• Ginssuart (unregistered)

    Frist

    Had it not been for "but she also made the system more observable", I would have sworn this was about the company I work for.

  • (nodebb)

    You'd like to think that only one level of indirection - gathering information so that they can see where they can improve the metrics - would be simple enough for metrics-driven managers to be able to understand. The fact that it's too complicated for them really explains the overall state of industry.

  • Vera (unregistered)

    Zane must have mistaken the factory for a particularly messy game of Cookie Clicker.

  • COBOL Dilettante (unregistered)

    Reminds me of a workflow "dashboard" (i.e. Excel spreadsheet) requested by an MD that we used in a previous role. These tracked "cases in" versus "cases out", and if the two numbers were the same, then the team was on track.

    Of course, there was nothing to indicate whether those numbers referred to the same cases ... 4 cases in and shoved under the carpet + 4 cases finally worked after sitting round for three months = on track.

    The MD moved on after a few months. None of his successors looked at the "dashboard", but filling it in was now A Task, so it still got completed for at least two more MDs ...

  • (nodebb) in reply to COBOL Dilettante

    These tracked "cases in" versus "cases out", and if the two numbers were the same, then the team was on track.

    So if there's a backlog of cases (seems that there was), and something happens so that N new cases come in and N+M cases are finished (M > 0), the team is not on track???

  • (nodebb) in reply to COBOL Dilettante

    "MD" == "Metric Destroyer"?

  • (nodebb)

    Given the reality that even in the absence of changes to the production system the widgets/day or widgets/dollar metric is going to "breath" up & down due to random factors, how does management react to that reality?

    Do they panic each day, take credit when it ticks up and issue scathing blame when it ticks down, or ignore the daily numbers completely and only look at the monthly average? I'm sure it'd be interesting to know which it mostly is. And which manager from Zane to the Owner or CEO does what.

  • (nodebb) in reply to dpm

    @dpm ref

    "MD" == "Metric Destroyer"?

    If that's a serious question, it's almost certainly "Managing Director." In US large companies that's the typical label for the position just below VP (or regional VP) of whatever. In other words, somebody about 4-5 levels removed from the people who actually move the cases.

  • (nodebb)

    Sounds like she should have just deleted the CV system. Zero defects, nothing to count, better metrics!

  • Mirko (unregistered)

    Cue Goodhart's law (for those who don't know it: "When a metric becomes a target, it ceases to be a good metric").

  • Officer Johnny Holzkopf (unregistered)

    Welcome to the СОВИЕТ НАИЛ ФАКТОРИ, comrade Numbers Go-Up.

  • (nodebb) in reply to Steve_The_Cynic

    So if there's a backlog of cases (seems that there was), and something happens so that N new cases come in and N+M cases are finished (M > 0), the team is not on track???

    Maybe. I agree that managing that the units out keep up the with units in is a real metric, but so the concerns are real too. So, you add "SLA violations" as an additional metric and agree on an SLA for start-to-end on a work item basis. Then, if you are keeping up on flow, but still accumulating work items that take too long, then either the team is cherry-picking easy stuff and leaving the hard stuff, or some things genuinely need more attention and the process put in place to address them isn't sufficient.

  • Former Intern (unregistered)

    Oh man, as soon as I saw the word 'PLC' I thought this was going to be some ladder-logic WTF. I was a compsci co-op student in the late 90s doing that stuff on Siemens PLCs, and... my god, whoever invented the ladder-logic 'language' should be tried for crimes against humanity

  • COBOL Dilettante (unregistered) in reply to Steve_The_Cynic
    So if there's a backlog of cases (seems that there was), and something happens so that N new cases come in and N+M cases are finished (M > 0), the team is not on track???
    Well ideally it should have shown that the team were exceeding expectations, but I suspect what *actually* happened was the spreadsheet showed a nonsense value, and someone manually overwrote formulae until the conditional formatting went green
  • pah (unregistered) in reply to Former Intern

    The "Ladder Logic Language" is just an emulation of physical relays and wires, which is how those controls were implemented before PLCs existed.

  • Acronym (unregistered) in reply to WTFGuy

    Obviously they point to the time it went up, say that is what should be normal and look for someone to scream at and demand answers why that number isn't THE number all the time.

  • (nodebb)

    It sounds like what Rachel needed was a Slow Down Loop -- https://thedailywtf.com/articles/The-Slow-Down-Loop

    Addendum 2026-09-02 14:02: (Alternately, a Speed-Up Loop would do the same thing. Just from a different perspective. https://thedailywtf.com/articles/The-Speedup-Loop )

  • Hmmmm 🤨 (unregistered)

    It seems absurd to not observe that successful production rates are directly at odds with failure/reject rates. On the other hand, I could envision a case where a manager already knows that the CV algorithms have been chased to the ground...

  • (nodebb)

    Maybe.

    Sometimes you can run faster but that will also create more defects.

    SO the number of widgets started goes up, the number of defects created goes up, but the number of good widgets out also goes up. Not as much as we'd wish for, but still up. In that case jo just looking at a metric of output quantity that's a win. It's not a win on widgets out per dollar of raw materials in. But it may well be a win on overall value of good widgets produced versus the fixed and variable costs of the factory.

    Of course if you run faster and the end result is there's more increase in defects than starts and the net result is fewer good widgets, that's a loss no matter which metric you look at.

  • (nodebb) in reply to pah

    The "Ladder Logic Language" is just an emulation of physical relays and wires, which is how those controls were implemented before PLCs existed.

    Yup, I can confirm this. Many years ago (1989/90), I worked with folks who worked regularly with PLCs, and this was exactly what they told me.

  • Conundrum (unregistered)

    "Suppose you run a car delivery business. Your employees deliver cars to their destination, one car at a time. Sometimes, they drive into a brick wall and ruin the car; when that happens, you give them a new car, and have them deliver it instead. You record the number of cars delivered, but not the number of cars crashed, since a small amount of crashing is an acceptable writeoff."

    "Now, suppose I propose a new route, through this active construction site in the middle of town. It increases your throughput enough that you can deliver two more cars per day. But it also causes ten more crashes per day."

    "Since you only track delivered cars, this seems like a win: I improved your number by two. But what you miss is that I didn't actually improve it; I actually decreased it by eight, it's just that the system only records half of the data. And because it catches the +2 but ignores the -10, the number goes up... but the profit goes down."

  • (nodebb) in reply to Conundrum

    Sure. But, on the other hand, what if you are validating green clay before entering a kiln and all of the rejects simply get thrown back in the raw material hopper? Then anything that increases the rate of good output is a win no matter how much bad output is created.

    Adjust your metric to the work being done and don't worry about this abstract "what if" stuff that might or might not apply.

  • (nodebb)

    'Zactly. My point was that "first glance" or "rule of thumb" thinking is usually wrong. Which is why managers use it almost exclusively.

    Those of us who are paid to think things all the way through logically and also understand something of business should know to reject slapdash thought and the slapdash results thereof.

  • Brad (unregistered)

    Sad situation, but her approach was great. I use this one myself - I call it the "McHappy Meal Approach". Bundle up a shiny toy with some food. You can feed them whatever you want while they are focussed on the new shiny thing.

  • 516052 (unregistered)

    The more layers of management there is the worse things are. Plain and simple.

  • (nodebb)

    Replace "widgets" with "trouble tickets" and you pretty much have the same discussion.

  • (nodebb) in reply to WTFGuy

    For a small company, MD and CEO are the same person. For a big company, there's potential for all sorts of seat warmers to be stuffed in around and between them.

Leave a comment on “What You Measure”

Log In or post as a guest

Replying to comment #:

Log Out

« Return to Article