Contributors: Steve Tangsombatvisit, Guillaume McIntyre, Kevin Heung
Search marketing is an area of vast possibility and vast potential where there are hundreds of different ways to do similar things. Global sites like Netflix serve millions of different search ads to people all over the world each day. We try to optimize this traffic to make sure that the right ads are served to the right people.
There are many tools and methods that teams can use to optimize traffic. To go over all that we do at Netflix in the area of search traffic optimization would consume much more than a single blog post. Instead, we will focus on one potential and important area of traffic optimization, negative keywords.
Negative keywords allow advertisers to direct traffic to the right ad group, and assuming you have relevant ads for every ad group (this assumption applies throughout the entire post), can increase click-through rates (CTR), which will positively impact quality score and, all else being equal, lower average cost-per clicks (CPC).
Let's look at a common Netflix scenario. At Netflix, we have ad groups for a variety of keywords related to streaming movies online. We have different ad groups for different keyword combinations and permutations. For example, we have an ad group for the keyword “stream movies” and another ad group for the keyword “stream movies online.”
Sample Netflix Ad Groups
The differences are subtle but important. If the search query is “stream movies online,” we do not want Google to direct traffic to the “stream movies” ad group because it would not be as relevant as the ad group “stream movies online.”
You can shape the traffic using negative keywords. This is a fairly easy operation to do. Unfortunately, for large search marketing teams with many accounts, thousands of ad groups and millions of keywords, managing negative keywords is a time-consuming task.
This is the problem we set out to solve: to manage negative keywords on a large global scale in an automated and relevant way.
The first thing we defined was a set of rules of when and where to apply negatives. There is no real industry standard for negative keywords. After browsing around, we did not find anything that fit our needs. Per the Netflix culture, we defined our own. Here is the process we came up with.
We typically create "exact", "phrase" and "broad" ad groups.
'Exact (E), Phrase (P) and Broad (B)'
Search Query Report
Let’s take a look at a couple of examples.
Search Query: “stream movie”
Ad Served From: “stream movies - E”
Search Query: “stream movie online”
Ad Served From: “stream movies - P”
At Netflix, our Traffic Optimizer audits the search query report daily and apply these rules to all queries, adding new negative keywords as necessary. In the first few days of execution, the Traffic Optimizer generated hundreds of new negative keywords and slowed down to a few per day, as expected. The Traffic Optimizer seemed to have helped increase CTR (see below), but we will be conducting a test to quantify the impact.
We use Traffic Optimizer along with several other in house-tools that help us move fast and streamline operations. This allows us to manage hundreds of campaigns in an automated way without much manual involvement.
Traffic optimizer is just one of the many in-house tools that we’re developing to make sure we deliver the right advertisements about Netflix to the right audience at the right time. If you’re interested in helping us stay on the bleeding edge of advertising technology and spreading the word about Netflix, we’re always looking for talented engineers. Head on over to our jobs page to learn more!