Bootstrappers have to figure out how to market via the web. It’s the lowest cost, lowest friction, highest leverage medium available to get the word out. And, it is the easiest place to collect metrics so your decision making can be informed rather than just seat of the pants. But there are many dangers with metrics. First and foremost is optimizing the wrong one.
John Pozadzides, CEO of the Web analytics company Woopra, has a great article in ReadWriteWeb about the leading sources of referral traffic. There really is some awesome stuff in there, and it is tempting to go through the article’s major conclusions and think you have a plan:
– Facebook trumps all the other Social Media by a large amount with 68% of referrals. Twitter only gets 25% and LinkedIn 4%. Translation: Bet hugely on Facebook, maybe experiment a little with Twitter and LinkedIn, but ignore the rest.
– Of the Social bookmarkers, Stumbleupon gets 51%, Digg 30%, YCombinator 12%, and the rest is noise. So, we need to bet on Stumbleupon with a little follow up on Digg.
– Google is delivering 92% of search referrals. Guess we won’t worry about anyone else there.
– Media site referrals are 84% YouTube. Hey, Larry, we better get going on some videos!
What’s wrong with that marketing plan? It’s backed up by well-researched numbers. The statistics are compelling. How can we lose?
The answer, as in so many similar cases, is we may have optimized the wrong variable. It’s reasonable to assume that you want to get new traffic to your site in order to sell more product. After all, if prospects don’t know your better mousetrap exists, its hard to beat a path to your door. And these various venues are only too happy to help. In fact, Google would love to send clicks your way for just a small fee. However, even if all visits were created equal, we have no idea what the relative sizes of each pool are. So what if YouTube is 84% of media referrals if media referrals turn out to only be 1% of overall referrals? Worse, all visits are not created equal.
Look at the YCombinator number as just one example. I don’t mean to pick on YCombinator, but go take a look at it. It’s a very specialized audience. Clearly they can swing that audience to go look at something in droves. But is your product really going to benefit from that audience’s interest? What will be the symptoms of making a mistake there? Well, you might successfully capture a bunch of referrals from YCombinator only to find that they bounce as soon as they hit your site. Yes, your content cleverly got them to click, but your product’s substance couldn’t cash the check you’d just written with your clever content.
I’ll give another hint. The article suggests Stumbleupon may be an awesome place to focus some attention. I get a fair amount of referrals from Stumbleupon, and I’m not even trying, so I’m not surprised they show up well in a study like this. It was the #7 source of referral traffic, in fact. There is just one problem with focusing on it. When I checked this morning, Google Analytics tells me that it had a 96.9% bounce rate. The traffic arrives, doesn’t like what it sees, and leaves immediately. Average time on site is 49 seconds, versus a site-wide average of nearly 4 minutes. Maybe, with some effort, I can redirect Stumbleupon to do a better job. But, if I’m just looking at the raw referral number, I have optimized the wrong metric. I need to look at something else that gets closer to what I am trying to achieve.
So what metric should we be optimizing? I’m a huge fan of closed loop marketing. I want to be able to track a lead, let’s call that lead a visit to my site, all the way through to a purchase. I want to understand that path as well as I can, from source all the way to revenue. I want to assign costs, volumes, and conversion rates to these sources and tie the effects of my changes in marketing back to their impact on these variables. It’s a lofty goal, but it’s the only way to know what’s working. If you optimize the wrong variable, you’re in for a potentially very expensive mistake either in opportunity cost for having worked on the wrong thing, or in real cash if you’re paying for clicks that didn’t translate into sales.
How do we get this kind of end to end analytics? The good news, is that it isn’t that hard even for a tiny bootstrap company to get a start on it because we have Google Analytics. It works like this:
Google Analytics will do a good job tracking where people who come to your site came from. It knows what pages they land on and where they go from there. Your job is to make the act of landing on a page correspond to the metric you want to measure. Let’s say you want to track people buying your software. Perhaps there is a page you could create that people only go to if they bought the software. It isn’t reachable in any other way. Maybe it is a Welcome page, or a page that activates their product, or better a page that confirms activation. Make that a Goal page and you’re ready to work backwards to analyze your funnel. You probably want to differentiate content that attracts people to your site and facilitates SEO from the actual page where you pitch your customers to buy the product. Perhaps there is a trial close along the way where customers sign up for a free trial. So on and so forth.
Once you start analyzing these patterns, you’ll come to understand what really works for your world. Tie up Google Analytics to Google Adwords and you get an even more powerful analytics combination. You start to understand the impact and interplay of keywords, ad copy, traffic sources, and even factors like time of day, geographic region, or languages. I’ll have more to say about how to use these and other tools over time as I write more about bootstrapping.
Some will argue that its impossible to get these metrics, so we shouldn’t worry about it. I couldn’t disagree more. The Internet makes it easier than ever before. Marketing is Tragically Knowable. I will always remember that phrase from my mentor, and co-founder of Netflix, Marc Randolph. Working through these kinds of metrics and conducting the A/B experiments to know what works better is all part of avoiding the “Tragic” part of “Tragically Knowable.” Tragic is when you think you know, whether by incomplete information, optimizing the wrong metric, or just going by gut, you make a big bet on it, and it turns out to be wrong. To avoid the Tragedy, make sure you understand how each factor you’re trying to optimize feeds your ultimate goal of selling product. Don’t optimize the wrong metric.
Software developers say that premature optimizing is the root of all evil in software. Optimizing the wrong metric may be the root of all evil in marketing.