Advertisers always want value and they want it at scale.
We’re often tasked with the goal of driving value for Facebook advertisers. While our definition of value differs from campaign to campaign, (registrations, app installs, purchases), advertisers always want value and they want it at scale. So what are the steps to driving value? Here at Nanigans our approach is to:
• Start with what you know works. Use historical data to identify multiple target audience segments
• Test various creative elements and ad types
• Collect enough data to optimize to the advertisers’ definition of value
Once we can deliver value, then we can figure out how to deliver more.
In terms of scaling a Facebook Advertising campaign, there are multiple ways to do it and almost all of them add to its complexity.
• Add more locations into the traffic mix such as expanding to more countries. Expanding to new countries means that there are multiple languages required for the creative copy in addition to testing different creative elements.
• Find target audience segments. This requires even more attention to reach, and interest targeting by geography.
• Ensure the campaign set up and strategy is incredibly precise and closely monitor the results. This is important because costs vary country by country and the average return can look different depending on where users are coming from. Plus, traffic flows in at different rates, depending on the time of day and who’s online.
Start with what you know, never stop testing, and continuously iterate based on your learnings.
The first step was to take a close look at the traffic already run to date. This helped to determine where performance looked good, why and how we could get more….Quickly!
Next, we focused on expanding reach. The campaign’s location parameters were extended so that it had active ads running in every country on the map. If there was translated ad copy available, it was used where it made the most sense. To rapidly scale the campaign, we split up the task of creating new ad plans for each language and matching the appropriate languages to each country. For any new country that we wanted to test, we made sure to start with creative that had performed well previously. Although we wanted to scale fast, we also wanted to make sure that we were testing only a few variables at one time
We took a similar approach to defining different audience segments within each country. By breaking up each country into various audience segments, we were able to allocate more budget and bid higher on the best segments within that country, while at the same time, weed out the underperformers. If we were targeting an interest group in one country and it achieved good results, we made sure to test it in other countries. By the time our team had completed the ad plans and pushed the ads, we had almost 70,000 ads running in in over 120 countries around the world.
Creative fatigue and market saturation proved to be among the biggest challenges.
Creating the ads proved to be the easiest part of this challenge. The hardest part was maintaining performance at such a high level of volume. Our team broke up the days into four, six hour shifts so that we had someone working on the campaign at every hour of the day. Creative fatigue and market saturation proved to be two of the biggest challenges that we faced, and the team was eager and ready to take these on.
To keep the creative fresh, we ran small-scale tests and pushed new creative almost every hour of the day. Whoever started the day at 6am set up creative tests to be run later in the day, while launching creative tests that were set up by the previous person on shift. As tests completed, we’d analyze the results and add the winning creative combinations to our existing ad plans to be pushed out to different audience segments.
We also closely monitored ad frequency and configured the strategy group settings to maintain an optimal frequency throughout the day. If the ads were displayed too often, particularly in the smaller reach segments, performance declined at a much faster rate. We also leveraged Ad Engine’s “Audience Dampener”, which can pause down the lesser performing ads within each segment and only allow for a certain number of ads to be active at a given time. With this, ads automatically roll into the testing queue, go through an evaluation process, and then they are either selected to run at scale or paused down.
Market saturation also presented a problem. At some point, no matter how often we iterated on creative, we anticipated that we’d see a natural decline in performance because there was so much traffic running during a short period of time. Following are some ways we worked to slow this process.
- We paid close attention to frequency and made sure to define segments in the cleanest way possible.
- Interest targets were grouped according to size and overlap to create segments that would be large enough to deliver the volume of traffic we were looking for, but small enough that we could assume performance would be consistent for all users targeted by the group and bid accordingly.
- By grouping interest targets that heavily overlapped, it provided better control over frequency so users could be targeted with one ad as opposed to many.
Paying this much attention to detail made a big difference in the long run, and allowed the campaign to scale.
It is possible to efficiently spend one million dollars a day on Facebook advertising.
In the end, we learned a great deal from managing this campaign, all of which has helped shape our campaign management best practices that we have used to manage hundreds successful large scale campaigns. Some key takeaways are:
- Understanding the importance of teamwork and how communication is key when you are working closely in a high-pressure environment.
- Creative iteration is imperative when you are running at scale, and customizing creative to hyper-targeted audience segments can provide lift to performance.
- Frequency is of utmost importance–it’s very important to consider both reach and overlap when defining your audience segments.
And most importantly, we know how to efficiently spend one million dollars on Facebook!