“It is a capital mistake to theorize before one has data.” - Sherlock Holmes (not a real person).
When we talk about our hiring platform we often describe it as "data driven", but what does this actually mean?
It seems to be an oft-touted saying with not much substance explaining how to actually go about it.
I generally despair at the dearth of practical steps to help recruitment functions move to a truly 'data-driven' approach.
So this is the Applied guide to what being an advanced data driven recruitment team looks like and the steps (starting with baby ones) you can take to get there.
Firstly, let's set up the context of modern recruitment to explain why measuring and gathering insights from data is so, so critical.
Recruitment is the most important thing a company does. Full stop.
This is a mantra oft-repeated but rarely internalised or actually fully understood by executive teams.
All of us, at one point in our careers, will think about our teams and colleagues and say "if only I could clone Sarah". But what if we could not only clone them, but also give her a different perspective: ways of thinking and working that made the team operate at a level well above the sum of its parts.
Getting great people will make or break your business. This should not be a platitude but should be an integral part of your organisation, because the rewards are massive: The productivity gap between average and high performers in complex jobs is 800%.
On the flip side, getting it wrong can also be costly. Not only from the time spent to find and interview a new employee, but also the damage or opportunity cost they bring with them.
A poor hire at mid-manager level with a salary of £42k can cost a business more than £132k.
HR needs a seat at the CEO's table and it needs to be considered as much of a business critical function as finance or sales.
Why is it not already for most organizations? Because historically, HR and talent has lacked the data driven insights and recruitment metrics to prove its business worth and lay out the resources it needs to impact the business.
Talent Acquisition is an over-worked, under-appreciated discipline
In most organizations, talent teams are under immense pressure to bring new employees through the door. There’s a huge focus on time and cost to hire and get people in as quickly as possible.
There’s also an unhealthy tension between most hiring managers or vacancy holders and the hiring process.
Lots of people think they know the best way to recruit people, or the best interview questions to ask or the tell-tale signs on a CV - but unfortunately, they’re generally mistaken. Our much-loved secret tricks and tells are actually counterproductive.
Recruitment teams have a tough time convincing hiring managers of the best methods and their job is made even tougher by a lack of functional respect and dearth of data.
I think it’s up to recruitment teams to break this unhealthy cycle by arming themselves with data-driven insights and a clear route from what they do to top and bottom-line business drivers.
Like any other function, HR must show why the issues it addresses matter to the business and that it has sensible ways to manage them. (Harvard Business Review, Peter Cappelli)
Other functions already operate in this way. If you compare recruitment to marketing, the missing data feedback in the former is stark and below is another one of my favourite quotes illustrating this point:
Imagine if the CEO asked marketing how an advertising campaign had gone, and the response was ”We have a good idea how long it took to roll out and what it cost, but we haven’t looked to see whether we’re selling more” (Harvard Business Review, Peter Cappelli)
Technology now enables us to better collect and garner insights from the full recruitment funnel, right through to how an employee performs in the role. Now is the time for HR to claim its rightful place as the most important business function.
Beware false dichotomies! Data-driven recruitment ≠ emotionless recruitment
Using data in your hiring doesn’t have to equate to robotic, soulless interactions with people. In fact, I would argue it enables you to better understand where the human touches are most needed.
Recruitment will always require some level of human interaction and nobody is saying that spreadsheets or algorithms should replace this. Instead, data should enable the empathy of your team.
The three stages of hiring metrics maturity
Having spoken to hundreds of in-house recruitment teams across all sorts of industries and company sizes, the three broad stages of metric maturity that I have come across are shown below:
Stage 1: No metrics
If you're just starting out, you’ll have no metrics.
This is absolutely fine and your first few recruitment processes will involve using your energy to create the process.
The danger is staying in this stage for too long, as you will be nothing but reactive in this state.
Startups that are starting to scale can sometimes get stuck here and then you're in trouble as the pull of 'bums on seats' might start to railroad your best data intentions.
Stage 2: Transactional
This is where most organizations sit. They collect data on how long it takes to hire someone and how much it costs.
These metrics are primarily concerned with working out how efficient the hiring process is, which is important but shouldn't be the raison d'être of the function.
Stage 3: Predictive
Very few organizations are at the stage, which is where data is being used to determine just how predictive of performance a recruitment process is.
This stage is all about working out how effective the recruitment process is and it involves some sort of feedback loop from how the recruited employee is going in the role they were hired for.
The predictive stage is where talent functions urgently need to move to and here's why...
Why you need to progress to predictive hiring metrics
In many organizations, the talent function is seen as an administrative function that handles the mechanics and minutiae of running a well-oiled process of bums-on-seats.
But this couldn't be further from the truth, since the process and methods used to find the next great humans to join your organisation are nuanced and ripe for data-driven insights.
Talent functions need to find great talent, and prove it.
The main purpose of a talent function has to be finding the best possible person to do a job. The end outcome is not just a person, any person, it has to be a great person who brings so much to the team on multiple dimensions.
If your main metrics are all around the efficiency of your process, such as costs and time taken, then you will be viewed internally as a cost centre open for sniping and tweaking at the cost of long-term value.
However, if your metrics speak to business drivers that are many times more valuable than the cost saving of shaving a few hours off of a process, then you'll have people's attention and their respect for your profession.
Increased fluency around metrics which impact directly on the top and bottom lines gets you more airtime and further influence to double-down and bring even more value to your organization.
If I was to break down what each of a talent function's main stakeholders were thinking at each stage, this is what it would look like…
As you move to predictive metrics, buy-in from your stakeholders increases. It's important to note that this change will be perceived externally too, by the candidates that are coming into contact with your process that is tuned for effectiveness, not just efficiency.
Below is a list of transactional metrics... and if you were to google recruitment metrics, these are the accepted de-facto standard set that you will come across.
As you can see, the majority of these metrics are concerned with analysing speed, costs and efficiency, with some limited forays into effectiveness and candidate experience.
Time to hire is the main metric for many organizations and it is an important one (but should never be your sole or main metric).
Our steer on transactional metrics is that they are necessary and you should measure enough to understand the size, speed and shape of your recruitment process, but they should not be the primary metrics or raison d'etre of the talent function.
The key with all of these metrics is to monitor them over time, so that you can benchmark against past performance and understand the impact of your initiatives or projects.
Here is an explanation of some of the top transactional metrics:
Time to hire
What? The amount of time between when the first candidate applies and when a candidate accepts an employment offer
Why? Speed and efficiency of process. Ensure candidates don’t lose interest.
How? Record the dates of each step in your recruitment process for each role then use an excel DATEDIFF function to calculate the number of days between each. Take the average across your roles to get your average time to hire. If you have an ATS, then it should do this as standard, out of the box.
Time to fill
What? The amount of time between when a role has been approved as a business requirement and when a candidate either accepts the role or walks in the door for their first day (changes between organisations).
Why? For resourcing purposes, to understand the full lead time to requiring a new employee and having them start,
How? As per time to hire, except extend the recorded dates from role approval, through to when then employee starts.
Cost per hire
What? The average cost, including hiring team time, advertising, assessments, software of all of your recruitment activities, split across the number of hires you made in that same period.
Why? To understand how much it costs to hire someone and to allow for budgeting of future hires and to make sure you are not overspending on things like advertising.
How? On a monthly basis, record the costs of all of your team, from people's salaries (maybe using averages) to software paid for that is directly involved in hiring. Divide this by the number of roles that were hired in that month. Consider doing a three-month rolling average if your hiring numbers are low.
Moving from no metrics to transactional
Starting from a clean slate is great as you'll have none of the inertia of BAU metrics. The order in which you should start implementing transactional metrics are as per below:
- Time to hire
- Cost per hire
- Pass-through rates
- Candidate experience scores
- Fill rate
The first step is as simple as starting an excel spreadsheet, using some basic excel functions like DATEDIFF and then updating the data monthly. After 3 months, you'll start getting some great charts that will be starting to give you some insights.
Beware of vanity metrics!
A vanity metric is one which looks great on the surface, or makes you feel good, but actually doesn't give you much insight.
A common one in recruitment is volume of candidates. The focus on this metric is often driven by the hiring manager who wants the daily dopamine hit of hearing how many new candidates have applied for their role.
High volumes of candidates can be problematic in recruitment as it is usually expensive to both source and sift through the sheer numbers in an effective way. Organisations with strong brands do often benefit from much higher candidate volumes, but they can often struggle to review everyone's application thoroughly, leading to shortcuts which actually decrease the effectiveness of the process.
It is much more helpful to link volumes to success rate through your recruitment funnel and understand the source of where those candidates come from. You can then double down on the sources that are working for you, because they are where you've sourced fantastic employees previously, not because you get thousands of irrelevant applications from them.
So, ditch the vanity metrics and push back on the hiring managers that ask for them. They aren't helpful!
You are what you measure!
"What gets measured, gets managed..." is the well known saying, falsely attributed to renowned management scientist Peter Drucker.
It’s pithy and makes a lot of sense sometimes, but my favourite part is the tongue-in-cheek follow-up quote: "...even when it’s pointless to measure and manage it, and even if it harms the purpose of the organisation to do so”.
Essentially if you’re measuring and following the wrong things, then you’d be wasting your time, or worse: doing more harm than good.
Predictive hiring metrics - these are the ones you should be tracking
Less than 10% of companies actually do these types of metrics and those that do often do it ad-hoc on a project basis. These metrics are all about determining how effective your process is so that you can continuously improve it.
The key here is to collect and connect pieces of data outside of your immediate hiring process, be it sourcing information from job boards or on the job performance data, in order to work out which pieces of your process are working.
Here are some common metrics which can speak how predictive your process is:
Quality of Hire
Quality of hire is the holy grail of recruitment and is usually a contentious and imperfect measurement.
What? A measure of how effective a hire is in their role
Why? To understand how effective your talent function is at sourcing and placing great talent
How? Every organizations does it differently. The generally accepted approach is to blend information from multiple sources and then come up with a score out of 100. Here are some commonly used
Quality of hire on its own is useful to prove the talent function is fulfilling its full value, however in isolation it is still just a bit of a pat on the back.
The real power of this type of metric truly comes into the fore when you compare it with the assessment scores you made at recruitment. However, it can be a bit like comparing apples to oranges, if your assessment scoring methodology is completely different to your quality of hire one.
What we recommend is that you align your quality of hire measurement with how you assess during recruitment, after all, if you’re truly testing candidates on their ability to do a job well and adding to your organisation on other intangible levels, then the way you measure during selection, should match how you would measure someone excelling in their role.
We understand this can be quite a leap, so below are some intermediate metrics and feedback loops that you can put in place with easier to find data, before you build up to a more fully aligned model.
Recruitment score vs 1st year retention rate
What? Comparison of how candidates fared during the recruitment process with whether they stay for 1 year
Why? Retention is a (noisy) proxy for performance, as if a new employee is doing well and enjoying their work they are likely to stay.
How? Ensure your recruitment process is data-driven and assessment scores are recorded. Equally, ensure that your retention data is recorded. Then connect the two sets of data on a candidate basis (anonymised or using a candidate ID rather than name/email) and eyeball the data.
The graph on the left shows a random relationship where there doesn't seem to be a relationship between recruitment score and retention - the insight here is that if the recruitment score is not predictive of if someone stays, the way you're scoring may not working properly.
The graph on the right shows that new employees with lower recruitment scores are much more likely to leave in their first year. Which is good, because it hints that your recruitment score is doing what it is supposed to do; predicting success in a role. In fact, you could argue that hiring people with low recruitment scores (which may happen in tight labour markets or in desperate times) is counterproductive on average, as the majority leave within the first year.
Recruitment score vs. appraisal score
What? Comparison of how candidates fared during the recruitment process with how they went in their 6/12 month performance appraisal.
Why? The performance appraisal is a proxy for performance, albeit an imperfect one. Therefore you can work out the relationship between recruitment score and how their manager (or preferably 360 degree peers and stakeholders) perceives their performance.
How? This one is quite hard given the reluctance of organisations to score employees during their appraisal process. Doing this is often perceived as quite 'old school' harking back to the days of forced curves pioneered by GE. We agree, this is tricky stuff, but we think if you can translate part of the performance appraisal from 'greatly exceeds' and 'below expectations' to scores out of 10, for example, you can start to do some analysis.
The graph on the left shows no real relationship between recruitment score and appraisal score (for the simplicity this has been set at 0-100%. The second graph does show that, as expected, the higher the candidate scored in their recruitment assessments, the higher they perform, which would be good news. With two continuous analogue variables you can start using the regression analysis inbuilt into excel (or a more advanced tool) to start seeing how strong your relationship between the two variables is. It's really getting exciting now!
Skills score (recruitment) vs. Skills score (in role)
This is the holy grail for us and how best-in-class organisations do this. It involves assessing candidates on skills during recruitment with a clear marking rubric, and then their 360 degree set of stakeholders using this same set of skills and marking rubric 6/12 months into the role.
What? Comparison of how candidates fared during the recruitment process broken down by skill with how their 360 degree stakeholders perceive they actually perform for those same skills.
Why? At a granular skills level, you are able to work out how well your questions, interviews and assessments are predicting the actual skills of the new employee once they are doing the job. If you see a disparity, you are able to fix the questions that are not predictive and double-down on those which give you the answers you seek.
Diversity & Inclusion funnels
Building diverse and inclusive teams is a business imperative as bringing diversity into your organization adds so much value to multiple aspects of the organization.
However, many organizations struggle to do this for a plethora of reasons, some within our immediate control and others that require more systemic changes. Most organzsations we speak to understand that they are falling short, however they often are not sure where the problem lies. In recruitment, this translates to do they have a sourcing, selection or onboarding problem.
By collecting and monitoring (on aggregate) the equal opportunities information throughout the recruitment funnel you are able to understand if there is a significant underrepresentation or drop-off of any one group, and take appropriate actions to fix it.
This information should always be optional - and never a requirement.
Here is an example of a D&I funnel within the Applied platform:
We've already discussed that volume of candidates is a vanity metric and should be viewed with some skepticism.
Sourcing reports however, are extremely useful to help you measure diversity and find out where to shout about your roles in order to get the best candidates.
An example is shown below:
This shows all of the different sources that a job was advertised at broken down by volume, scores throughout the process and the equal ops info of that group. You can use it to redirect your advertising spend to the best sources for the next role that you do.
Moving from predictive to transactional
Moving out from transactional metrics to those which get to the root of how predictive your process is, is hard, but it is imperative for the empowerment of your talent function and the untapped value it can bring to your organisation.
Hard but not impossible. Don't let imperfect data stand in your way of moving ahead, as even a light-touch form of quality of hire feedback into recruitment can give you powerful insights.
As with any major change, we recommend that you follow a change management process. This is the one that I like below. Depending on the size of your organisation, you will need to make this more or less formal; for startups just internalise these steps and execute upon them, for much larger orgs it might pay to put these steps to a schedule and create an explicit project.
For any of the above to work, you need to start being data-driven during your recruitment process, so the first port of call should be ensuring that each step has a marking rubric and that data is scored. So if there is a gut-driven initial sift of the CV, this should be upgraded to include a CV marking rubric at the least (or even better, a move to work samples).
So, pick your first battle carefully and build on those early small wins.
How to build a business case for recruitment
Step 1 of the above process requires building a case for change. The most commonly accepted way of doing this is via a business case, where you show how the change you want to make will impact on the top or bottom lines of your organization (or the value equivalent in your organisation, such as impact).
It can sometimes be a challenge to translate people focused initiatives into dollar value, but it can be done. Check out the calculations we use to demonstrate impact here - complete with the sources cited which we used to estimate the scale of the impact.
Key takeaways for data driven recruitment
So there it is, our manifesto and steps to implement an effective metrics and reporting system for your talent function. Here are the key takeaways:
- There are 3 broad stages of metric maturity: no metrics, transactional and predictive
- The goal should be to move to predictive as these metrics assess the effectiveness of your function
- The majority of organizations sit within transactional, less than 10% are truly in the predictive stage
- If you report out on and focus in on costs and time spent, they will start to shape the way your talent is viewed and views itself (i.e. as a cost centre only)
- The insights from even basic predictive measures are hugely impactful.
- Moving to predictive metrics empowers your function and gains it influence to deliver even more value.
- Change is hard. Start small and build upon your small wins, using an overarching change management plan.
We built Applied to make data-driven hiring a piece of cake. We’ve swapped gut-instinct and guesswork for the most predictive, behavioural science-based assessment methods - with metric tracking and reporting built into the entire process. For a tour of our platform, please don’t hesitate to book in a demo, or have a read of our resources.