Advertising in search is often difficult to conceptualize for many brands because of how different the live auction and targeting methodology is compared with traditional media planning and buying. Media strategies are built on market research, which always includes target demographics or personas.
Specialty channels like search — where the targeting strategy is based on keywords and not necessarily on demographics or personas — cause a methodological disconnect for some brands and traditional planners and buyers. The result is that search often becomes just another impression/cost buy, which is not how it should be used.
Google has made strides in providing additional information on demographics and the ability to target personas, specifically through the Google Display Network, but when buying on the search engines themselves, most demographic information is not great. However, here is a potential way to help you bridge that gap between demographics and search keyword strategies.
Although this tool is supposed to be more for display, you can still get demographic information based on keywords. The data comes from Google DoubleClick network, but it still represents user cookies that are looking for and engaging in content related to the keywords you enter.
It can be found here: https://adwords.google.com/da/DisplayPlanner/Home
Once you go to the Display Planner, click the drop-down arrow that says “Search for new targeting ideas using a phrase, website, or category.” Enter the keyword for which you want demographic info, then click the button that says “Get ad group ideas.”
This will bring up a screen with gender, device and age bucket information — similar to the information that Google Universal Analytics provides. The benefit is that you can use this to be proactive instead of reactive.
Now, there isn’t a way to get a meaningful download with the demo data, even though it appears on the screen in these charts. The reason for this is that the purpose of this tool is to convince you to buy more by giving you additional display suggestions.
However, if there is a chart appearing, that means those numbers have got to be in the source code:
This means you can capture that information by extracting it from the source, which will provide you with nice, clean datasets of demographic information on your keyword.Now, if you are like me, one keyword really is just not enough. I want the information on all of my keywords… and I tend to have a lot of them.
To get the information for ~700 keywords, it would probably take someone manually collecting it for four or five hours — which is what I had one of my tremendous associates do for me to see if there were any real insights, and there were! I’m in the market for a new smartphone, so that’s what we looked at: smartphones and providers.
To make the data easier to understand, I put it into TIBCO Spotfire (Tableau also is a good choice) and made this dashboard to see what we got:
We also created some filters by word type, brand, stage in the funnel, product type and so on.It was interesting to note that people were more likely to search for purchase-based words on their desktops and that some brands had 10- to 15-percent difference in gender and age searches.
Additionally, MetroPCS and Google phones had the largest group of older searchers, while Samsung and Apple had the youngest and were also searched for 10 percent more on mobile.
Regardless, when keywords have the added benefit of some representation of demographic information, it creates a bridge of communication to traditional channels that buy based on cost, impressions and demographics.
If you want to collect this data in a more efficient way, there are several choices, but there are risks. We don’t allow any automated collection/scraping of Google for many reasons, including its terms and conditions, which you would be violating.
And Google is very likely to catch you if you don’t understand its systems, resulting in an hour ban of your IP address. This can be very bad, especially if you are buying advertisements in Google, as it disables your ability to do that.
With all that said, since Google collects everyone’s website content (unless not allowed in robots.txt), this is how you would be able to accomplish collecting their information. I would probably use something like Selenium (a browser-simulating add-on), unless you’re a senior Java developer, as the time control rates are easier to control and understand.
I would use the Selenium nodes in KNIME, as it allows you to import variables and manipulate and store data more easily than the web browser application. And that flow would look something like this:
Services such as Experian Audience View also offer panel-based demographic search query information. However, you’ll need to pay for such services, while the Display Planner tool is free.