benchmarking process.
If there is a legitimate scenario where a role is clearly performing the duties of two or more roles, I still wouldn’t usually advise blending. The reason for this is that if someone is truly performing the duties of multiple roles, then we probably want the market data to reflect the highest value across those possibilities. If we blend and average data points, then the resulting benchmark will end up lower than at least one of the data points. In this scenario, I suggest picking the highest value of the roles available.
Where I do agree with blending is if we are using multiple data sources as part of our philosophy. An example is a not-for-profit company that competes with the private sector. Philosophically, they may want to position themselves between the two sectors in order to compete for talent in which case blending two datapoints from two surveys makes sense.
Broad guidance aside, here are some of the common questions that come up:
How do we decide between Managers and Senior Managers?
For all intents and purposes, a Manager and a Senior Manager are the same thing, as are Directors and Senior Directors. As a general rule, Managers manage Individual Contributors, and Directors manage Managers – the same applies for the Senior equivalents which are essentially bigger, badder versions of the same role. Practically speaking, this might mean Senior Managers have bigger teams, or more senior direct reports, or more strategically important teams reporting into them than their non-senior counterparts. Because of this, the distinction can end up fuzzy, particularly in smaller companies.
I have often seen the Senior versions of these levels being used predominantly as a means of recognition and promotion for high-performing individuals where there is no Director or VP jump available. The jump from a Director to a VP, for example, is a big one and it may take time for the opportunity to materialize so having the Senior Director level available provides a promotional opportunity in the meantime.
A word of warning – internal titling can often throw us off here because if someone had the “Senior” title in a previous job, it is unlikely that they would be willing to relinquish it in their new role. The question I ask for the purposes of job matching is “Are there truly two tiers of Manager/Director in the organization and what distinguishes them?”
To my earlier point on looking at distribution, this remains a helpful exercise across the managerial roles. For example, it doesn’t really make sense that there could be more Managers than Directors. Bear in mind though that the distribution is often bimodal in nature with peaks at the Manager and Director levels and fewer individuals falling into the Senior equivalents.
Should we use years of experience?
I am a fan of years of experience for job matching purposes – so long as we use it in the right way. If it is used to assess individuals, however, it can be a risky attribute. For example, Mark Zuckerberg had 0 years of experience when he started Facebook, did that make him entry-level? How about the person on the accounts payables team that has been doing the same job for 20 years – does that make them a principal?
In conceptual terms, though, years of experience can be a helpful way to visualize the type of experience we would expect to look for in the role. To rephrase my earlier question “If the person were to leave, how many years of experience would you add to the job posting?”.
How do we ensure manager buy-in?
Ultimately it is hiring managers that will use the ranges and if they don’t have confidence in market matches then all that hard work could be for nothing. As such, it is important to get buy in across the business and the bigger the business, the more important it becomes.
The challenges are often both time and engagement – Being good at job matching takes training and practice and survey job matching materials are notoriously cumbersome.
If job architecture is coming up, it may be more efficient to wait until you have that customized set of roles, titles, and leveling guides to engage with people leaders – this means that the materials are more relevant, digestible, and useful going forward.
In the meantime though, you may just not have the institutional knowledge to match certain roles and need that input. In such scenarios, I generally avoid going through a full job match training with managers, and try to narrow down the outstanding job matches to a handful of laser-focused questions that can be addressed over Slack or email. These questions might be along the lines of “Could you give me 3-5 bullet points on the duties and responsibilities of this role”, or “Please review this description and let me know how accurately it describes this role”, or “Does this level require regular guidance or is it autonomous by nature”. This approach saves time in the early stages of the project, paving the way for deeper educational sessions and alignment review once custom job architecture and leveling guides have been developed.
How often should we review our matches?
Truth be told, probably all the time. Pressures from recruiters and hiring managers on data points, new roles, promotions, changes to the job architecture or internal JDs – It is better to think of job matching as an ongoing process rather than a point-in-time task. There will often be a broader annual review that takes place at the time of the survey submission and it is important to first familiarize yourself with any changes to the survey architecture since the previous year. Fortunately these are often summarized on an annual basis.
Like any skill, job matching takes time and practice to get right. Even then, given the room for interpretation, there are always grey areas. No matter how many times I have gone through the job matching process, I still come unstuck with unfamiliar roles or distinctions between levels, and still recommend matches which are changed at a later date.
Take your time, don’t get too attached, and embrace uncertainty.
However, I am sure there are plenty of other good job matching ideas and tips out there so let me know in the comments below:
“Happy are they that hear their detractions, and can put them to mending.”Much ado about nothing – Benedick