In 2020, contextual advertising will become a much more important part of the digital advertising media mix. As the prominence on contextual advertising continues to grow, here are the ins and outs of running contextual advertising campaigns to drive business performance.
Contextual advertising is the use of on-page keywords to first identify the meaning of a page, and secondly to make ad-serving decisions based on that meaning. In practice, contextual advertising allows advertisers to serve up ads on individual pages based on the content of that page.
The uses cases for contextual advertising are robust. This tactic makes it onto most of our ad campaigns. Here are five use cases:
Serve ads to people on pages about a specific make and model of car, which is great for car manufacturers that want to reach people who are researching their next car purchase.
Market to people whose media consumption behavior indicates an interest in a complementary product. For example, if we are selling tickets to the next Adele concert in Los Angeles, then a person reading about news, gossip, stats, or song lyrics for Adele songs should definitely see ads for her next show.
Reach people in hard-to-find demographics, such as LGBTQI. While it’s hard to ascertain someone’s orientation, it’s easy to serve ads on content that speaks to that audience. This is used for any product that wants to align with LGBTQI communities.
Restaurant Consumer Targeting
A restaurant uses contextual advertising to reach people who aren’t just foodies, but people who are on pages about specific types of food and dishes. Let’s say the advertiser is Five Guys and they want to reach burger lovers; they’ll want their ads to appear on listicles about the best hamburgers in Los Angeles.
Catering Company Advertising
A catering company will want to serve up ads on pages that speak to special events in their city. Ads may appear on pages that talk about the specific special events they’ll be serving, such as weddings, bar mitzvahs, and birthdays. Diving deeper into weddings, the ads will appear on pages about wedding registries, wedding planning, and floral services.
The cost of contextual advertising strategies is lower than the cost of other types of advertiser data tactics. This happens for two important reasons. The cost of data by the providers is lower, so the overall share of media spend that is used on data is lower. Secondly, there is an abundance of impression inventory available for contextual advertising, giving ad buyers more flexibility for when, where, and how the ad is delivered.
There are three overall strategies that roll up to contextual targeting:
1. Off-the-shelf contextual segments
These are segments offered by companies whose business it is to track the contents of web pages. So companies will create a predetermined segment of data that consistently and repeatedly reaches people interested in categories like sports, baseball, home and garden, cooking, entertainment, movies, and horror movies.
2. Custom keyword segments
Similar to the off-the-shelf segments, we have the ability to create custom segments in which keywords that we define are the basis for the contextual targeting. This is a custom recipe for our advertisers that is much more granular than off-the-shelf segments. In the case of reaching potential fans of Adele, we’ll create a keyword list of terms related to Adele, such as “Adele,” “rolling in the deep,” “we could have had it all,” and “someone like you.” These key phrases give us people who are interested in Adele’s music.
Based on the sample results, we can then filter out keywords, or add negative keywords. For example, we may exclude “Dell,” relating to Dell computers.
3. Target by site
This is another form of contextual targeting that is more low tech. It doesn’t require the technology partners to do any work. What it requires is some upfront legwork to define the types of sites that your target consumers go to.
There are two examples of this. The first is a very narrow trade example. If I want to market something that is super expensive, such as getting funding for a technology company, and I need to reach people who are in the private equity business, I may target people who are on private equity publications and business publications. I would target a white list of sites that include:
The challenge with this particular list is that these trade publications are owned by media companies that don’t make their inventory available on the open exchange. So the targeting white list still happens after I form a direct deal with the inventory owners to reach users on their sites. The upside is I have a direct connection to the publisher and we agree on terms, deal amount, and CPM. The downside is we don’t have the flexibility of open ad exchange targeting, which lets us run inventory without having to make commitments to the advertiser.
In another example, we want to serve ads to fans of the Los Angeles Dodgers. Maybe we are selling a product like Farmer John’s hot dogs for purchase in store. Since these are the official Dodgers Dog, we’d want to leverage the affiliation with the Dodgers. There are substantially more sites that cater to this interest since this is a much broader target. We would serve ads on the following sites:
Of course, the MLB site for the Dodgers is available too, but they may have special requirements for how partners can advertise on the site.
A common use of contextual advertising is the deployment of local media properties. These are brand-safe environments that everyone is familiar with. A Los Angeles-based brand that wants to reach people in LA-native environments would run on properties including:
Contextual Advertising Vs Behavioral Advertising
Contextual advertising and keyword targeting are often paired up on an ad campaigns. In some cases, they are separated out as separate tactics that reach the same types of users using different methods. Behavioral advertising, or audience targeting, is the deployment of data from either third parties or the advertiser, and focuses on who the user is, not the context of the ad.
So, for example, if I’m deploying behavioral targeting to reach Dodgers fans, the ads may target a Factual segment of people whom we’ve seen at Dodger Stadium. The ad can run anywhere, such as on Yahoo, or while the user is reading news, on Flixster, or searching for movies, or even a Roku when the user is watching on a connected TV.
In some cases, we do combine the two data sets. We may target everyone in a Factual segment who has been to Dodger Stadium, and also limit the contextual advertising to pages that are about sports or Los Angeles news.
Contextual Advertising With Video
There are platforms that allow contextual advertising with video. In addition to the partners listed above, Google Ads and YouTube allow advertisers to reach people who are on videos about specific topics and who are on specific video URLs.
Contextual Advertising and Hyperlocal
Tony Price, our Chief Operating Officer, created this quote and it really got the point across in a recent client meeting on the topic of hyperlocal advertising. “People are the new pixel. Where people go in the real world is as important as where they go in the digital world.” Physical world location is an important part of contextual advertising. The physical form of contextual targeting is where you are at the moment. If the user is at a Starbucks on their mobile device, that physical context is a factor in their mood, thinking, and interest. A person browsing on their mobile at Starbucks is relaxing. They are sipping coffee. They smell coffee. It’s a social environment. Maybe they are in the mood for entertainment, so an ad for an upcoming Adele show may work better in this environment than when this same person is back at work.
A person on mobile at LAX is traveling. By pairing audience data, we can granularly target people who are going to specific cities. The overlap of data sets will be small, but the scale that you do have is pinpoint targeted.
You may know that Google is constantly using code to track what a web page is about. Other companies use similar technology to do the same thing, and use the knowledge in different ways. These companies generally offer three related solutions: contextual targeting (which is the focus of this article), viewability tracking and targeting, brand safety targeting, and fraud prevention.
- Viewability tracking and targeting is the ability to understand when an ad is seen by a user. The definition of viewability in 2020 is that a banner is on screen for one second with at least 50% of the ad’s pixels appearing on screen. For video, the ad should be in 100% view for at least 50% of the duration to be counted as viewable.
- Brand safety targeting uses the same technology to make a determination about the environment of the ad. So, if an advertiser doesn’t want an ad to appear on a page about violent news or with swimsuit photos, these pages can be avoided.
- Finally, fraud prevention lets advertisers avoid serving ads to devices that exhibit behaviors that lead these technology companies to believe there is no one behind the screen. Examples of this are scrolling behavior or clicking behavior on pages that is not consistent with patterns that are seen in people, or excluding advertising on apps that are more ad spaces than content.
Each of the companies below have proprietary technology that offer these capabilities. These technologies are black box, so advertisers take the seller’s word on the technology behind the companies. In practice, we interchange these tools based on the targeting offered, the nuanced needs of the campaign, and past results. More often, we use combinations of DoubleVerify, Integral Ad Science, and Grapeshot, though we use all on a regular basis.
- Integral Ad Science
- Oracle Grapeshot
Beyond the technology companies that make the core contextual targeting capabilities accessible, there are some simple tools and methods that you can use to deploy contextual targeting.
This tool costs $149 per month and allows us to define a list of websites that speak to specific audiences. So, if I want to understand where hip hop music fans congregate on the web, I’d do a search on Alexa and it will give me lists of site that rank well for this search term.
To get site list data, after logging in, go to Competitors, Site Screener. This report outputs a list of sites that speak to that audience. This way, you can get an easy-to-output list of media properties that you can serve ads on.
This tool is primarily marketed as a search engine optimization tool. The platform costs $179 per month, and its primary function is allowing marketers to understand the keywords their site and pages on the site rank for. But a really powerful functionality is the Keyword Explorer—I can type in a search term and get a long list of the top-ranking websites that rank well for the term I entered.
For more information, you can go to the Content Explorer tab that shows URLs and sites that rank for the keyword you are searching for.
We go a step further on some campaigns and define the individual URLs that are important to us. We then go into Peer39, upload the URL list, and target ads on a specific set of known URLs.
Google manual site list search
Google is free. Simply pop in a search and see the URLs that load for the search term. When you search “top sites for hip hop,” often a site called Feedspot pops in, and they have a ranking of the top 100 URLs. I find this method usually overall slower, but that’s the trade-off for free.
List trade publications for your industry
Finally, you, your agency, or your advertiser will inherently know the top sites that are important for your industry. Those URLs, whether they are trade publications or key sources of consumer-facing information, are usually the first place to look when it comes to contextual targeting white lists.
DSP – Demand Side Platform
Finally, you’ll need access to a demand-side platform to serve ads to users with the tools and methods that I mentioned. Some of these platforms include The Trade Desk, MediaMath, and DataXu.
Contextual Advertising Trends
Contextual ads are becoming a topic of discussion as the CCPA, California’s privacy regulation, goes into effect in January of 2020. As eMarketer notes, the market isn’t predicted to make a massive jump to contextual targeting over audience-based or behavioral ads, but in the coming years it’s possible there will be greater emphasis on it. If it’s tougher for advertisers to connect a person to the sum of their behavioral signals, one clear and direct signal is contextual targeting. We know a user is interested in Dodgers when they are on a page about the LA Dodgers.
Secondly, in an opinion piece in AdExchanger, Ed Carrey, CEO and founder of Audience Town, suggests that verticalized ad networks will make a comeback. He says that we’ll see similar companies set up to offer access to consumers in a specific verticalized market based on contextual adjacencies. In this example, an ad network may appear offer targeting to ads on pages or sites that are all about LA Dodgers. In fact, any brand, such as the LA Dodgers, that has a center of gravity actually has an advertising opportunity to develop that ad network. Imagine the LA Dodgers selling access to audiences of LA Dodgers fans based on that site list that I published at the top, layered on with additional targeting data and insights.
These campaigns are infinitely scalable. For our local marketers, contextual targeting is deployed to a small site list or keyword group, or a larger list to a smaller geographic region. We set up campaigns this way because every day, we see the power of digital advertising, and a core tenet for our business is to give local marketers access to the best targeting opportunities in the marketplace.
For our large advertising clients, there is a lot of scalable and available inventory to run across contextual targeting tactics. When budgets are larger, we develop three to five theories about the targeting performance, and set up campaigns to test that targeting performance. For example, we may run a media plan with the following contextual targeting tactics:
- Off-the-shelf contextual targeting from DoubleVerify
- Off-the-shelf contextual targeting from Integral Ad Science
- Custom keyword list from Grapeshot broad
- Custom keyword list from Grapeshot narrow
- White list of sites broad
- White list of sites narrow
Then, as the campaigns run, we look at performance and scale within the campaigns. The core question we want to answer with this setup is which strategy, or strategies, have the perfect combination of:
- Highest scale
- Best business performance (i.e., highest return on ad spend, highest return on investment, lowest cost per lead, lowest website bounce rate, or most pages visited on site)
- Enough inventory to support the size of the campaign
During our optimization decisions, we’ll turn on new sites, cut underperforming sites, and test iterations of these strategies to get to the performance we’re looking for.
Since I first started programmatic ad buying, contextual targeting has been one of the two big data pillars in our business, with the other being audience, or behavioral, data. These campaigns work well because costs are moderate, there is lots of great inventory, and there is flexibility for contextual targeting increasingly built into the ad buying platforms that are used across the marketplace. With the autonomy that the contextual targeting ecosystem provides, we have opportunities to be nimble, launch campaigns quickly, pull data, define insights, and make data-driven decisions decisively.
Contextual advertising will continue to be source of strong campaign performance for the foreseeable future