Posted Monday, September 29th, 2008 at 3:21 pm by John Brian (91 posts)
Inspired by a post on Occum’s Razor, I recently applied for and received access to Google’s new Ad Planner (it took about a week for my application to be processed, so if you’re intrigued, don’t delay – apply right now. It’s okay – I can wait…). Test driving the tool, I found three things:
- It’s immensely cool for a marketer, voter targeting guru, or demographics aficionado
- It’s yet another example of how awesome Google is for giving us tools like this for free
- It will make some privacy advocates likely go bonkers, as happened with some other Google innovations
Here’s the reader’s digest version: the Ad Planner leverages Google’s gigantic barrel o’ data to help users understand what sites people browse, based on a variety of demographic information and their other online behavior. It then helps you to build a list of sites to run advertising on, and even provides the ability to export your target list in MediaVisor (so that’s where that DoubleClick acquisition went).
The long version, as always, is a lot more complicated – the tool is very powerful, and not just for advertisers. Follow me below the fold to learn more…
First, let’s try an experiment to show how the Ad Planner works. I’ll include some screenshots for the benefit of anyone still waiting for their account to activate (if that doesn’t include you, just turn on AdBlock – I’m pretty sure that I’ve talked about ads enough in this post for the images to get hit). Let’s start with a hypothetical goal: we want to get young people to use Barack Obama’s Vote for Change site to register and early vote.
Let’s start by going to the research tab and entering in some criteria. We start out with an audience of 230 million potential visitors, 100% of the country (the Ad Planner currently only works in America. Fortunately for our hypothetical, so does voting for President of the United States). Ad Planner lets us pick from a variety of languages, and Vote for Change includes English and Spanish versions, so let’s start with those sites. This has virtually no effect on our reach, unsurprisingly, but if we’d chosen a less common selection of languages in the US, such as German, French, or Japanese, we’d be looking at a combined 2.8 million visitors, 1.3% of the total market. Unfortunately, there’s no state level targeting yet – we could cut out states that don’t require pre-registration to vote, like Wisconsin and Minnesota.
Next we start diving into the regular demographics – they include four core demos that marketers like to divide by: gender, age, education and household income. Since we’re running a campaign in 2008 and not 1908, we won’t touch the gender criteria, but let’s go ahead and segment by age. We just want to reach young people, but they have to be able to vote, so let’s check off the 18-24 box only. That cuts us down to 15 million hits already, so we’ll go ahead and add in the 25-34 group, bringing us up to 45 million visitors, 20% of the web browsing public.
For our next audience, we’ll go ahead and select just those individuals who are at least attending college, as well as those who’ve graduated. This brings us down to 28 million visitors, 1 in 8 Americans online. We’ll leave the household income demo filter alone – here’s one place that I wish that Google was more precise, as we might want to allow everyone with “some college” but segment only certain income levels of graduates, with the assumption that income levels of individuals currently in college are wildly off from the demographic they most resemble. So leaving this alone, we hold steady at our current levels. Here’s what our settings look like now.
Now we get to the really cool part: we ask what other sites people are visiting. For this campaign, we’re going to need to recruit people who are willing to give out some information about themselves on the internet. That means social networking sites: Facebook, Flickr, and (sigh…) MySpace are three of the most likely to pay off. We can keep putting in more sites – I’ve got the filter set to “or” at the moment, so every site we add will increase these numbers. With just these three, we get 13 Million visitors, which is probably a manageable number.
Now it’s time to start going through the actual sites. For each site, we see how much of their audience we’re likely to reach with each site. These range from about two-thirds, with Facebook or Yahoo, to around 23% with Amazon, to 9% with the New York Times, all the way down to 0.7% with Penny Arcade. There are some weird results in here (the Pirate Bay and WordPress are more popular than Fox News… heh.), but since they’re backed by the power of Google, I’ve got to assume that they’re credible.
Beyond this, though, we can go through the sites on the list and see if they accept advertising, and, if so, what types we can place there. If we want to learn more about a given site on the list, we can just click on the tiny green icon () to learn more about that site. If we look into the Apple.com, for example, here’s what we see. This data comes from a variety of sources and given Google’s concern for privacy, I think they’ve done a good job of balancing the desire to have a useful tool with not revealing anything about individual browsers – most of the aggregate data is nothing that Compete couldn’t provide as well.
With all that said, Google did put some restrictions on the tool. A big one is that if the sample size gets too small, they won’t provide recommendations: this means that you won’t be able to pull data on very niche sites, or combine enough demographics to get a user pool smaller than about 10k or so. Sorry, folks, but you won’t be able to get any data on the traffic of Beaconfire.com, though I checked several large national non-profits and was able to get some recommendations. It’s definitely worth checking your own organization’s name to get some idea of where else your users are visiting – as they say, the results may surprise you.
I think that this is a fantastic tool for a variety of purposes – I’ll probably dedicate a separate post to describing a few. Sign up for the beta today to give it a try, and leave your thoughts on the tool in the comments, then stop by again in about a week to hear what some of the Beaconfire team found by using the Ad Planner when building campaigns.