Saturday, August 11, 2007

The Myth of the Long Tail

I've had a few interesting discussions lately regarding the long tail. As Chris Anderson has made very clear, there is a ton of potential in the long tail concept, what concerns me is the common perception that the long tail is a strategy unto itself, rather than a tool for gathering data on what parts of it work for you in particular. As it relates to search marketing, the romantic notion that this single concept will, and does, work for anyone is quite flawed.

In this post, I’ll be addressing how to maximize returns from the land of long tail, and how to determine if you should even be paying there at all.

Why It Works For Amazon
The reason the long tail is so very effective for Amazon is due to the relative cost of maintaining that inventory. They leverage an entire infrastructure to capitalize on the minimal costs of serving photos and and text about products. This is an incredibly effective demonstration of the economics of scale.

Searcher Intent VS Advertiser Desire
One of the greatest values in the long tail is actionable insight on buyer intent. That is to say, what buyers are willing to accept your product/service as, rather than what you tell them it is.

People vote with money and action. If you are selling a nationwide service, and you have long tail keywords associated with every city from
New York to San Diego, you may find that a large percentage of those geo-targeted users in fact want their service provide to be in St. Louis or Denver or wherever they are, not where you tell them your service applies.

Problems continue to arise when::

  • Marketing metrics are tied to lead counts rather than sales totals
  • The cost of maintaining the additional keyword inventory rises as a proportion of total costs
  • Competitors blindly bid the easier long tail keywords into the stratosphere because of the, seemingly rational, thinking that long tail keywords are a good investment

This is a crucial insight, and allows search marketers to view the subconscious perceptions of their target audience through bounce rates, on page times and conversion rates, which I'll address in a future PPC 101 series. For now, let's consider a universal success metric, the conversion, in determining where and when you should be wagging the long tail.

How to Use the Long Tail

Ah, conversions.
The thing about conversions is they are very often what I call "measured at the wrong goal post", particularly in my b2b experience. A conversion in regards to a business website will often be a lead rather than a sale, due to the long iterative nature of the b2b sales cycle.

I have found this to cause problems for two reasons:

  • Long tail searchers seem more open to registering (or converting) on websites because they are aware of what they want, and measuring leads as conversions can lead to suspect lead counts.
  • Those same long tail searchers are in fact more selective when deciding on actually doing business with a service provider because they are further along in the decision making process, meaning unless you have what they're looking for, they're moving on.

How do you then ensure that the huge mass of long tail keywords you're dumping money into is actually working for you? Easy!

  1. Drive traffic
  2. Measure conversions to actual sale
  3. Remove non-performing money pits
  4. Rinse, wash, repeat
That's it. Just make sure you manage the long tail keywords as well as your core set, and do not be afraid to stop using large quantities of them if the performance is not there. The point is not to invest in those keywords that, for whatever reason - your branding, location, whatever - simply will not work.

Segment, Segment, Segment
The simplest way to control long tail keyword sets is to place them in their own Campaign. You can use dynamic placement to minimize build out or (my favorite because I'm a bit of a control freak) build as any ad groups as you can stand, preferably only for the top 20% of your long tail keywords. This emphasizes the keywords that stand a chance of "moving the dial", and prevents you from relying too much on bulk keywords, which offer little control.