sensible lines "How do I measure the value of my "upper funnel" keywords?" People ask me this all the time.

Upper funnel keywords are typically those that are used by Customers who are early in their consideration life cycle, they have not made up their mind.

They are keywords that introduce people to you. They don't convert, that happens later in the life cycle (say as they move from "curious" to "interested" to "consideration" to…).

So how to measure value of those keywords? [You can also apply this to targeted email campaigns or other marketing adventures.]

Before I answer that question let me take you down memory lane and give you some critical context, then we'll answer the question.

A while back I had written a post about the long tail of search.

You can read that post (it is a good one :) for more details, but the essence of the phenomenon is that your website most likely has a "thick head" (say ten or fifteen keywords that drive massive number of Visits) and usually a "long tail" (hundreds of thousands of key words and key phrases that drive five, ten, fifteen – few – Visits).

As an example for the month of Jan '09 for my blog:

the "head" is 11 keywords (100 or more Visits) and

the "tail" is, and I am always astonished by this, 22,181 key words!

Can you believe that?

For such a deeply focused blog on one topic there were a total of 22k key phrases that delivered traffic from search engines.

That's both a reflection of all the love/work I put into SEO and a real living breathing example of the long tail.

the long tail of search

My long tail post also outlines how the head and the tail have very unique characteristics when it comes to the kinds of:

key words that are present in the head (usually brand keywords) and the tail (usually category / generic / ecosystem) and

people / visitors that use who come with head (people who know you already / have made up their mind) and tail (people new to your franchisee) key words and key phrases

Read the post on the search head and long tail to see how to uniquely use Organic Search (SEO) and Paid Search (SEM / PPC) strategically to plan your world domination.

My recommendation there was that most businesses should have a very focused and efficient long tail search strategy to ensure they keep attracting newbies (prospects) to their company / product / franchise. These people have not made up their mind. Find them first, make them fall in love with you, retain them.

[Bringing the story back to where we started....]

"Upper Funnel" has a very brotherly correlation to the "long tail" keywords. Both introduce new people to you, people who have not yet decided what they want to buy / sign up for etc, people who will take a little bit to buy.

So what's the problem? Why don't more people do this?

Mental Model & Measurement.

It turns out that most Online Marketers tend to think of life, online atleast, to be all about "one night stands". Come. Wallet out. Convert. I have addressed this issue a lot, most recently in my Measure Latent Conversions & Visitor Behavior post.

The "single session" mindset is very corrosive. Mostly because typical customer behavior looks like this:

conversion funnel

People are introduced to your firm / products / services. Great. They are early in the consideration life cycle. They might go away. Come back again. Perhaps now with a branded search query. Go away. Then some will come back and convert later.

You on the other hand are measuring single session conversion.

That translates into you devaluing all the "upper funnel" / "long tail" keywords as losers.

For example I placed an order for Mobile Madden 09 at by clicking on search ad at for Madden 09. But the reality it I only considered EA because a couple days back I had clicked on an ad for Mobile Football Games on

That keyword got nada credit.

So how do you value the "upper funnel" / "long tail" keywords?

My recommendation is simple:

1) Understand each stage of the customer purchase life cycle.

2) Map your keyword portfolio to each of those life cycle stages (curious, awareness, interested, consideration, purchase etc).

3) Measure success of keywords in each step differently. No: "everyone's got to convert!"

Early stage / upper funnel / long tail keywords are significantly cheaper, let's hold them to a lower standard.

"Purchase stage" life cycle products can be expensive. Let's hold them to a much much higher standard.

Here's how your potential measurement strategy would look like:

metrics kpis for customer conversion lifecycle

Use Bounce Rate to measure early customer consideration lifecycle keywords. You are paying 30 or 60 (or whatever) cents per click, getting that traffic to come and stay on your site (give you one pathetic click) might be a worthy goal. They came, they saw. Rather than they came, they puked, they left (my definition of bounce rate ).

For the next set of keywords (active consideration) there is a lot more competition. They are more expensive. Demand that those keywords drive traffic that spends a lot of time on your site. Explore what you have to offer, get to second base .

Then for your high cost brand keywords expect that they will bring visitors that will show a high degree of Visitor Loyalty, i.e they will come for multiple visits, getting very so close to converting (third base! ).

Finally for your conversion keywords (mesothelioma! :)) only value success by measuring conversions, leads, multi channel offline impact .

You may pick different metrics, your customer life cycle might be longer / shorter, your keyword portfolio might be fatter or leaner than mine. Consider this my attempt at sharing a adaptable framework.

money success

The Outcome:

You have put together a flexible framework that helps value your paid search keyword portfolio in a significantly better way.

You still demand that every keyword you bid on delivers its pound of flesh (value).

You also map things back to the customer consideration life cycle, achieving better targeting and landing page experience on your site.


Think of this contrast: You just "blew" $3 million on a 30 second ad at the Super Bowl. The ad was irrelevant to most people who saw it (or heard it as they ran to the bathroom). Now contrast that with executing the above strategy with Microsoft, Yahoo!, Google, for 30 or 40 or low cents per click (or even dollars) getting in front of relevant people, those that actually typed in keywords related to your business, people who are looking for you! A hugely higher ROI awaits you.



You should be able to do much of the above with your Web Analytics tools. Especially if you unleash the power of data segmentation .

If you work with Search Agencies then demand that their tools do this for you. They can. You may have to push. Do it.


Most web analytics tools today stink at truly doing multi touch attribution analysis. That's the art (sadly not yet science) of measuring truly pan session customer behavior, the campaigns (search, display, email…) that caused each visit, and then attributing some element value to each touch point.

Doubleclick has such a report, "exposure to conversion" report. It is ok, not as sexy cool as it could be. Atlas has something similar as well. ClickEquations, for Paid Search, has something close, you should check it out. [Disclosure : I am on their Advisory Board.]

A couple web analytics vendors, in exchange for $1 to $2 mil, will give you additional addons on top of the web analytics tool you are paying for. These addons don't have multi touch attribution analysis reports built in. But if you want to you could built it yourself (though you'll have to come up with your own attribution models).

It is my hope that we as an industry will slowly move towards allowing you much easier access to really advanced features like multi touch attribution analysis.

It is a very tough problem to solve. For example it is computationally very intensive (and expensive), and you have no idea how hideously complex attribution models can be.

But I hope we'll get there.

It is important to realize you don't have to wait for that. In this post I am not requiring multi touch attribution analysis. You can do what I am suggesting tomorrow.

You can start to reshape your approach and mental model today.

Good luck.

Ok now your turn.

Do you have a long tail? Do you see the behavior I have outlined above on your site? How do you value your upper funnel keywords? Do you think my model will work?

I would love to have your thoughts and critique.


Couple other related posts you might find interesting:

Social Bookmarks:

  • services sprite
  • services sprite
  • services sprite
  • services sprite
  • services sprite
  • services sprite
  • services sprite
  • services sprite
  • services sprite
  • services sprite
  • services sprite