For as long as people have been buying ads, they have been targeting their desired audiences. The science behind this obviously has changed over the years. In the beginning -- say, back in ancient Greece -- it was as simple as putting the name of your pottery shop on a few of your clay pots. This evolved to more location-based models over the millennia, of course, and today we can geo-target your mobile device. End of story? Not quite.
As we think about the evolution of targeted advertising over the past 50 years, there are panel-based "currency" data providers such as Nielsen, Arbitron, and others. These services allow buyers to place ads on specific published content across numerous media, with an understanding of the overall audience breakdown that views this content. Buyers can place their ads on content where their desired audience makes up some percentage of the audience that consumes that content. By doing this across a certain number of publications or shows, they can be relatively confident that they are reaching a certain number of members of their target audience.
This is easy when you're selling a product or service that has a very broad audience -- say, toothpaste. But when you have a very targeted customer you're trying to reach, it can be much more difficult. Other than niche publications clearly aligned with your target customer -- say, knitting magazines or websites -- it has been hard to find enough touchpoints to reach prospective customers easily.
That has changed significantly over the last few years. Let's focus on digital media for our purposes. The core types of targeting available today include the following.
Panel-based data is the most broadly used today, from providers such as Nielsen, comScore, and others. These panels are used as described above -- to understand the overall audiences that consume content provided by a publisher. This "whole milk" approach works well for brand advertisers that have large audiences that are easy to find.
This category includes geo-targeting and geo-derived information such as Nielsen PRIZM clusters that merge information about households in specific geographies. This is much more important today than in the past, given that mobile devices offer information about where audiences are at the moment of the ad delivery, thereby taking location-based advertising to new heights. In mobile devices, this matters a lot, as some of the mechanisms available on the web are either not available on mobile, or much less available due to technical limitations related to cookies.
First-party audience data
First-party audience data is available from either the advertisers directly (data they have about their existing customers) or from publishers directly (data they have about their individual audience members). First-party data is derived either from explicitly provided information or from observed behavior.
On the advertiser side, this is typically CRM data; generally these are either customers or prospects with whom the advertiser has had direct contact. Perhaps the person in question has purchased from the advertiser before, or perhaps that person has signed up for a newsletter. In the case of e-commerce, perhaps the user has visited the site but hasn't purchased, in which case a click-path analysis might derive some information about the person's interests.
In the case of publishers, this information can be captured through registration (which actually tends to be much more accurate than professionals believe; as it turns out, many people don't put in fake information) or from observed behavior (users who read financial news get put into a finance bucket to be targeted when consuming other kinds of content).
Third-party audience data
Third-party audience data is available from numerous providers. Typically these data points are derived from observing the behavior (anonymously) of the end users as they're moving across numerous websites. Sometimes this data is derived from other sources, such as credit card activity matched anonymously to users via cookie matching.
Third-party retargeting data
Third-party retargeting data is available from numerous providers. These companies will typically place targeting tags on both the advertiser and publisher websites and then link those together in order to execute media buys. Because the provider needs to have matched cookies on both the advertiser and publisher websites, typically these services run as ad networks, since they need to close the loop directly. But there are providers that allow advertisers to create their own retargeting cookie pools and reach their customers and prospects over ad exchanges and through their own direct publisher relationships. This is frequently being referred to as second-party targeting.
Look-alike targeting is available from numerous providers as well, which enables the buyer to provide the look-alike vendor or network with a pool of cookies or data definitions. The providers will then find matching audiences who "look like" the users you've provided to them. This allows the buyer to get value similar to retargeting campaigns, but for much larger audiences.
Custom micro-segmentation is available from a few providers. This enables the buyer to specify extremely targeted audiences that are orders of magnitude more targeted than what is available over the open market and that match their ad campaign goals exactly or much more closely. This type of targeting can be used for brand campaigns or for performance.
The types of targeting above are broad bucket definitions, and there are now literally hundreds of thousands, if not millions, of available targeting segments on the market. Vendors should be more than happy to educate buyers (and sellers) on the opportunity and methodologies behind the data segmentation. I highly recommend that one or more buyers within every buying group become an expert in the types of available segmentation and the data models involved.