Easy Tricks for Great SEM

Five Mill Tree Method, explained

Let me now dive right into the “Five Mill Tree Method”, which has really been the impetus for 95% of my Search Marketing methodology.  In the previous post I discussed how we we were stuck in 1998, and as much as I loved that song by Next, there have been so many opportunities to optimize accounts that have been shadowed by “long tail” tunnel vision.

As I said, the advertising landscape changed from a list of keywords to one advertising space with many different ways to dissect it.  We have been given the opportunity to become more granular (my favorite buzz word).

I like pictures, so here’s my graphical explanation:

[NOTE: At this point I'm starting to regret choosing "socks" as my example.  Oh weeelllllll... looks like we're committed to thoughts of warmer feet]

In the new keyword landscape you can cut the advertising landscape into as many parts as you want.  You can target by Geography, Match Type, Daypart, and Network.  Using just those few options, you can create quite a few different match types, but I’ll get to that later.

Picture Time!:

Previously, I noted that the way to build out and optimize campaigns was using the following methodology:

Step 1.  Build Adgroup(s) for “Head” word(s)

- Test ads

- Pick best performing ad

- Bid keyword

- Re-bid if/when performance changes

Step 2: Build Out “The Long Tail”

- Keyword tools: use whatever “proprietary” and non-proprietary tools to do so

- Spreadsheets:  form a dirty relationship with the concatenate function

- Web logs:  analyze the queries that actually bring users to the site and add them to the keyword list

With this very fundamental change in the the Search Marketing algorithm, I feel that it is absolutely necessary to throw out 1998 and look at and utilize the tools that are in front of us.   My methodology changed to one better explained by a graphic.  In the below graphic think of “KW1″ as the circle that defined Socks (in broad match).  As we develop more information, we break up the traffic source by performance, and performance alone.

Now, let’s imagine we’ve gathered enough data for KW1, to make a decision.  In other words, let’s say we find out that after running the broad match for “Socks”, that [socks] performance better and the remainder performance worse.  We then branch out and bid accordingly.  Let’s say that we then look just at [socks] and find that traffic from California performs worse and, conversely, the remainder of the [socks] traffic must perform better.  Etc…. etc…  Of course, we are also adjusting bids accordingly.

Here’s a graphical explanation, as per usual:

Note:  *For the most part*, the most efficient and rapid way to optimize your account in this fashion is by making sure that each branch/split in traffic is a roughly 50%/50% split.  There are some caveats to that statement, but for the sake of not entering some complex match, let’s just try to keep splitting 50%/50%.

Everything I have explained above is simple.  In fact, it’s painfully obvious in many respects.  But, for some reason, many advertisers still think in terms of “head keywords” and “long tail” keywords.  One major added benefit of the Five Mill Tree Method, which should not be overlooked, is it’s ability to create awesome Barriers of Entry.

Barriers of Entry:

The Five Mill Tree Method is great for optimizing and becoming more granular.  But, it’s ability to form Barriers of Entry is just plain fun.

So let me give you a very simple example of the usual extreme case where the Five Mill Tree Method is “useless”:

“Why should I optimize?! I’m already #1 on my keywords and I’m #1 EVERYWHERE!”

I see this scenario often.  Occassionally, a “Mom & Pop” company will enter the Search Landscape (or for the sake of being SF PC, how about we say “Pop & Pop” or “Mom & Mom”?  Eh, I will maintain the more famous idiomatic expression “Mom & Pop” in order to make this story less confusing).  They typically come in, bid some arbitrary sum which, in fact, tends to be much higher than is typically worthwhile.  We’ve all seen it.  We typically then assume: “My product/offering is better than the next guy and they won’t last long”.

It’s really fun to make that assumption because it’s the easiest thing to do.  Unfortunately, though, it’s not always right.  Some one will come in, with a highly competitive product or offer at lower prices and rock your world… at least in the short term.

Dear Five Mill Tree Method, how are you going to help me here?

Well, let’s say we determined that the the northern half of the US converts at a higher rate when click on ads for Socks then does the Southern half.  So, what do you do?  You bid accordingly… obviously.  So, no we’re advertising $5 per CPC is in those colder Northern states and $1 in the Southern states.  Per our previous statements about always being “#1 Everywhere”, we are not shocked in this hypothetical scenario, that we will maintain the #1 position everywhere.

Let’s say “Mom & Pop” comes in and puts out a $3.50 Max bid per click.   Now, we have the #1 position in the North and “Mom & Pop” has #1 in the worse performing South.  The overall mix has left our competitor with worse traffic and you with the better traffic.  Overall, we are roughly ad position 1.5 and the competitor is roughly 1.5.  [NOTE: i'm making gross assumptions that our ad CTRs and Conversion Rates are relatively or proportionally the same and there aren't really many other true competitors on the landscape.  We can nit-pick this example all day, but it doesn't deny the barrier of entry that we have created].  Here’s yet another graphic to help explain.

As a result of this Geo-Targeting use of the Five Mill Tree Method, here are some important assumptions we can make:

1.  Competitor gets good ad position.

2.  Competitor gets relatively worse traffic.

3.  Competitor does not analyze traffic closely enough to realize that conversion rate has been worse due to disproportionate traffic.  Typically, competitors dip their toe in the water, run away, and call Search Marketing unsuccessful.

I can almost feel people rolling their eyes thinking that this little change doesn’t make much of a difference, but I really must disagree.  As you continue to Optimize, the Barriers of Entry pile on.  You constantly make the landscape more and more complicated for your competitors.  Imagine the complexity created by the Five Mill Tree Method in this purely Geo-targeted version of the Five Mill Tree Method:

1. As you “slice and dice”** …

a.  Competitor CPA will increase and/or conversions will decrease.

b.  Your CPA will decrease and/or conversions will increase.

2. Optimization will protect your margins from competitors who start to bid more aggressively.

3.  As a side effect…

a.  Your newly targeted adgroups will [hopefully] lead to the creation of better targeted ads.

b.  Your history with optimized ads will lead to high quality scores and low CPCs

c.  Increased granularity and profitability.

In sum, Optimization with the Five Mill Tree Method will nearly always lead to…

1.  Increased conversions (or revenue) at the same overall CPA

2. Decreased CPAs (or increased margins, depending on your “metric of success”)

3.  Increased Barriers of Entry.

ONE FINAL NOTE!!!: A very important reminder that the Five Mill Tree Method is not just match type manipulation or geo-targeting manipulation, it’s a combination of all these things.  Further, there are tricks you can utilize with this fairly simple method to make optimization even MORE interesting.

Next entry, I will explain some tricks and techniques that can be utilized with this method.  If not that, I will add in my most recent Google rant…. :-/

Social Bookmarking