A Walkthrough of Google Ad Planner Beta

I took a few minutes this afternoon to play around with the new beta of Google Ad Planner.

After logging in to my Gmail account, which I rarely check, on a whim earlier today I saw an email with subject "Welcome to Google Ad Planner!" How exciting... although it was dated July 14th. The good news was that the link to begin using Ad Planner inside the email was still active and within 30 seconds I was on my way.

Here are my impressions after playing for about 15 minutes:

It's Pretty Simple

It takes a good 15 minutes of playing around with the tool to really understand what the heck you're doing (I spent a few minutes just selecting and unselecting things to see what fields changed in result). Once you figure it out you realize that it's really simple. Any time you need something from the tool you'll be in for 5 minutes and then on your way to do something with the list of sites you've built.

Maybe Too Simple

The beta does feel like it's holding a bit back. Google is known for their ability to take big data sets and understand what things mean in human terms and so far the tool allows you to select and restrict data sets in a pretty direct way. It sort of feels like an Excel template wrapper built on top of a linear data set. But this isn't bad, just not earth shattering.

They do mention that the category that each site is placed into is determined by automatic tools. I learned this after digging for more detail on why gigaom.com, Om Malik's technology blog, was placed into the ISPs category. It looks like the ISPs category includes mobile technology which is a direction that I know Om is leaning these days but a quick peruse of his homepage turned up one out of six posts on mobile and a single category heading on the topic.

Focused Toolset

This tool is going to be great for a very small subset of people (a set of people which Google Ad Planner can likely tell you more about)... what other sites do people who visit Google Ad Planner also visit?

What It's Good For

The key intuition you can get from Google Ad Planner right now is really this: assuming I know a site that I convert a lot of good visitors from now (likely the result of data I've collected over time from a CPM campaign) I can use this tool to find other sites that get similar visitors. The composition index score is a number made of pure magic that when higher numerically represents a higher correlation to sites I'm using to baseline my audience.

In English Google says it is "how concentrated your audience is on a specific site relative to internet users within the country you have specified" and in Googlish (the language used by the smart people who work within the Google empire) it "is calculated by dividing the site's reach into the audience by the percent reach into the audience on the internet within the country you've specified." Holy guacamole. I'll take your word for it Google. I think that sentence needs some parenthesis in there, order of operations?

Demographic and Geographic Data

The secondary intuition you can get from Google Ad Planner right now is a list of sites that contain a high percentage of visitors within a range of country, gender, age, education, and household income. Gender and country are probably less of a range than the others.

Just for the heck of it I spent a few minutes configuring the tool as if I was theladders.com, a job site for $100k plus jobs, looking for people who I figured would be most likely in my target audience using the filters along the left side of the interface. Interestingly opentable.com had the highest composition index among well educated, high income, middle-aged men and women.

Other not so surprising sites included Barrons.com and Williams-Sonoma.com. DrudgeReport.com and JetBlue.com also made the list… I'm learning more about how the wealthy stay that way. The resulting list is a good representation of the secondary value I mentioned above.

Now for the key value, I entered a site into the online behavior list that I figured would be a good source of prospects for theladders.com. I guessed msnbc.com. The real value of course would be in using a site that performs extremely well among real online advertising campaign data from theladders.com.

Anyway, the list was quickly revised to include a large number of sites I'm not familiar with. I had ever heard of four out of the top twenty results. I think that's the point of Google Ad Planner. These are not sites I would have otherwise added to my online campaigns and they have a good concentration of people that are similar to my best online prospects. Value delivered.

It is worth mentioning that I am making an assumption about the value of the 'online behavior' comparison which is that I'm assuming Google knows something more about each site than the demographic details it lets me filter by. I figure they have online user path information that shows visitors surfing around the web and when these sessions include two or more sites they draw a correlation.

This correlation is of course stronger than any comparison of age, gender, income, etc because it represents what people actually do online, which if you know user-reported data you'll know is much more valuable than what a user directly tells you. Other than just figuring that this is how they do it, they also don't adjust your demographic selections when you enter a site or list of sites you want to compare visitors against. If they were just scoping your search to the demographic data of the sites you enter I figure they would go ahead and realign your demographic filters to match your site list. They still could just be doing this, I don't think so though.

Anyway, back to the data in action. For first time advertisers the value here comes from the progression from the demographic matching list to the site comparison feature. First you start with a list of sites with a high density of visitors that fit your basic demographic requirements and you run some online campaigns to see which sites perform the best for you. A few hundred thousand bucks over three months and I figure you've got a great list of sites that are performing the best for you.

Then you come back to Google Ad Planner and enter your original target demographic filters (or maybe don't, there is interesting value here either way) and then add your list of online behavior site filters. Now you get another list of even more targeted sites that you can put into your online advertising campaigns.

Assuming Google's data is good pretty much all of these sites should outperform the original list of sites you pulled from Ad Planner. If so, value delivered. Some sites will ultimately perform better than others and your ad management platform will take you from there. For example Doubleclick with pixel optimization implemented will automatically allocate your budget in the best places as the winning sites take statistical significance over the rest, just like it did with your original list of demographically correlated sites.

The process I mentioned here involved "go plug the list of sites into your online advertising campaigns" twice which is why Google created Ad Planner and precisely why the export button on the page has an 'export to Mediavisor CSV' option conspicuously presented. Mediavisor of course being Doubleclick's online advertising platform (owned by Google). Back duly scratched.