Fixing employee performance management (again)?
Revolut, The Office, forced ranking, cicadas, GE and more.
History repeating and all that.
Karl Marx wrote a deft phrase in his 18th Brumaire, dissing Hegel. “Hegel remarks somewhere that all great, world-historical facts and personages occur, as it were, twice. He has forgotten to add: the first time as tragedy, the second as farce.”
Marx wasn’t the first person to comment on history repeating itself, and there have been other adroit renditions: “History doesn’t repeat itself, but it often rhymes” attributed without real evidence to Mark Twain. Many economists make a good living coming up with ever more complex math to remind us to learn from the past. Heck, at its most fundamental an LLM is just a whole lot of regressions on historical data.
Employee Performance Management pops up noisily in the HR Tech discourse, cicada-like, every 10 years or so. This has been going since ancient Babylonians left tablets describing pay for performance. It seemed to involve extra beer. It has been disliked since then, and reinvented every decade or so.
Here’s one from the US army, in the 1940s.
My favourite TV show scene, Big Keith’s Appraisal.
In the 1980s and 1990s General Electric, and especially its CEO, Jack Welch, were deified for their performance management culture. It was taught at every business school. Top employees were well rewarded, and low performers were ruthlessly moved on. Microsoft and many other organizations utilized a similar model. It was called forced ranking or stack-ranking. At one point most large organizations used some form of this model.
Lars Dalgaard brilliantly deployed Jack Welch to position SuccessFactors as the technology to support modern performance management, and turned SuccessFactors into one of the most successful SaaS IPOs.
There are many valid criticisms of the Welch model (see below).
But there is one thing that is well worth learning from Welch. Employee performance is CEO business. Building a high performance organisation is the job of the CEO. Let’s hold that thought a moment.
When I was at Gartner I was mainly critical of forced ranking, and I remain so. The paragraph below is Gartner-speak for this is unlikely to end well.
We saw a pushback from the Welch model, and performance management swung to being all about employee development and engagement. Adobe and then many other organisations dropped performance management scores altogether. It then heralded a new set of HR technology features and vendors, Lattice, Betterworks and many more. We all learned the catchy phrase continuous performance management, and we rediscovered the dangers of the halo and recency effects. Understanding the drivers of employee engagement took on new importance, powered by tools like Peakon and CultureAmp.
Pay for performance fell out of fashion, but it was never completely clear what replaced it. This phase of it’s all about development happily coincided with robust hiring.
The history of performance management continues to repeat itself.
Revolut
Thanks to a Linkedin post from Martin Mignot of Index Ventures, I read through the performance management playbook of the Revolut CEO and co-founder, Nik Storonsky. It makes fascinating reading. It is worth a deep perusal. Quantumlight have done a brilliant job in making this accessible. Check it out here (but please don’t do your performance appraisal in googlesheets though).
Revolut is one of Europe’s most impressive neo-banks.
According to Reuters.
“The nine-year-old fintech business has now been valued at $45 billion after its employees sold shares to investors, surpassing in terms of worth some of Europe's biggest lenders, including Britain's 334-year-old Barclays (BARC.L),and NatWest (NWG.L)”
It is a remarkable business. Congratulations to the founders and employees, and well played early investors too.
My thoughts on their model: lots of questions and comments.
First, a caveat, I’m going by a couple blog posts, the Quantumlight notes and bit of Internet sleuthing, I’ve not had a proper demo, or talked to anyone from Revolut. I’m not expecting answers either, just pondering out loud.
Storonsky has spun performance management out of HR, and made it a seperate function, reporting directly to him. I’m not sure if this is a merely a reflection of the importance of performance management, or a sort of disdain for the HR function. But having a team that is very senior, and really focused on figuring out what makes a high performing organization in your company’s context makes a whole lot of sense. Reporting lines are always contentious, but the key point here, like with Welch, is that Stronsky sees managing employee performance as a big part of his job. I like this a lot.
Looking through the methodology explanation and the product demo, I can’t help thinking that this is forced ranking all over again.
Exhibit A: Revolut.
“A high-performance organization is A-player centric (top 15%-25%), focusing resources to retain and promote top talent while exiting underperformers as fast as possible.”
Exhibit B: GE.
“It rated workers on a three-step scale, based on individual goals and performance. The top 15% are told they are "1s," the middle 75% are designated "2s" and the bottom 10% are assigned "3s," according to the Wall Street Journal. And the bottom 10% got canned.”
The Revolut model uses this framework.
Unsatisfactory: Indicates performance that does not meet expectations.
Developing: Indicates performance that is below average and needs improvement.
Performing: Indicates satisfactory performance that meets expectations.
Exceeding: Indicates performance that surpasses expectations.
Exceptional: Indicates outstanding performance that significantly exceeds expectations.
The labels are mix of static and continuous (ing), which isn’t really coherent. Developing is potentially misleading, as an employee could move from performing down to developing, which would mean they weren’t developing, but regressing. GE’s model is cleaner, as is the Army model from the 1940s. Was there a scientific reason for the 5 point scale?
How do you manage calibration between departments? Department size variance can make this really awkward. A manager with 12 reports has a different aggregate scoring outcome than a manager with 11 reports will have.
I didn’t see a concept of team or department evaluation. The implicit assumption is that all teams are performing equally, this isn’t really the case in reality. Performance isn’t really Gaustian. One team may have 70% low performers, another 70% high performers, when measured “objectively”.
How do you cope with bias, as the answers in playbook didn’t cover gender bias for instance. See Textio’s research. and Joshi’s paper.
How do you objectively and scientifically evaluate managerial skills? The playbook is thin on this.
I didn’t see a concept of potential in the model. Do you promote only on current role performance?
Why do you dismiss 360 degree feedback? “this review process is not a 360 review, which typically offers limited benefits at the expense of employee time - we found that employees tend to be too generous with each other, so 360 reviews usually don’t provide useful feedback.” I can’t see how one can objectively assess manager performance without asking their employees and peers. There is a good bit of academic research supporting 360.
Some of the most important researchers in management theory have been strong in their critique of forced-ranking, Cappelli, Lawler, Ulrich, Weibel etc. Did you consider the research when you developed your approach?
This paper provides an excellent overview of the ratings v no-ratings debate.
About applications.
UX design and screen resolution has moved on. Modern UX has so much more pixel real estate than 15 years ago.
The Revolut solution looks of the moment from a UX perspective. Having built significant consumer apps definitely carries through. Well played, the software looks slick and aesthetically appealing. If Farrow and Ball did UX, it would look like this.
But I can’t help thinking the functional scope is pretty much identical to what SuccessFactors was doing 15 years ago (It is weirdly hard to find old screenprints, but here’s one).
This one is from about 2010.
2011 or so (note cool ‘AI’ writing assistant feature).
My conjecture
I don’t have any scientific backing for this, but it would make for interesting research. Steve Hunt?
Remember, forced ranking worked for Welch for a while. GE was bloated, slow and smug when he took over. He had to blow it up. Neutron Jack needed a method. He found one, and used it to turn GE into the world’s most valuable company. It then stopped working.
Forced ranking may be a useful, if painful exercise for later stage pre-ipo companies that have scaled very dramatically. They may have over-hired, especially if they had funding largesse under ZIRP. To be IPO ready, they need to show sustainable growth and discipline.
Done right, it enables them to reset organization structures, size and even culture, while retaining top performers. It sends a grown-up message to the investors that you are “operationally disciplined” now. But it is a short-term intervention, for 3-4 years, beyond that you end up with a toxic culture, entrenched biases, dehumanisation and highly gamed system that destroys shareholder value.
It is organizational surgery.
Closing thoughts
I’m thrilled to see a CEO / Founder so engaged with performance management. I wish more took it so seriously.
I’ve some concerns with the approach (based only on what I’ve read) and I hope that their team have learnt from past failings of forced ranking. It may well be an appropriate approach for Revolut today, after all they have moved to profitability with this model. I’ll be at Slush in a couple of weeks, so I’ll get to hear more about the Revolut processes then.
I don’t think there is one correct way to do performance management. Humans and organizations are complex. What works for a time, stops working. Check the evidence, build on the research. Remember that human motivation is complex.
I’m wondering too, how AI can make performance management fairer and better. Our AI overlords are helpfully assessing our performance already, far more so than many employees can imagine. I vacillate between despair and excitement. I’ve seen a couple of exciting startups tackling the thorny problem of improving management performance. I’ve also seen some egregious examples of AI over-reach.
We can strive to make performance management better. Software will play a role, as will robust research, but having CEOs that genuinely take accountability for employee performance is the best starting place.
And History, with all her volumes vast,
Hath but one page,—'tis better written here- (Byron needs a mention too)
I’m also not convinced companies should be building their own HR Tech, but that is for another day.
As usual, I’ll end the post with a song. It is a banger. Shirley Bassey…. History Repeating…
So many thoughts, largely in agreement with your views. I think too many organizations don't have a real performance management structure as a result, people don't really understand performance expectations. This a a leadership issue and your point about it starting with the CEO is critical.
The Resolut approach is interesting but there are a few areas where I have problems with it.
One is absolving management of their performance management responsibility and putting it in the hands of an independent part of the organization. The job of every manager is to maximize the performance of each person on their team. As a result, they have to sit down with the person to establish performance expectations, review, coach, evaluate. This cannot be delegated.
Second, the approach tends to think of performance management as a static thing. They don't address the role of managers in coaching and developing their people to maximize the performance. What do you do to prepare an A player to continue that performance and be prepared to step into a more senior role? What do you do to coach and develop a B player to be A players. What do you do to coach a C player to improve their performance becoming a B player.
When we look at performance failures, it's partly a reflection of the performance failure on the part of management. It's not just on the employee.
I'll stop here. A good article, mostly provoking thinking about performance management, whether one agrees or not.