I recently spoke with Confirm's CEO David Murray. Confirm uses AI to improve the Performance Management process. In and of itself, that wouldn't be much of a story but what is important is how Confirm frames the current state of Performance Management.
Let's dive right in with the question of real performance vs. the appearance of performance. Murray and I got into a solid conversation about how managers can be misled into giving a better (or worse) performance review. Some great workers are quietly and modestly toiling away doing a great job while also not beating their chest to make sure someone's noticing their effort. Other people may be doing far less work but are constantly in their boss's face creating the appearance of being a hard worker (while doing little or only putting forth the minimum effort).
This problem, sorting genuine vs. perceived performance, is real and troublesome. Many managers only prepare performance reviews once per year. They often forget more about a person's contributions/achievements than they remember. They also tend to confuse the attention they receive from some employees as a sign of productivity from that worker. In other words, many performance reviews are highly flawed due to old data, missing data and the distortion that occurs when emotions/feelings are allowed to sway the data.
To solve this issue, some HR vendors have tried to use other data sources to supplement the manager's incomplete or erroneous perspective. Years ago, when this idea was novel, a vendor used 'heat maps' that showed who were the most influential employees in a firm. These were the 'go-to' people that others sought out for answers/guidance/direction. It didn't matter what the person's job title said. They were the de facto thought leaders, experts and doers within the firm. These are the folks that get things done and they know how to do them, too.
In contrast, political animals, like toadies, yes-persons and hangers-on, don't really do anything except play up to the boss. No one seeks their advice and counsel and colleagues know them to be an empty suit. Or, as they say in Texas, these people "are all hat and no cattle".
Another innovation followed. Performance management vendors started to identify other data feeds that might contain relevant bits of performance data that a manager might want to consider or include in a performance review. For example, emails, text messages and even social media could be mined for compliments, criticisms, thank you statements, expressions of help received, etc. Another source of this data could be customer service solutions as they might have customer feedback comments that speak to the help and knowledge (or lack thereof) given by a customer support person.
Generative AI tools were particularly effective in identifying potentially relevant copy and placing it in a format suitable for performance reviews. What these tools do is provide a more complete view of the employee. And, they can help remove some of the assumptions, biases and personal opinions that might not align with data-driven facts.
In the end, these new tools make for a more fair, insightful and productive review process but it can also produce a number of other positive knock-on effects. These include:
That's great progress but other problems can remain.
A colleague of mine never met a job applicant he didn't want to hire. He also thought every employee deserved an "Exceeds Expectations" in their performance reviews. The real issue is that he was either a people pleaser, a hopeless optimist and/or incapable (or unwillingly) to be objective. He also could have been uncomfortable with giving bad news to people. He wanted to be liked by everyone - unfortunately, you don't get to be a manager if you can't make tough personnel decisions.
The opposite of this is the excessively tough grader (ETG). The ETG is not just a tough critic of a person's job performance but they might also believe that no one can do the job as well as they did/can. Those unreasonable ETG's are being excessively rough because they can get away with it. They'll rate most everyone in the lowest possible performance rankings. A great worker might be lucky just to be rated 'Acceptable'.
Another challenge is the manager that habitually ranks everyone no better than an 'Acceptable' performance ranking. These leaders often do this for two reasons. Either they are doing this to avoid granting larger raises (thus eating into their budget) or to hide their top talent from other managers in their firm. The former rationale really speaks to the manager's poor budgeting/planning capability. The latter speaks to their selfish and lazy approach to management and talent development. They don't want other managers to poach their best and brightest workers even if these other career opportunities might be great for the firm and the employee. In that situation, the manager is selfish, lacking empathy and someone not interested in developing people.
These extremes are possible when little or no supporting data is available to challenge the evaluator's point of view/commentary. This is why tools that collect many relevant data points are so helpful to an organization. Better data and tools provide the company (and the employee) with a truer view of their workforce and its capabilities.
Finally, there's the Ladder. This is the insidious technique where a manager must rank order all employees from best to worst. This mechanism, incorrectly, assumes that only the top one or two people in the ladder are worth lavishing praise, promotions, raises and career opportunities. Those at the bottom of the ladder are candidates for termination. The reality is that the top-ranked people might be terrible or that those at the bottom of the ladder might be pretty good as poor performers had already been sorted out some time ago. Again, a lack of data is what allows poor systems like the above to exist.
No, businesses need more than extra data or proof points. They need to achieve a more balanced view of their workforce and allocate their development energies to more than a few persons.
Murray shared the following bell curve graphic with me. It shows how more than 60% of a workforce get rated 'meets expectations' with small percentages allocated to the extreme edges (i.e., unacceptable and exceptional categories). I'm familiar with this type of distribution curve with some firms mandating that their managers force fit their workers into this scoring distribution.
These allocations are often unfair as you can't fit a distribution like this when you only have, for example, four people in a department. Likewise, a very loyal, senior and experienced team might have a lot of great people and the resulting distribution should represent a top-heavier grouping.
While there are myriad issues with these distributions, Murray had the most piercing observation: Why is a bell-curve the right distribution? Shouldn't there be more similar quantities of workers in each category? He believes poor performance management practices are what produce this bell curve distribution..
These allocations are often unfair as you can't fit a distribution like this when you only have, for example, four people in a department. Likewise, a very loyal, senior and experienced team might have a lot of great people and the resulting distribution should represent a top-heavier grouping.
While there are myriad issues with these distributions, Murray had the most piercing observation: Why is a bell-curve the right distribution? Shouldn't there be more similar quantities of workers in each category? He believes poor performance management practices are what produce this this bell curve.
Murray's pithiest point of the interview was simply this:
A lot of the people in the middle (i.e., "Meets Expectations") belong somewhere else on the curve.
Murray added that if managers had better data then they might identify more people who should go into every category save that for 'Meets Expectations'. More/better data would likely identify more people with great talent even though those individuals may not be as politically savvy or cozying up to their leaders as toadies are. These unspoken leaders should fill more of the 'Exceeds Expectations' and 'Outstanding' categories. Likewise, more of the toadies need to be moved from the higher categories and moved to the 'Needs Improvement' and 'Unacceptable' categories. Murray believes these people are better at 'managing up' than in doing their assigned job well. I'd agree.
One management consulting article described the problem as the 'middle 80% challenge'. This name came about as too many operational and HR leaders only care about the 10% of workers that are either outstanding/high performers or the 10% that are unacceptable and should be on a performance improvement plan (or are actively being counseled out of the company). That middle 80% encompasses a huge portion of a company's workforce and it gets little to no attention.
Worse, the middle 80% are forgotten, overlooked and definitely not getting developed. There could be some real superstars waiting to be discovered/developed except that the company's performance management tools aren't picking up the best data points to identify these people. Firms that only focus on the extremes of the performance dimension are wasting their human resources.
Better data, more data and a more discerning performance improvement process/technology would deliver more acceptable business outcomes. This flattening of the bell curve should improve morale.
Murray and I also discussed Confirm's latest innovations. These included:
Confirm is already integrated with a number of major HRMS vendors
Within my consulting client base, I remain unimpressed with the performance management practices, processes and technologies that many firms have in place. The reviews are too few and too infrequent to really drive meaningful discussions with employees. Moreover, the focus on these reviews is almost always backward focused (i.e., what the employee did/didn't do the last year). A forward-looking focus is often non-existent or poorly thought out. Tools like Confirm should make the part of the conversation where people dwell on the past a shorter one as the content will be more complete, less political and briefer. With that, time CAN be spent looking ahead.
It also bothers me that these imperfect performance reviews are being used as a key input to setting employees' future pay. If more than 60% of employees are getting a middling performance ranking, why should they be happy with this?
Also, if more people get moved out of the undifferentiated blob in the middle (i.e., meets expectations), then a flurry of other benefits should accrue to the employee and employer.
This product can definitely help mid-to-larger firms. One customer of Confirm's has approximately 12,500 employee equivalents. Proof of their market acceptance is visible on their website's customer list.
