4 ms·

Amazon's success at being a grinding place but reaping major business results align with what I've seen of my previous employer (a transportation icon) that tig

Amazon's success at being a grinding place but reaping major business results align with what I've seen of my previous employer (a transportation icon) that tightened the screws, re-engineered processes, encouraged thousands to leave. Revenue, profit, stock price, and bonuses soared. All of what we've been told about the war for talent, desire for caring work places, etc. seems to be exaggerated. Companies can literally have a policy of "be happy you have a job, punk" for professionals for YEARS, and people will suck it up. Morale does have an impact but not as much as you'd think once your company is large enough AND it is successful. One commonality in both companies is that they are very data driven. Eliminating a costly perk? Do it and see what the data says. Make a change to a major process that people are resisting? Do it, fire any insubordinates, and see what the data says. Basically don't take management policy or process for granted unless you have data to back it up. There's a line between tough and abusive, however, and this seems to be where Amazon's (and other tough cultures) walk the line.

jaawn 11y ago
The data-driven management you're describing is based on the significant assumption that you are able to gather accurate data, and interpret that data accurately. For example, if you are making a management decision based on how much time employees are spending on X, you are assuming that you actually know how much time is being spent on X and what the value of that time is. Simply measuring data is not sufficient. You have to measure the right data in the right way for your stated purpose. Companies operating like Amazon does (apparently), are wildly over-confident in their data collection and interpretation abilities, in my opinion.
- parasubvert 11y ago
  Most companies I have consulted or work for make decisions based on politics and conjecture, with data being invented at a bare minimum to back the argument up, but no one doing serious analysis of it. The culture of having data collection and an ability to challenge it (whether it is flawed or not) indicates a desire for a fact-driven culture vs. a power-driven culture. They might be over-confident, but one would think that reality in the market will eventually catch up to that.
  - jaawn 11y ago
    This is entirely possible. However, when the data being measured is about complex human behavior (i.e. time allocation, productivity, work habits, etc...), I don't think we are even a little bit close to accurately representing reality. If we were "pretty close," then sure you could justify making decisions based on that data, but I don't think we are close enough for that. Even just the "time worked on X" example is too complex to track. It is deceptively "simple." It seems like (especially to managers) employees should be able to work on a task for awhile, and afterward record how much time they spent working on it. However, it isn't that simple. In reality, "working on X" might actually mean working on X along with several other things such as email or web browsing or talking to a co-worker or answering the phone. With reliance on self-reporting, and without some sort of monitoring system, it is unreasonable to expect this metric to be accurate, yet this is how many (most?) time reporting systems work. Managers are making large-scale decisions based on this data. It looks like accurate data, it has fancy graphs and charts and reporting...but it isn't actually very accurate. An employee might report an hour spent on a project, when really only 40 minutes of that hour were spent on the project work, while the remaining 20 minutes were spent on various interruptions and extra tasks. This inaccuracy isn't much of an issue for informal uses, such as sticking to a personal schedule, but for driving decisions as part of a greater pool of data, it is misleading. At my organization, upper management is trying to use time reporting data to come up with a total cost for various initiatives. This is the kind of scenario I am talking about. Managers don't typically sit and watch everyone work, nor do they discuss time reporting entries individually, so all they end up seeing is the data. This separates them from how that data is generated and leads to inappropriate reliance on that data. From their perspective, it feels like reliable data that can be used to assign a cost to various projects, but the data probably doesn't adequately support that use case.
    - parasubvert 11y ago
      I feel you. Time tracking analytics is God awfully complicated and mostly impossible to get right unless you have a very disciplined culture. I've railed against such poor systems in my younger years. But sometimes there are ways to make it work. At Pivotal Labs for example, you are allocated to a pair daily and then to a project or product over a few weeks. There are no interruptions or meetings beyond townhalls and daily stand ups... Unless you and your pair occasionally want a ping pong break. Productivity and resourcing becomes a bit more reasonable in this culture as its related to stories/requirements in Pivotal Tracker being completed relative to the number of pairs rather than tying it to individual worth. It's like an incredibly kind and collaborative culture that also gets good data. Stepping back to a general management perspective, To me the most effective data is if you keep it relatively simple and do controlled experiments that are systemically tied to an external market result. A lot of companies in the logistics (and now software services) space get good at this.

jaawn 11y ago

The data-driven management you're describing is based on the significant assumption that you are able to gather accurate data, and interpret that data accurately. For example, if you are making a management decision based on how much time employees are spending on X, you are assuming that you actually know how much time is being spent on X and what the value of that time is. Simply measuring data is not sufficient. You have to measure the right data in the right way for your stated purpose. Companies operating like Amazon does (apparently), are wildly over-confident in their data collection and interpretation abilities, in my opinion.

parasubvert 11y ago
Most companies I have consulted or work for make decisions based on politics and conjecture, with data being invented at a bare minimum to back the argument up, but no one doing serious analysis of it. The culture of having data collection and an ability to challenge it (whether it is flawed or not) indicates a desire for a fact-driven culture vs. a power-driven culture. They might be over-confident, but one would think that reality in the market will eventually catch up to that.
- jaawn 11y ago
  This is entirely possible. However, when the data being measured is about complex human behavior (i.e. time allocation, productivity, work habits, etc...), I don't think we are even a little bit close to accurately representing reality. If we were "pretty close," then sure you could justify making decisions based on that data, but I don't think we are close enough for that. Even just the "time worked on X" example is too complex to track. It is deceptively "simple." It seems like (especially to managers) employees should be able to work on a task for awhile, and afterward record how much time they spent working on it. However, it isn't that simple. In reality, "working on X" might actually mean working on X along with several other things such as email or web browsing or talking to a co-worker or answering the phone. With reliance on self-reporting, and without some sort of monitoring system, it is unreasonable to expect this metric to be accurate, yet this is how many (most?) time reporting systems work. Managers are making large-scale decisions based on this data. It looks like accurate data, it has fancy graphs and charts and reporting...but it isn't actually very accurate. An employee might report an hour spent on a project, when really only 40 minutes of that hour were spent on the project work, while the remaining 20 minutes were spent on various interruptions and extra tasks. This inaccuracy isn't much of an issue for informal uses, such as sticking to a personal schedule, but for driving decisions as part of a greater pool of data, it is misleading. At my organization, upper management is trying to use time reporting data to come up with a total cost for various initiatives. This is the kind of scenario I am talking about. Managers don't typically sit and watch everyone work, nor do they discuss time reporting entries individually, so all they end up seeing is the data. This separates them from how that data is generated and leads to inappropriate reliance on that data. From their perspective, it feels like reliable data that can be used to assign a cost to various projects, but the data probably doesn't adequately support that use case.
  - parasubvert 11y ago
    I feel you. Time tracking analytics is God awfully complicated and mostly impossible to get right unless you have a very disciplined culture. I've railed against such poor systems in my younger years. But sometimes there are ways to make it work. At Pivotal Labs for example, you are allocated to a pair daily and then to a project or product over a few weeks. There are no interruptions or meetings beyond townhalls and daily stand ups... Unless you and your pair occasionally want a ping pong break. Productivity and resourcing becomes a bit more reasonable in this culture as its related to stories/requirements in Pivotal Tracker being completed relative to the number of pairs rather than tying it to individual worth. It's like an incredibly kind and collaborative culture that also gets good data. Stepping back to a general management perspective, To me the most effective data is if you keep it relatively simple and do controlled experiments that are systemically tied to an external market result. A lot of companies in the logistics (and now software services) space get good at this.