Death to the Merit-Pay-Increase Matrix!
After hating the merit-pay-increase matrix for 25 years, imagine my delight in seeing that Saba plans to debut a Big Data and machine-learning product that encourages you not to treat every employee the same, tells you who needs to be treated differently and even indicates how much the good ones may cost to keep them from leaving!
By Bill Kutik
I have hated the merit-pay-increase matrix for 25 years.
As you likely know, it is the paper form (spreadsheet or compensation application) that companies use to control growth in wages (except for those at the top, of course) and for HR to fulfill its traditional role of treating everyone the same. Ironically, often without regard to merit!
You've probably done one before. All your direct reports are in the first column, with their salaries in another and (if the apps are properly integrated) their performance-review number in a third. In a fourth column, you give employees percentage increases in base salaries because of their review number or maybe on how you're feeling that morning. All your increases must total a company-wide standard: lately, 2 percent or 3 percent or, in some golden past, as high as 5 percent or 7 percent.
If your total goes over the standard, you have to give someone less money to get the percentage down. If you have a great reason for not complying, the rules generally require the approval of someone incredibly senior, even the CEO. In other words: Best not bother.
Behind this are more pernicious rules. Every salary band has a midpoint, often expressed as a compa-ratio. The idea is to drive everyone's base salary as close to the midpoint as possible. Which means, in practice, giving employees below the midpoint larger percentage increases (often regardless of merit) and those above it smaller increases despite their merit. Workarounds include giving short- or long-term incentive payments in place of a salary increase.
Is this any way to run a successful business in 2014?
Not really. Talent experts have been telling us for years to identify top performers, incent them like crazy, and watch them every second so competitors don't steal them away. In short, treat them better than you treat everyone else, not the same way.
HR software today rarely embodies that modern point of view. HR people and managers may talk the talk of HI-POs and all the rest, but compensation software doesn't really walk the walk. Certainly not the merit-pay-increase matrix, where salaries and bonuses are specifically determined.
(Background on my attitude, if you care: I came out of daily print newspapers, where union contracts determined salaries up to five years of experience, and after that it was all about finding great stories that sold newspapers. Never met an HR person after signing my payroll form. I remember a managing editor at the New York Daily News giving me a bonus on the spot for a front-page scoop that beat The New York Times. Talk about pay for performance!
(My only job in trade journalism (business-to-business, rather than to consumer) was starting an HR tech magazine for what's called a professional publisher. Suitably there I was handed my first corporate merit-pay-increase matrix form. Once HR explained it to me, I waved it in front of the small and very young staff and said, "This is keeping me from paying you what you're worth and what you deserve. Everyone's assignment is to study it and figure out how to break it!" Of course, they couldn't, but it really motivated them to learn HR, which was a major job requirement.)
So everyone is concerned about identifying their high performers and keeping them from leaving. A lot of vendors now claim to do that through analytics or predictive analytics. At least 15 years ago, my old friend Row Henson presented PeopleSoft's first workforce-analytics package. A huge, powerful, unstructured beast that maybe 20 customers bought and fewer than five tried to implement and use.
But it identified who was a flight risk! Wow, ahead of its time. I asked, "How?"
"It tells us when their options have vested," Row said shamefacedly, knowing there was so much more. A Silly Valley solution.
Though Saba lives in the same valley, it has come up with something not at all silly for this critical modern problem, which it plans to debut at the HR Technology ® Conference. Recently, I got a sneak peek and the first demo given to an outsider.
First, Saba smartly puts the new solution in its Compensation@Work module so comp administrators can create a merit-pay-increase matrix. Later, managers will have direct access to it.
Second, it uses Saba's new Big-Data and machine-intelligence platform called TIM, or The Intelligent Mentor, to offer concrete identifications and suggestions for corrective compensation actions for each employee. Naturally, they can be ignored or over-ridden for that gut feeling so many managers claim to have.
And the demo actually worked! Not always true for early-stage products. That doesn't mean the final product will, but it's a strong sign months before it is generally available at the end of November.
TIM performs this magic by examining three big buckets of data about all employees, currently labeled @Risk, High-Value and Under-Paid. While much of that data can be imported from your HRMS, obviously some of it has to be bought from third parties (such as salary surveys or specific job demand by geographies) and still loaded manually. TIM doesn't Google for you.
Each bucket has multiple data points. @Risk, in fact, has as many as 20-plus, and uses various algorithms on them (Saba likes to call them "signals") to calculate unique weighted values for them. Obviously, those are the jewels in the crown for TIM and Saba, but here are two points for each of the three buckets:
* @Risk: skills in demand, last promotion;
* High-Value: influence on company social network, "critical person";
* Under-Paid: market trends, peer comparisons.
The system looks at each company's employees and builds complex relationships among the signals. Based on how each company defines the weights of each signal, TIM defines each of the three buckets for every employee.
Obviously, what you want to know are the employees at the intersection of all three: the highly valued people who are a high flight risk, and are being under-paid.
That's exactly what TIM does. You get a page of their pictures with their names in the second column, current pay in the third, cost to replace them (always staggering) in the fourth, and "Proposed Retention."
Yes, TIM recommends exactly how much more to pay them to get them to stay! Not sure it can identify the terrible boss making the employee leave no matter how much money is offered, but that's the job of analytics, not compensation.
So Saba likes to claim TIM is so individualized that two people in the same job at the same company at the same level and in the same location may not get the same recommendations.
Historically, Saba has served mostly large customers, so it generally does not deal in disruptive technology, though TIM could be very disruptive to current best practices. So there are multiple human interventions possible before or after the recommendations, including checking out the employee's compa-ratio (thus preserving the sanctity of the midpoint!) and determining whether the proposed retention money should be paid as a base salary increase or one-time incentive comp (the traditional way around the midpoint rules).
Smart and politically correct for Saba to do that. Hopefully the more fully-baked product will also expose more of the data and reasoning behind TIM's recommendations: required for any end-user to take them seriously. Nobody wants to trust a black box anymore. We even know how our GPS works!
Still, hats off to the first HCM system I know of that not only tells you who needs to be treated better, but also how much better. Once that's adopted widely, it's death to the merit-pay-increase matrix.
HR Technology Columnist Bill Kutik is co-chair emeritus of the 17th Annual HR Technology ® Conference & Exposition, returning to Las Vegas, Oct. 7-10, 2014. Listen to The Bill Kutik Radio Show ® to get a code for a large discount. You can comment on this column at the Conference LinkedIn Group, which doesn't require prior or future conference attendance to join. He can be reached at firstname.lastname@example.org.