The Optimum Ratio in The Long Tail of Search

HitTail Best Practices Apr 13, 2007

Recently, HitTail forum user bvadel asked an insightful question. What’s a healthy site in terms of the ratio of “head” keywords to “longtail” keywords? He generously offers his site’s statistics of 13% in the top-10 keyword head, and 87% in the long keyword tail.

Yes, bvadel. That’s quite good. Here’s how I answered…

Let’s look at the life of a site.

Upon launching a brand new site, first there are zero search hits.

Then your first Google hit occurs, hopefully about 7 days in. We know that’s not realistic for everybody, but stick with us, and we’ll show you how.

On that first hit, your ratio is 100% head keywords, 0% tail keywords (using the Top-10 methodology that HitTail employs).

This ratio continues right up to and including your 10th unique search hit. 100% / 0%.

On the 11th unique hit, your ratio starts to change. You’re 90.1% head keywords and 9% long tail keywords. It’s still very skewed towards your “most popular” even though the hit count of your 11th word isn’t really any different. What’s in the head and what’s in the tail (such as it is) is arbitrary at this point.

Time passes.

In about 3 months, considering you’re publishing diligently, your tail starts to form. The traffic that resulted from your top-10 keywords starts to proportionally shrink compared to the totaling of the less popular tail keywords.

90/10 becomes 80/20 becomes 70/30 becomes 60/40, until finally they meet at 50/50.

If you’re doing your job well, this is only about 6 months into a brand new site. You are blogging every day, right?

Now, the rate at which the ratio flips slows down.

You creep to 30/70. And in about a year, you settle down to what is the average of all our HitTailers, which is ironically 20/80.

That is, 20% of your traffic is resulting from your top-10 keywords, and 80% of your traffic is from everything else.

This is one of the FEW places HitTail will ever look across everyone’s data–getting the head vs. tail averages, because it is of great value to the industry at large from a statistical standpoint.

Now here’s the rub.

Every once in awhile, a mega-popular site signs up for HitTail. They hardly need it. They’re massively popular, to the point that we either have to charge them for the heavy volume premium service, or trade service for service (which we occasionally do).

And those people have ratios like 5/95.

That’s right.

Their top 10 keywords are responsible for maybe less than 5% of their overall traffic.

This spectacular fact turns a lot of blockbuster economics on its head, in which 2% of the inventory selection accounts for 80% of the revenue–even at “long-tail” retailers like Amazon.com.

It takes awhile to digest, but it’s true.

The more popular your site becomes, the less you rely on any particular keywords.

Popular sites are diversified, and skew heavily towards the tail.

3 comments
  1. Peter Koning

    But the tail does thicken somewhat for those 95/5’ers right? I’m seeing 2’s and 1’s all the way down at the moment. Over time I would expect the tail would grow very long, but also get a little thicker too after the 5% steep curve flattens. Does that make sense?

  2. Mike Levin

    Alex Barnett had one of the best articles I read on the thickening of the long tail. And you never see anyone explicitly saying this, but yes, what we see with Chris Anderson’s economic model is actually NOT a 1/x curve. It’s a much more complicated logistics curve that has the “S” shape. It’s built into most HP calculators for easily running regression analysis, and was Sam Walton’s secret to success. But it’s otherwise unknown. And it’s why the seeming long tail can thicken. I don’t see it discussed much, but it probably should be. Hmmmm. Maybe another blog post.

  3. Justin

    That actually fits in with what chris anderson is saying in his book The Long Tail: Why the Future of Business Is Selling Less of More, he makes an excellent argument that the long tail accounts for a greater percentage overall overtime as the “infinite shelf space” grows and grows and grows

Leave a Reply

Your email address will not be published. Required fields are marked *