Most companies measure keyword performance – and especially PPC keyword performance – based on one factor: did that word or phrase bring converting visitors to the site on the visit in which they converted.
So the natural thing to do is trim non-performing words and phrases in order to increase the efficiency of your PPC spend. And that’s exactly what one client did, except rather than increasing his efficiency, he dropped his sales by 30%.
Because, depending on what you sell, lots of people buy on their second, third, or umpteenth visit to your site, rather than the first visit. Those visitors are building confidence in you as they move through their buying process. But most systems don’t (or can’t) track user behavior over multiple visits. So when those early and middle buying-stage keywords shown up as non-converters, they get cut.
The shame is that not everyone is able to track the following sales drop off, which may not occur for days, weeks, or months, back to the act of cutting those keywords.
Would you trade Dennis Rodman for non-performance? Of course not, right? Rodman’s defensive stats alone tell the tale. At his prime, Dennis was pulling down a truly astonishing 18.7 rebounds per game. For reference, the previous year’s league leader in rebounds (David Robinson) averaged 13 per game.
But if the only stats you looked at involved scoring, you’d get a different picture. Comparing Rodman’s 8-9 points per game against other star players’ 20 or more points per game, you’d likely have been misled into trading Rodman, only to find yourself wondering why you started losing games and everyone else’s scoring stats went up against your team.
Think of your assisting keywords terms as the Dennis Rodman’s of your PPC campaign, except you’ll get all the assists and none of the off-court shenanigan’s.
A recent eConsultancy post discusses how Google’s default window for tracking cookies can distort traffic data. Left in its default cookie window setting, Google Analytics (GA) will classify visitors as “search”-driven traffic for six months following a single search based click through to your site – regardless of how they got to your site previous to that search or how they might arrive at your site following that search. Here’s an example of how this might skew your results:
Let’s say you’re driving traffic to your site via radio ads and that a listener, after hearing your ad, types your url directly into his browser. Later, he comes back but this time, he types your business name into Google and clicks through on a displayed search result. Following that, he visits your site three more times via bookmark or directly typing your URL into his site. That’s a total of 5 visits.
Question: How many of those visits would GA classify as search-driven?
Answer: 4 out of 5.
GA would count the first search-based visit and then all of the remaining 3 visits, despite the fact that the following three visits didn’t use search and may have taken place several months after the initial search. Multiply that by all your visitors/visits, and you can see how your understanding of what drives traffic to your website might be distorted in favor of search. And under the impression that your traffic was mostly generated by search and not, say, your radio ads, you might be tempted to cut them from your ad spend. Obviously, the same thing could apply with e-mail campaigns, magazine ads, etc.
Any experienced Web Analyst or Website Optimizer could extend this list of “gotchas” and “classic mistakes” almost indefinitely. It’s just not that uncommon for an uncareful analysis of data to lead online marketers either to analysis paralysis or sub-optimal optimization strategies. Is it any wonder that 70% of businesses collecting wed data fail to act on their analytics data?
Obviously this issue has been central to Bryan and Jeffrey Eisenberg’s Web careers since the beginning. It’s why they helped found the Web Analytics Association; why they published The Marketer’s Common Sense Guide to eMetrics, Call to Action and Always Be Testing; why they created Persuasion Architecture; and ultimately why they’ve built the OnTarget program.
The central theme amongst all of these issues is bringing clarity and actionable insight to Web improvement and online marketing efforts. They are all answers to the business owner who feels confused or disoriented by the data he’s given and want’s a clear direction toward more sales/conversions and improved website performance.
So, if you find yourself struggling to make sense of your online marketing data, or frustrated by non- or counter-productive optimization efforts, ask yourself: are you giving credit where it’s deserved? Or do you need help achieving greater clarity and actionable insight from your optimization efforts?