The exception doesn’t prove the rule. An exception proves that the rule is an imperfect analogy that you can’t count on. It’s not quite as catchy, but it is pretty much the entire foundation of modern science. Finding exceptions to rules is the only way we have to confidently separate the good ideas from the bad ones.
Here are ten factors that are easy enough to measure, but terribly dangerous to rely upon to tell you any more than exactly what they tell you.
1) Number of twitter followers
The number of followers you have on Twitter is a good indication of the amount of time you’ve spent on Twitter, but it’s not a good measure of your ability to influence people. If it was, Gustavo Dudamel (24,472 followers) would sell a lot more records than Anna Netrebko (1,932 followers).
2) Number of blog subscribers
Blogs acquire subscribers over time in much the same way as stones of the not-rolling variety gather moss. Those people aren’t necessarily reading your blog.
According to Google Reader, Alex Ross has 10x as many subscribers as I do. If subscriber count was proportional to readership, I’d see a huge spike in traffic every time he links to me. In fact, the links generate about as much traffic for me as I do for him, when I return the favour.
3) Number of retweets
My April 1 post was entitled “Social Media Marketing,” and lots of wannabe social media experts shared it on their Twitter feeds. For a while, the Twitter shares outnumbered the total page views, which means that not only were all these people’s followers not reading it, but they were sharing it without having read it themselves. Some of them will probably do the same with this post. Morons.
4) Wikio ranking
Your Wikio ranking is based on little more concrete than the number of times your stuff was shared on Twitter which, we’ve just established, isn’t a good measure of anything in particular.
5) Hits
1997 called, and they’d like their buzzword back. “Hits” means whatever you want it to. Be more specific. Items requested from the web server? Page views? Unique Users? Mention “hits” to me in a meeting, and I’ll be double-checking all your other “facts” on the assumption that you either don’t know what you’re talking about, or, worse, that you’re hoping I don’t.
6) Views on YouTube
Susan Boyle scored 100,000,000 YouTube views and sold 700,000 albums. This gives us Boyle’s Constant – 14,285 – the number of times your stuff has to be seen on YouTube before somebody will buy it.
It’s a neat theory, but a lot of people don’t even do that well. If they did, the smart money would be on a platinum first week for Rebecca Black. You don’t know who that is? Count yourself lucky. Whatever you do, don’t click here.
7) Conversion rate
Define it clearly and then compare it to something concrete, not an arbitrary baseline. A conversion ratio depends upon so many factors that about the only useful thing you can do with it is look at how it changes over time.
8) Clicks
I don’t care how many people you sent to my shop. I care how many people bought something. All the others were just slowly driving up my bandwidth bill.
9) Impressions
You’re going to have to clarify what you mean here, because counting the number of times a picture was downloaded from a webserver will not tell you how many people saw it.
10) Anything with “Top” in it
Why do the charts on iTunes say “Top Songs” and “Top Albums”? Simple. It’s because “Albums ranked according to a secret formula based on today’s sales and an exponentially decaying portion of historical sales” is not very catchy. It is, though, more informative. Pretty much anywhere that “Top” is used to indicate the best of something, it’s a sneaky bit of reductionism, waiting to pollute your decision-making process.
“These stats might not be perfect,” says the consultant, “but they’re the best indicators we have, so we should keep using them.”
That’s almost always total crap.
If you’re trying to find your way from Portugal to Jamaica, a map of the Pacific isn’t going to help you. It’s how Columbus found the West Indies, but he spent the rest of his life believing he’d made it to Asia, so this analogy isn’t on your side.
We’re terrified of admitting that we have no idea what we’re doing, but it’s quite liberating. Be suspicious of anybody that tells you otherwise. There’s a good chance they’re wrong, and they don’t even know it.
Proper Discord 
“Boyle’s Constant!” I love it.
Thanks. Now go share this on Twitter to inflate my Wikio ranking.
Nice one Mr Discord. The ‘hits’ metric is one I’ve been trying to get across as phoney for years, ever since a high profile orchestra claimed it had “a million hits a month” and I was asked by my colleagues why our site didn’t.
Ask your IT people how many http requests are served up each month. There are 67 items on your homepage. 14,926 views on that page will get you to a million hits. Hmm. That’s suspiciously close to Boyle’s constant.
This one feels a little overly contrary to me. Dudamel selling fewer albums than Netrebko is a weak argument. Clearly the more twitter followers you have, the more effective platform it is to get your word out. That doesn’t translate to album sales, but that doesn’t make it a bad metric, provided you understand what’s being measured.
I think you’re making the case that people shouldn’t give these numbers more weight than they deserve (and you explicitly say as much), but the entirety of the article comes off as implying that the numbers are useless at best. Thinking is more important than measuring, and knowing what you don’t know is as important as knowing what you think you know, but in the right hands, measurements, even imperfect ones, are useful tools.
I didn’t use the title “10 worthless metrics” because they’re not worthless. Each of these measurements can tell you something very specific. The trouble is, that’s rarely the piece of information we want to know. It’s tempting to treat these as the answers to different questions, and that’s dangerous.
More Twitter followers absolutely does not necessarily mean a more effective platform, because not all followers are equal. Some are spammers, some are bots, some are children, some are dormant accounts, some are marketing people, looking for good ideas, some are adults with credit cards. It’s not safe to assume that one person’s 4,000 followers are worth twice as much (or even more), for any specific thing, than another person’s 2,000. Only a small number of Twitter users are actively engaged in the service at all.
If you want to sell something, you need real users with real credit cards. If you’re a venue, you care most about followers in your local area. Personally, I’m trying to reach people who are interested in digital marketing, so Twitter is a really useful outlet, but I still know that many of my >1,100 followers only followed me because they hoped I’d follow them back, and they can’t possibly be reading my stuff.
Well done. Nice to see another person bringing up quality-over-quantity in regards to social media reach. With Twitter, there are so many ways to get large numbers of followers through different “follow-back” systems like you mention, but most of those followers aren’t looking at anything you tweet. As someone who does work in digital/social media, my suggestion for the stats-junkies (because they are important), is to use everything you can get ahold of, and then look at it all as a whole in order to compare the results with your goals. No one metric is the end-all-be-all. These things don’t work in a vacuum. If you’re trying to determine sales-growth, looking only at Twitter follower numbers does nothing for you unless you go deeper.
The only issue I might have here is with your (seeming) disdain for the “click-through” statistic. Just as it is with a “brick-and-mortar” shop, you have to get customers in the door before they are going to buy something. The more people you get to the store, the better your chances of getting sales. If the average purchase-per-visit was 0.01%, you’re still going to be better off if you have twice as many visitors purchasing at that same percentage. But, click-baiting (like your “whatever you do, don’t click here” example) should not be a goal or the only statistic considered. I actually think if you compare click stats and your “incoming traffic” stats from your site, you can get some pretty good, and reliable data. But again, you really can’t look at one statistic as an end-all-be-all.
Thanks for the post and for being another voice preaching quality-over-quanitity.
Fair points, all. Just wanted to emphasize that Dangerous != Useless (which a less than careful reader may not pick up on).
I’m sure someone’s already thought of measuring influence by weighing each of your followers by how many followers they have. A ‘follower-quality’ metric. That would weed out the spammers and follow-baiters (that sounds dirty).
Funnily enough, Wikio uses something very similar to your “follower-quality” metric.
A lot of commercial twitter accounts use auto-follow software to follow thousands of people, in the hope that they’ll return the favour. A lot of my followers have, themselves, many thousands of followers, but they also follow thousands of people and can’t possibly be reading what I’m saying.
The data doesn’t become more useful when it tries to stand on its own shoulders.
Some social media tracking services count mark individual users as being influential, because their stuff gets shared a lot. It requires really complicated tracking, but it is a slightly more useful indicator or reach than raw follower numbers. Still, it doesn’t tell you anything about the interests of those followers, or whether or not they’d respond to a call to action more demanding than “retweet this”.
Isn’t there some sort of Donald Trump corollary to this somewhere? You’d figure there has to be…
Having a lot of money, an apparent knack for business and a dead hamster on your head does not necessarily mean that you’re not a hypocritical racist?
I suddenly feel light as a feather!
Thank you, sir!
Hi, just wanted to drop by and say thanks again for this post. I’ve been working through Gary Vaynerchuck’s “The Thank You Economy” on my blog and writing about how to apply it to classical music, and this post inspired me to get through the part about “Intent Mattering.” The reason your post was enough to get a blog out of me was because you drive home the idea that stats alone can’t be your end goal. So, for what it’s worth, here’s my post “Intent Matters”: http://ow.ly/4Ztzv. Thanks again and keep up the good work.