Contextual advertising is the approach most advertisers use and there is merit to this approach, but ultimately it is the user you want to target and not the content itself. Yes the content can lend context to what the user may be interested in, but it is the user who will make the purchase not the content.
The app name is a common piece of information passed in a mobile advertising request, or in the case of mobile web the page URL. This information helps the advertiser determine the content of the app or page that the ad will be displayed on, and the advertiser attempts to show an ad that relates or will appeal to a user viewing it.
If an automobile advertiser wants to target users interested in buying a new vehicle, then it is a good approach to show them an ad if they are viewing automobile content. But if that user is interested in buying a new car, the content they are looking at should not matter, the automotive advertiser wants to reach them not just when they are using an auto app. The user may be more in the mindset to buy a new car when viewing automotive content, but ultimately this person is the target user the advertiser wants to get their ads in front of regardless of the content.
Unfortunately, it is harder to track users on mobile than it is on desktop. Users may have multiple mobile devices and cookie tracking does not work with apps, so advertisers must use alternative methods to track a user between devices and across apps and mobile sites. Fortunately, the mobile industry has matured in this respect with the advent of unique Device Advertising IDs which do not personally identify a user, but allow them to be tracked between different mobile devices. Using this approach is called deterministic matching and has a high accuracy rate.
Device Advertising IDs are not currently accessible on mobile websites, so advertisers must use a less accurate method called probabilistic matching. This process involves comparing a set of data points such as geo location, user agent, screen resolution, browsing and behavioral data, etc. to try and identify separate visits as being from the same user. Many companies have their own proprietary algorithm to determine these matches, but regardless of what data points are used it is much less accurate than deterministic matching.
Since advertisers have a much harder time tracking and targeting individual users on mobile, they must fall back to doing contextual targeting. If done properly contextual targeting can be effective, but usually there is limited requests from content that matches the targeted criteria. If an advertiser could target the same user regardless of what they are viewing, then there would be a lot more ad opportunities for them to display ads on.
Newer apps face the challenge of being unknown to the market in terms of what their content type is and ultimately what their user demographics are, which is why they typically receive a much lower fill rate. Typically, their fill rate does not increase until advertisers become familiar with their content and start to recognize their users.
Advertisers must do whatever they can to serve ads to their target audience as often as possible, but it is not a perfect process. Publishers can help advertisers by providing as much information as possible about their app in the different app stores and passing in as much information (device advertising ID, app name, app store URL, etc.) as possible with each ad request. Both the advertiser and publisher need to remember that ultimately the user should be targeted and not the content. As the mobile ad industry matures this will hopefully become easier for advertisers to accomplish without sacrificing the privacy of users.