Optimization Best Practices: Part 1

July 13, 2022by Michael Bellitte

Optimization Best Practices: Part 1 


A Demand Side Platform (DSP) is a tool to facilitate programmatic buys. Over the years we have seen massive improvements in the technology powering DSPs, but having a hands-on keyboard operator (HOK) remains essential to use a DSP effectively. One of a HOK operator’s greatest responsibilities is optimizing campaigns for performance. In part one of this series, let’s take a look at some best practices to keep in mind: 


Pacing is Your First Performance Lever

Pacing should always be your first performance lever. Before you can dive into more granular optimizations, you need to make sure that your campaign is set to deliver in full, with relatively smooth budget distribution over the campaign’s flight (slight fluctuations in daily spend are acceptable). Effectively spending all of your allocated media dollars with targeted advertising should be considered a success in itself. Leaving money on the table often results in reduced budgets next time around.

Pacing to spend in full is also necessary so that the breadth of manual and automated optimizations remains as expansive as possible. If you are pacing behind, you will be unable to use optimization techniques such as cutting out or reducing bids on the lowest-performing media.

To make sure that your campaign is pacing in full, you should first investigate audience size, frequency settings, and that your bids are aggressive enough to win inventory. 


Consider The Spend Implications

An important concept to understand when optimizing campaigns is the need to balance optimizations on statistically relevant grains, while avoiding optimizations that will cut off too much spend and cause your campaign to pace behind. For example, it does not make sense to spend time adjusting bids on individual sites that make up less than 1% of your overall campaign spend. The juice is not worth the squeeze. On the other hand, you need to be wary of cutting out vast swaths of potential inventory that you need to deliver in full. Picture a scenario where you notice that performance is slightly lower on weekdays. You may be tempted to only serve on weekends. However, when you consider the spend implications, you would quickly realize that you cannot cut 5 days out of a 7-day week without severely impacting spend/delivery. 

As a general rule of thumb, you should not cut out anything that will drop your inventory by over 20% in any one instance. After you make an optimization, closely monitor your spend over the next few days to make sure you are still on track.


Know When To Optimize

It’s just as important to know when to optimize, as it is to know how to optimize your campaign. Let’s take a moment to focus on the when. We recommend optimizing campaigns as soon as there is enough statistical significance to do so. Usually, this means waiting at least 3-4 full days. Then align your optimization schedule so that you are pulling reporting and optimizing at the start of each week (Mon-Weds) so that you can monitor the results of your optimizations during the back half of the week. We do not recommend making any major optimizations on Fridays because you won’t catch any potential issues until the following week.

Another best practice is to limit yourself to making only 1 to 3 optimizations per week. You will want to stay in this range so that if performance improves you can easily tie back which optimization helped. Conversely, if the optimization did not produce the intended result, you will want to quickly figure out which optimization needs fixing. Sticking to these guidelines will ensure a healthy optimization schedule.