William Wong, Aug 2021
Metrics
Fine-Tuning Your Pipeline With Data
No single word strikes more fear and angst in me than metrics. And I'm probably not alone. Metrics are a powerful tool, but like Spidey always says "with great power comes great responsibility." Bad metrics (and even good ones) if used incorrectly can cause a lot of destruction.
What are bad metrics? Anything that is used to reward or punish people. Yup, even those that reward folks are evil cause when you're praising one group, the other is punished. Do not use metrics as a behavior shaping tool.
So what is it good for? The strongest recruiting teams thrive on data and use it to improve performance. They utilize pipeline analytics to allow sourcers, recruiters and leaders to view the entire talent funnel to identify weak spots, and then fix it.
This article will cover the best metrics to pull, an in-depth look at one of them, and troubleshooting strategies.
When deploying metrics, our goal is to make sure they track outcomes (not just actions). Tracking actions is easy but doesn't really capture the business impact. What do I mean by this?
Tactical vs Strategic Metrics
With so many choices on what to track - which metrics should we focus on? Below is a list, curtesy of LinkedIn's Future of Recruiting Report (2021). These metrics can be sliced into various combos to tell us a story about our TA strategy.
Most folks track Time to Hire (an action) but it doesn't tell the whole story. Ask yourself, "what is the real impact of this" (an outcome). A slow recruiter who hires 10 high performers is better than one who blasts out 25 poor performers (here we're combining # of Hires + Time to Hire + Quality of Hire). Or perhaps we're only gauging recruiters based on # of hires and TTH - which can give false readings on performance. Consider injecting Cost per Hire into the picture to get a clearer picture of their ROI.
Note: Should also mention certain metrics like TTH can differ depending on the type of role, location and various other factors. How? It will take significantly longer to hire an Android Engineer w/ a security clearance in Wichita, KS than a Full Stack Java Engineer remotely anywhere in the USA. Consider injecting a difficulty modifier into TTH to get a more accurate reading, or only compare TTH per pipelines.
Source of Metrics
Before we venture any deeper, let's take a quick pause and address a topic everyone should be asking themselves:
"How are you currently obtaining your metrics?"
Is it via an automated system that is user friendly, allows you to configure the data as your heart desires and presented in a visually stunning fashion? Or is it thru a tedious manual process?
I still see TA teams exporting data from their ATS into excel and then manually splicing the data into useable content. I'm all about pivots, formulas and macros, but there are just so many problems with this practice.
Key Issues
-
Data is not real time (e.g. reporting is done on Monday, but analyzing with team on Thursday).
-
Takes too much effort. The manpower is better spent elsewhere.
-
Ugly visualization. Many of the data is not presentable. Think Tableau vs excel charts.
-
Not easily configurable. Let's say I want to filter these metrics by recruiter, job types, managers and demographics. It'll be a Herculean task to deliver.
Are you spending more time gathering metrics than using them? Please please please utilize a tool to automate all of this. Recruiting is a strategic role and our leaders are leaning on us to provide thoughtful solutions. Don't get stuck in the weeds.
I normally don't plug vendors, but GEM is a perfect solution for this. OK, now that I've got this out of my system, let's continue.
"Recruiters need to analyze their funnel as rigorously as marketing and sales teams do."
Martin Beischl, Global Director at riskmethods
Hiring Funnel Conversion Rates
Out of the 8 metrics recommended above, there's only 1 that's directly tied to our teams production numbers. Want more hires? This is the one to use. It's a results-orientated metric.
How it works: These set of metrics look at how many people passes from one stage of our recruiting cycle to the next. Every company uses different stages and that's ok. Key here is to look at all stages from a wholistic viewpoint to see how one affects the other.
This gives us an objective view of the team's efforts so we can figure out where the hiring pipeline is stuck or leaky, make quick adjustments and then see those strategies pay off.
Example: a high [Submit To Manager] but a low [Sub to Int %] means we need to focus more on quality candidates.
"You can't build strong recruiting strategies if you don't know where to move the needle - or which needles need to be moved." - GEM
Talent Funnel Overview
Alright, let's get the show on the road. When we talk about the talent funnel, it's often broken up into two different parts (top and bottom). When it comes to metrics, the top and bottom of the funnel requires a very different set of lens and strategy.
Top of the Funnel
The diagram below depicts what a typical talent pipeline will look like. At the top we capture candidates from two different buckets - one is passive and the other active.
[ 1 ] Application Channel
First let's talk about the passive bucket on the right. When a job req is created it gets posted at various sites thru the company's job syndication alliances. These sites could include your typical job boards like LinkedIn, Indeed, Monster, Career Builder, Dice, as well as government sites, diversity groups, etc.
The purpose is to publicize the role in hopes of attracting inbound applications. This works pretty well for the G&A space, but the yield for tech roles is fairly low (~20% of applicants are actually qualified for the job). This bucket should be seen as a bonus and not your only solution.
[ 2 ] Sourcing Channel
In the war for talent you can't just rely on inbound applications for several reasons but the most important one is this: LinkedIn reports 90% of people are open to learning more about a new job but only 36% are actively searching for one. We need to get off our butts and actively source candidates that are a fit for our needs.
Top of the Funnel Metrics
Tracking outreach campaigns is something that's been done by marketing teams for decades but this concept is still fairly new to the recruiting world. There are tools now that can dig pretty deep into stuff like which part of your automated email campaign was the most effective, or which talent pool is more receptive to which content. While those are pretty neat, there's 3 main stats that we must track.
[ 1 ] Number of Reach Outs
This is the unique number of candidates your Sourcer should be targeting. This number will vary based on what your plate looks like but generally my target for a dedicated sourcer is 200 per week.
[ 2 ] Response Rate
This number represents the percentage of people that actually responded to your messages. The goal is 35% (industry average). Meaning if we reached out to 200 candidates, 66 will respond.
[ 3 ] Interested In Job
Just because someone responded to your inquiry doesn't mean they're interested in being an applicant. In general, I would say over 50% of responses are interested in your role. Increasing this number isn't always as black and white as the two metrics above, but I still track it so we know how many applicants are sourced into our talent pipeline.
A drip campaign helps dramatically improve your response rate. If you're not already utilizing one, beg your leaders to implement an automated email solution.
Talent Pipeline Metrics
OK, now that we have a bunch of candidates via applications and sourced, what now? We employ three types of metrics: one tracks the number of candidates in a stage, one focuses on passthrough rates and the last tracks time in stage.
[ 2 ] Passthrough Rates
How many people passes from one stage to the next. We analyze conversion rates to identify the weakest points of our funnels to understand how to fix it.
Example: a high [Submit To Manager] but a low [Sub to Int %] means we need to focus more on quality candidates.
[ 1 ] Candidates in Stage
This one is simple, how many candidates are in each stage. Do we have enough candidates at the top of the funnel, how many are interviewing, down to the amount that gets hired.
[ 3 ] Time in Stage
How long are candidates sitting in each stage. Shines light on areas that we can fine tune to speed up the hiring process.
Example: a significant decrease in amount of [Hire] to [Starts] combined with long delay between [Hire] to [Start] could point to BG checks and immigration.
Important Drivers
While there may be a lot of metrics to consider, not all of them generate the same value. Some are primary drivers while others are secondary.
Here is what a typical technical search campaign looks like for a month. The passthrough rates will vary by types of roles, complexity, company, etc. But the numbers below can be used as a benchmark to identify potential areas that can use some tweaking.
Sourcing Metrics
Everything starts at the top, and here we have the [# of Reach Outs] with their corresponding [Response Rates]. The number to aim for and their results will greatly differ from one person to another. Create a plan tailored to your Sourcer's strengths.
GOAL: Try to get 200 reach outs a week. That type of volume can sustain a decent return. The industry average for response rate is 35% - use that as a bar. Half of those candidates will most likely be interested in applying for the role.
Troubleshooting: A low # of reach outs can mean various things like not understanding the job, unclear on tech, playing god, inefficient search strategies, help with search strings, etc. Again, don't use metrics as a punishment, but to figure out where the challenge lies, and then fix it. Anything less then a 30% response rate usually means either incorrect targeting, spammer or the messaging needs improvement.
Submit To Manager
This number is probably the most important metric to keep track of. Nothing matters more than the quantity of subs you provide to the manager. Too much and you're just throwing spaghettis against the wall to see if anything sticks. Too little and we don't have enough candidates to choose from. So what's the appropriate number?
GOAL: It all depends on your historical performance on similar roles. No history, then aim for 10. Work backwards to see how many reachouts you need to generate in order to yield 10 subs. Depending on the quality of your sourcing, it would say ~25% of all interested candidates during the outreach campaign will be qualified to be a submittal. So 40 candidates interested will yield 10 Subs to Mgr.
Troubleshooting: Low submits to manager could stem from many things. The most obvious points to our sourcing ability - did we reach out to enough candidates to yield the proper amount of subs? Was it our inability to convert interest candidates to subs? Is our comp package an issue? Do we have the proper sell for the company/role? Are we tapping into our BU to influence candidates? Are we up to par with our ability to counter objections? Or are we just playing god and holding an unnecessarily high bar for ourselves.
Sub to Screen %
This is the gauge of whether we understand the req and ability to partner with the manager to generate interest for our subs.
GOAL: We should expect 90% of all resumes submitted to the manager be selected for a phone screen. If we submitted 10 candidates to the mgr, we expect 9 to move into phone interview stage.
Troubleshooting: If interview % are low then we should look at the reasons why the candidates were rejected. Does it reflect our inability to understand the req? How are we with the technologies involved? Are we able to describe in laymen terms the functionality of the job? Is our candidate summary informative and talks to both the strength and weakness of the candidate? Would it be useful if we gathered a few sample interview questions from the team? Or should we have another req intake with the mgr to reassess the role. Is the mgr looking for a purple squirrel and do we need to reset expectations? Could we consider alternative profiles?
Screen to Loop %
True gauge of quality. Are our candidates strong enough to get pass the first round. This number is extremely important because a low % here means we're wasting a lot of engineering hours performing 1st rounds on candidates that never makes it to the main interview loop. Too high a % means we're being too picky and should probably sub more to mgr.
GOAL: Aim for 60-70%. An initial phone screen is meant to technically test the person. The heavy lifting will be done during the loop, which can include many additional interviewers. If we have 9 phone interviews, then 5 to 6 should proceed to the loop.
Troubleshooting: A low phone to loop% could mean the recruiter is not qualifying the candidate enough to screen past what is often an exaggerated resume. Is the candidate indicating tasks/projects that they were not a part of? Could we implement some additional screening on the recruiter side prior to a phone screen? Perhaps the recruiter has a very close partnership with the HM and they allow everything to go thru when it shouldn't. Or is the interviewer being too harsh? Are they asking similar questions to all candidates? Are they asking questions that are not part of the job description? Should we re-calibrate the job?
Loop to Offer %
This stage is a true testament of whether our HM can make a decision.
GOAL: If we provide the team with 4 quality candidates, the they should be able to make an offer to one. A 25-30% at this stage is fairly normal.
Troubleshooting: Anything lower then 25% really needs some serious attention. During the interview debrief session, the recruiter needs to have the ability to provide consultative guidance on how to proceed. Did everyone but one person gave a thumbs up? Is the recruiter able to influence that person to change their mind? How bout bringing in a 2nd opinion to counter that no? Bring data to back up your points.
Offer to Hire %
How strong is our offer compared to the competition? Is our equity, brand, opportunity above, in line, or below the market.
GOAL: This number really differs from company to company, and even groups within a company. I know a specific infrastructure team in one of the FAANGs has a 97% acceptance rate while their SWE is in the mid 80s.
Troubleshooting: If you're clocking less than 75%, we need to really examine whether it's the comp package, the brand, the opportunity? Maybe we need to revisit closing tactics. Are we leveraging the BU to help close? Or perhaps take a closer look at whether we're reaching too high of a talent for the role in mind.
BG Check to Start %
This almost feels like it should be 99%, but people drop off for various reasons from not passing BG, having cold feet, taking other opportunities, or whatever Murphy's Law musters up.
Troubleshooting: A very important calculation is time in stage. On it's own it doesn't really tell a story, but if you combine it with passthrough rates it may paint a clearer solution. For example, let's say your [BG Check to Start %] is 70% and your Time In Stage for BG Check is 3 weeks and immigration is 5 weeks. That's an very long time for a candidate to be hanging out in limbo, and you know what they say TIME KILLS ALL DEALS. Fix the delay in your onboarding process and you'll magically see your starts going up.
Final Thoughts
There's more than one way to skin a cat. What works in one firm won't in another. Be creative and try to simplify your solution as much as possible.
Your metrics are only as good as the laziest person in the company.
The war on talent is so fierce sometimes that we "forget" to status an applicant into different stages.
One of the FAANGs had a perfect solution - they engineered a process that is less reliant on multiple stages, especially those that relies on human updates. Their metrics are simple and only tackles 4 items:
[ 1 ] Reach Outs to Submittals %
[ 2 ] Screens to Loops %
[ 3 ] Loop to Offer %
[ 4 ] Acceptance Rate
Here's their reduced workflow
Go wide or go deep, it's all up to you. Below is a blueprint of the entire TA lifecycle so you can use it to slice up your metrics however you like. Have fun and happy hunting!