An introduction to
campaign optimization principles
Campaign optimization involves cutting the losses (or under performing segments) from a
campaign while scaling gains (or well performing segments). This optimization process occurs on
many levels, beginning with creatives, then moving on to placements and individuals sites, and
finally on a campaign wide level.
Let’s begin with the Creative level.
The most important element of your media buying campaigns is your creatives or banner ads.
Without going too deep into the details of statistical significance, let’s just say that you need to
have a reasonable amount of data in order to have confidence in your optimization decisions.
For display advertising, a good sample size to have before making a decision is roughly 10,000
impressions. Of course, there are many other factors that can influence the performance of those
impressions, such as time of day, or day of week, but we’ll keep it general for now.
While judging the Click Through Rate (CTR) might seem like the obvious metric to evaluate (it’s
certainly the easiest), the ultimate factor in a creatives effectiveness is it’s eCPA
Cost Per Action
In other words, what is the effective cost per conversion for a given banner? If
there are no conversions, then look at the CTR. If the CTR falls below 0.08% consider it a candidate for being put “offline”.
Another question to consider when it comes to testing your creatives is “How many creatives
should I test at a given time?” The minimum should be at least 2, so that you can at least have a
head to head comparison of performance. The more banners you test, the more data you will
need in order to reach a significant amount of exposure for each variation. You could easily throw
up 10 different banners, but you will need to make sure that if you have a small testing budget, it
doesn’t dilute the amount of the impressions amongst all the banners too much.
One of the most overlooked aspects of campaign optimization is placement optimization. When
looking at your campaign stats and optimizing sites, make sure to look at the placements within
sites. Many times you will find poor performing placements (within sites) that are often “below the
fold” and causing the overall stats for a site or campaign to look bad.
Then just look at the stats of each placement and set them offline as needed:
Placements with low CTR also effectively drive up the effective cost per click (eCPC) to
This is pretty much a required step for optimizing campaigns. This is because during the
campaign creation phase, you can only add entire sites, not specific placements within sites.
Once creative level and placement level optimizations have been made, the next step is to
optimize on the Site Level. Sometimes a site can be beyond redemption, in terms of performance,
and needs to be set offline completely. There can be many reasons for setting a site offline:
• Poor overall CTR(click through rate) on all Creative and Placement levels (with significant
• Poor overall CR (conversion rate) on the clicks coming from a site. (i.e., lots of clicks, but no
• Prohibitive Costs (high eCPMs) resulting in higher costs per click and action. Poor ROI.
Finally, after a campaign has been optimized from a creative, placement, and site level, one of
the last tweaks to optimize a campaign is Day Parting. Day Parting is an option in the Campaign Editor that allows you to specify which hours of the day and which days of the week your campaign will run.
The reason for doing this level of optimization last is because you need to gather enough data
with which to make informed decisions. This is easy by viewing the hourly breakdown of your
For example, we might notice in an Hourly Report that the majority of
engagement and conversions occur during the hours of 9:00 a.m. to 9:00 p.m.
With that knowledge we can now make intelligent Day Parting rules that will run our campaigns only during optimal times. The result is more efficient spending of ad budgets.
Another way in which you can optimize on the Campaign level is by tweaking the Frequency Cap
rate. It’s common to have it set to 3 impressions per user every 12 hours, but this can be adjusted
in order to either limit volume or open more up.