Jack Welch was one of the most successful CEOs in history. He took GE's market value from $14 billion to $410 billion — roughly thirty times over. That's not a typo, and it's worth sitting with for a second. One of the tools behind that run was something now called the Performance-Values Matrix, which he used to rate people on values and behavior side by side with performance.
Here's the pattern I keep noticing in successful teams: they're stacked with people who are both high-value and high-performing. Not one or the other. Both.
Check your team's performance
The Jack Welch matrix is refreshingly simple to run. Leaders score their direct reports on two separate 1-to-10 scales — performance on one axis, values on the other — and plot each person on a grid. Performance measures the impact someone has on the organization. Values measures how well their behavior lines up with what the organization says it stands for.
Two questions, two scores, one picture of your team.
How to interpret the results
Four combinations fall out of this:
| 1. High performance and high values | 3. High performance and low values |
| 2. Low performance and low values | 4. Low performance and high values |
Group 1 is the easy one. Keep them, promote them, invest in them, reward them — they're doing everything right.
Group 2 sits at the opposite end, and honestly, it's just as easy a call, even if it doesn't feel that way in the moment. Poor performance and misaligned values together don't leave much room for debate.
Group 4 — high values, low performance — is worth keeping too. These are usually people who just need a bit of coaching or investment to catch up on the performance side.
Then there's group 3. This is the hard one. High performers who don't live the values. I've seen leaders argue both sides of this: some say you let them go regardless of the numbers they put up, because the damage to culture outweighs the wins. Others can't stomach cutting someone who's clearly delivering. My honest read is that there's no universal rule here — it comes down to the actual score and, more importantly, whether the person shows any real sign of being willing to improve.
VBM: the second values matrix for management
There's a second model worth knowing about, called Values-Based Management, or VBM. The premise is straightforward: put your company's stated values into practice, and performance should follow, because clearer values make for more effective processes and better odds of hitting your targets.
A study out of Bradley University set out to test this directly, comparing VBM against the Performance-Values Matrix (PVM). Researchers started with 125 executives at a large production facility, and after excluding those without complete records, ended up basing the analysis on 103 managers, checking for a real connection between how people behaved and how they performed.
They found a correlation. But it was too weak to say VBM was actually doing what it claims to do. Based on the data, PVM came out as the more reliable tool for gauging staff performance.
Why the gap? To answer that, you have to look at how the two models are actually built, and how the study itself was run.
VBM assumes something fairly direct: the more committed a manager is to company values, the better they'll perform. Behave right, and performance follows. The model essentially argues that no management decision should be made without first weighing its impact on values.
PVM and VBM: comparison and contrast
The core difference is this: PVM treats performance and values as two separate things to measure. VBM folds them together.
To see why that matters, it helps to distinguish between "values enactment" and "values match." Enactment is behavior that lines up with company values. Match is performance that lines up with them. VBM leans on enactment. PVM treats values as a match — essentially, how closely someone's actions track the company's stated principles.
My own take, and I think the data backs this up: measuring the two separately is simply more honest. It lets you tell the difference between how a result was achieved and what the result actually was. Those are not the same question, and conflating them hides more than it reveals.
Researchers at Bradley University carried out a study to test exactly this — whether values and performance are genuinely separate constructs, or just two ways of measuring the same thing.
They ran it at a company that had recently gone through the process of defining its core values: Commitment, Integrity, Teamwork, Customer Satisfaction, and Mutual Respect. Managers had already been trained on what those values meant in practice.
Measuring core values
The university team and the company's HR department built an assessment together — a Behavioral Observation Scale, or BOS — made up of 19 items designed to measure values enactment.
Then came 360-degree feedback: each manager was rated not just by their supervisor but by peers and reports too. In this study specifically, the feedback came from direct reports, the manager's own supervisor, and five peers.
Raters used a 5-point scale to judge how often a manager actually demonstrated each value.
The scale ran in roughly 20% bands — a 1 meant the behavior showed up less than 20% of the time, rising through each 20% threshold from there, up to a 5 for more than 80% of the time. Demonstrate a value consistently, and you'd land a 5.
From there, the team averaged the scores for each item.
Measuring job performance
Job performance came from a different source entirely: each manager's most recent annual review, scored against 17 job-critical criteria.
The rating scale looked like this:
- Doesn't meet expectations
- Below expectations
- Meets expectations
- Above expectations
- Well above expectations
To compare performance and values, researchers pulled from two separate assessments — the Core Values BOS on one side, the annual review on the other — completed by different groups of people. That separation wasn't an accident. It was there specifically to keep bias out of the picture.
That produced three distinct correlation measures:
- Annual performance review
- Core values assessment
- Comparison of the above two
Findings in favor of the PVM as the better matrix
Here's the number that stood out to me: researchers checked over 300 potential correlations between performance and values enactment. Only 12 turned out to be statistically significant. And given a 95% confidence interval, the researchers themselves flagged that even those 12 could just be noise.
Put plainly — performance and values don't move together in any reliable way. They need to be measured as separate things, and that's exactly what the PVM is built to do.
Takeaway
In my view, this is the clearest finding in the whole study: the Performance-Values Matrix is simply the more precise tool for measuring how managers and staff are actually doing. VBM doesn't hold up as an accurate measure of performance on its own. It might be worth thinking of it as a holdover from an older way of running organizations, one that hasn't quite caught up.
Decision matrices for businesses
Step back from values for a moment, and there's a broader tool worth knowing: the decision matrix. It's genuinely useful for solving problems, prioritizing tasks, and building a case for a decision you've already made. It also helps when you're choosing between different approaches to how your business operates.
A decision matrix is just rows and columns of values that let you compare options by weighing variables according to how much they matter. If you're facing a decision with enough moving parts that gut instinct alone won't cut it, this is the tool. It turns a messy call into a structured one.
University professor Stuart Pugh created the decision matrix method, originally to help choose between design alternatives. It's since spread well beyond design and into business generally. You'll also hear it called grid analysis, the Pugh method, or the multi-attribute utility concept — different names for the same idea, which is to strip subjectivity out of a decision.
What I like about the Pugh method is that it surfaces the factors actually driving a decision and cuts through the noise around them. It's a quantitative approach, which means it can push emotion out of the equation entirely. You assign a real value to each variable and weigh it properly — a step up from a simple pros-and-cons list.
That said, it's a rational tool through and through. Use it for decisions that should be unemotional. It's the wrong tool entirely when the choice comes down to personal taste or preference — there, removing emotion isn't a feature, it's a mismatch.
Example of a Pugh matrix
| Concept A | Concept B | Concept C | |
| Criterion 1 | S | S | + |
| Criterion 2 | S | S | + |
| Criterion 3 | S | + | - |
| Criterion 4 | S | + | S |
| Criterion 5 | S | - | + |
| Total + | 0 | 2 | 3 |
| Total - | 0 | 1 | 1 |
| Total score | 0 | 1 | 2 |
Here's a filled-in Pugh Matrix comparing three concepts, A, B, and C, listed across the top and judged against five criteria. Concept A is the baseline — every one of its criteria gets marked "S," for "same." Concepts B and C are then scored against A on each criterion: a "+" if they're better, a "-" if worse, and "S" if it's a wash.
The process of building a Pugh matrix
Assuming you've already got your alternative options defined, building the matrix itself happens in four stages.
Stage 1
Start by identifying and defining your selection criteria. Design requirements — in full or in part — usually work well here, so long as they account for the user and any other key stakeholders, internal ones included. The reliability of everything that follows depends almost entirely on getting this set of criteria right.
It's tempting to rush through this stage. Don't. A weak set of criteria here tends to produce an unreliable pick and an outcome nobody's happy with later.
Stage 2
Pick one option as your baseline and mark every criterion "S" for it. If you have a previous design to draw on, use that as the baseline — you'll already have a good feel for how it performs.
An alternative here is a 1-to-5 scale, where "S" (the baseline) sits at 3. A 1 is much worse, a 2 somewhat worse, a 4 better, and a 5 much better than baseline.
Stage 3
Tally the pluses and minuses for each alternative. Whoever comes out with the most points is technically the "winner" — though I'd treat that as a strong signal, not gospel. Don't crown the top-scoring concept without at least pausing to sanity-check it.
Stage 4
This one's optional, but worth doing if you've got several alternatives on the table. Rarely is there a single clean best option in that situation — it's usually worth building hybrids that combine the strongest elements of each. Here's what that looks like:
| Concept A | Concept B | Concept C | Concept AB | Concept BC | |
| Criterion 1 | S | S | + | + | - |
| Criterion 2 | S | S | + | S | + |
| Criterion 3 | S | + | - | + | + |
| Criterion 4 | S | + | S | - | + |
| Criterion 5 | S | - | + | + | + |
| Total + | 0 | 2 | 3 | 3 | 4 |
| Total - | 0 | 1 | 1 | 1 | 1 |
| Total score | 0 | 1 | 2 | 2 | 3 |
Notice Concept AB ties with C on total score. But BC pulls ahead as the strongest option overall, with the highest score on the board.
Example of decision-making using a Pugh matrix
Picture an entrepreneur weighing up three locations for a new business. He lists the factors that matter — rent, location, and access to good employees — and weights each by importance.
Market share matters most to him, since it drives how many customers walk through the door. He also wants somewhere close to home, so he can get there fast if something goes wrong. And he wants to be near a labor pool he can actually hire from.
Once he runs it through a Pugh matrix, one location clearly comes out ahead. It offers the best shot at customers and qualified staff — enough to justify the higher rent on its own, on top of the best overall score.
That, to me, is the real value of a decision matrix: it doesn't just help you pick — it makes the reasoning behind the pick visible, which is half the battle when multiple options and variables are in play.
Sources
Discover your own core values
Take the free, research-backed Values App assessment — nine minutes, max.
Discover your values