Edward Deming and his system of profound knowledge come to mind here. The realization that many dashboards simply result in what Deming calls “tampering” by management. And unfortunately the unexplained wiggles are some of the most tempting data points for this…
That seems like one of the places where a lot of tech stuff *isn't* like operations though. As I understand it (which, admittedly, is somewhat superficially), Deming mostly focused industrial production problems (eg, how do we improve this factory?). In a lot of the tech world, and especially at smaller companies, tampering by management is the entire point. Companies can't but tamper.
Ya - I guess I am going with a little bit of an arbitrary definition of tampering. I'm thinking of it as activity for activity sake "because we have to do something." When the best decision is to do nothing.
Yeah, you could make the case that a lot of tech employees' jobs are basically that: Hired to shuffle stuff around on nothing but faith that it helps somewhere.
I do think Cedric's piece makes a pretty good effort at answering that, which is that our work dashboards aren't, as much as I hate the word, "actionable." We see the wiggles, and they're really hard to interpret, so we don't know what to do. Though the one gripe I'd have with his piece is that, while it's true the wiggles are hard to interpret, the real problem is that the wiggles are *hard to change.* We don't have any direct ways to move those charts. And even if XmR charts are a good tool for interpreting the wiggles, they don't address the fundamental problem that cause and effect are really far apart. It's easy to know what to do to make the gas gauge go up, or the apple watch rings to fill up, or whatever. But how do you move the business charts? We have no idea.
Insightful. Perhaps the broken link in the analytical value chain is not in the stack at all but in the habits of users. Is there hope this confusion will be resolved as the analytics discipline matures or will products step in to simplify and guide users to what they really needed from all those dashboards in the first place?
I think so? My guess is that it all eventually becomes mechanical and kind of boring? I've said before that I think it'll eventually be kind of like that. And if that's a little much (it doesn't seem like there will a data CPA), it seems like we at a minimum be more like engineers, where there's kind of a standard pattern for how to do it, and a bunch of generally accepted best practices and stuff. Right now, there's a lot of freewheeling and people trying stuff. That's probably good for a time, but not forever. Circle of life, and all that.
Edward Deming and his system of profound knowledge come to mind here. The realization that many dashboards simply result in what Deming calls “tampering” by management. And unfortunately the unexplained wiggles are some of the most tempting data points for this…
That seems like one of the places where a lot of tech stuff *isn't* like operations though. As I understand it (which, admittedly, is somewhat superficially), Deming mostly focused industrial production problems (eg, how do we improve this factory?). In a lot of the tech world, and especially at smaller companies, tampering by management is the entire point. Companies can't but tamper.
Ya - I guess I am going with a little bit of an arbitrary definition of tampering. I'm thinking of it as activity for activity sake "because we have to do something." When the best decision is to do nothing.
Yeah, you could make the case that a lot of tech employees' jobs are basically that: Hired to shuffle stuff around on nothing but faith that it helps somewhere.
Ya… kind of crazy when you think about it.
I have thought a lot about simple metrics vs flashy data insights recently.
The data/dashboards that make a difference in my personal life on a day to day basis are never “insights.” Here is a short list.
1) fuel gauge on my car - I get gas when it’s low
2) apple watch 3 rings - try to go on a walk when my exercise ring is low
3) scale - try to adjust my exercise to calorie ratio when the number gets higher
No amazing insights from data - just numbers with clear actions to take to change them.
As has been mentioned - accountants have this stuff nailed down pretty well. Not sure why it’s taking the rest of us so long… 😁
I do think Cedric's piece makes a pretty good effort at answering that, which is that our work dashboards aren't, as much as I hate the word, "actionable." We see the wiggles, and they're really hard to interpret, so we don't know what to do. Though the one gripe I'd have with his piece is that, while it's true the wiggles are hard to interpret, the real problem is that the wiggles are *hard to change.* We don't have any direct ways to move those charts. And even if XmR charts are a good tool for interpreting the wiggles, they don't address the fundamental problem that cause and effect are really far apart. It's easy to know what to do to make the gas gauge go up, or the apple watch rings to fill up, or whatever. But how do you move the business charts? We have no idea.
Insightful. Perhaps the broken link in the analytical value chain is not in the stack at all but in the habits of users. Is there hope this confusion will be resolved as the analytics discipline matures or will products step in to simplify and guide users to what they really needed from all those dashboards in the first place?
I think so? My guess is that it all eventually becomes mechanical and kind of boring? I've said before that I think it'll eventually be kind of like that. And if that's a little much (it doesn't seem like there will a data CPA), it seems like we at a minimum be more like engineers, where there's kind of a standard pattern for how to do it, and a bunch of generally accepted best practices and stuff. Right now, there's a lot of freewheeling and people trying stuff. That's probably good for a time, but not forever. Circle of life, and all that.