You might have heard that if you can\u2019t measure you can\u2019t manage. You might have believed it, too. And if you think you need metrics to manage you might be feeling guilty about not having enough of them.\n\nGuilt no more. Good metrics are hard to craft, harder to manage, expensive to maintain, and perishable besides.\n\nBad metrics, in contrast, are easier all the way around, but that doesn\u2019t matter. Bad metrics are worse than no metrics. If a bad metric isn\u2019t lying to you outright, it\u2019s directing your attention to what doesn\u2019t deserve your attention.\n\nNot to mention that knowing what\u2019s going on isn\u2019t the primary purpose of establishing metrics in the first place. That\u2019s a fringe benefit.\n\nWhat metrics do for you is a two-edged sword. Edge #1 tells employees what \u201cbetter\u201d means and lets them know whether they\u2019re achieving it.\n\nIt\u2019s two-edged because without metrics, employees will do what they think makes sense to make a given situation better. But with metrics, they will do what they think will move the needle, regardless of whether moving the needle is a good idea.\n\nMisled by metrics\n\nMetrics are tools, and like any tools they\u2019re as easily misused as used. To help avoid misuse I\u2019ve come up with a simple set of guiding principles:\n\nLewis\u2019s 1st Law of Metrics: The primary purpose of any metric is to alter employee and team behavior. Letting you know how you\u2019re doing? That\u2019s nice too.\n\nThis is followed by Lewis\u2019s 2nd Law of Metrics: You get what you measure \u2013 that\u2019s the risk you take.\n\nWhich in turn is followed by Lewis\u2019s 3rd Law of Metrics: Once you get employees in the habit of moving the metrics, they won\u2019t do anything that doesn\u2019t move a metric. Anything you don\u2019t measure, you don\u2019t get.\n\nThe 7 Cs of IT metrics\n\nClearly, if you\u2019re going to establish a system of metrics to help you manage performance, you need good metrics. But what\u2019s the difference between good and bad metrics, you ask? Good metrics adhere to what I call the 7 Cs. Okay, its name might be annoying, but if you ignore any of the 7 Cs you\u2019ll end up with metrics that are, under the best of circumstances unhelpful, and under the worst actively destructive.\n\nGood metrics must be:\n\nPutting IT metrics to the test\n\nAren\u2019t convinced the 7 Cs framework works as advertised? Let\u2019s test it against a popular industry metric: the \u201ctotal cost of ownership\u201d (TCO). It\u2019s well-known and widely cited. Surely it must exemplify what quality metrics look like. Here goes:\n\nConnectedness: Economizing is usually a worthwhile goal. TCO is connected.\n\nCalibration: Every TCO component can be specified; in principle, the cost of each component can be tied down to well-documented spending; and the rest is simple arithmetic. TCO is calibrated.\n\nConsistency: Uh oh. When TCO increases, does this mean the situation is worse than it was before and vice versa? Sometimes yes; sometimes no. Take, for example, the cost of connecting a PC to your network. Don\u2019t do this and TCO goes down. But your PCs are pretty much worthless.\n\nThe TCO of almost anything increases the more it\u2019s used. Conversely, you can bring TCO to zero through the simple expedient of turning everything off.\n\nTCO isn\u2019t consistent.\n\nCompleteness: TCO is as incomplete as it is inconsistent, because in business, cost doesn\u2019t matter. Value is what matters, the net of benefit minus cost (or divided by cost if you prefer ratios). TCO ignores benefit and value, and anything you don\u2019t measure you don\u2019t get. TCO is incomplete.\n\nOn a Continuum: In principle, researchers could survey the product category\u2019s marketplace, measure every competing product\u2019s TCO, and report where each product\u2019s TCO falls in the range of measured values. TCO can be placed on a continuum, but in practice it\u2019s unlikely.\n\nCommunication and Currency: Unlike the first five Cs, these are about your metrics practices, not the metrics themselves. Whether TCO is communicated and current is a matter of choice, not of TCO itself as a metric.\n\nTCO Metrics Success Metric: Taking practicalities into account, TCO fails three of the 7 Cs \u2014 consistency, completeness, and the need for a continuum. Which leads to the question, are yours any better? And if not, what steps do you need to take to fix them?\n\nWrapping it all up\n\nIf you measure well, you can probably manage better than if you don\u2019t \u2014 so long as, that is, the cost and difficulty of measuring well don\u2019t exceed the benefits.\n\nMaybe we need to compute the Total Cost of Measurement. Or, better yet, the Total Cost of Management.\n\nAny volunteers?