Tag : display advertising

Not all impressions are created equal.

We don’t think about run of site (ROS) ads frequently as we don’t often use them. We try to be very intentional with our targeting. However, we recently had an engagement where we were asked to design ads for a display campaign on a popular industry website. The goal of the campaign was brand awareness (also something to avoid, but that’s for another post). The client was engaging with the publisher directly. We recommended the placement, designed the ads, and provided them to the client, figuring that was a done job. The client later returned to us to ask for more ad sizes because the publisher came back to them suggesting run of site ads because the desired placement was not available.

Some background for those less familiar with display advertising

If you are familiar with placement-based display advertising, you can skip this whole section. For the relative advertising novices, I’ll explain a little about various ad placements, their nomenclature, and how ads are priced.

An ad which is much wider than it is tall is generally referred to as a billboard, leaderboard, or banner ad. These are referred to as such because their placement on webpages is often near the top, although that is far from universally true, and even where it is true they often appear lower on the page as well. In our example on the right, which is a zoomed-out screenshot of the Lab Manager website, we see a large billboard banner at the top of the website (outlined in yellow), multiple interstitial banners of various sizes (in orange) and a small footer banner (green) which was snapped to the bottom of the page while I viewed it.

An ad which is much taller than it is wide is known as a skyscraper, although ones which are particularly large and a bit thicker may be called portraits, and large ads with 1:2 aspect ratios (most commonly 300 x 600 pixels) are referred to as half page ads. Lab Manager didn’t have those when I looked.

The last category of ad sizes is the square or rectangle ads. These are ads which do not have a high aspect ratio; generally less than 2:1. We can see one of those highlighted in purple. There is also some confusing nomenclature here: a very common ad of size 300 x 250 pixels is called a medium rectangle but you’ll also sometimes see it referred to as an MPU, and no one actually knows the original meaning of that acronym. You can think of it as mid-page unit or multi-purpose unit.

As you see, there are many different placements and ad sizes and it stands to reason that all of these will perform differently! If we were paying for these on a performance basis, say with cost-per-click, the variability in performance between the different placements would be self-correcting. If I am interested in a website’s audience and I’m paying per click, then I [generally] don’t care where on the page the click is coming from. However, publishers don’t like to charge on a per-click basis! If you are a publisher, this makes a lot of sense. You think of yourself as being in the business of attracting eyeballs. Even though to some extent they are, publishers do not want to be in the business of getting people to click on ads. They simply want to publish content which attracts their target market. Furthermore, they definitely don’t want their revenues to be at the whims of the quality of ads which their advertisers post, nor do they want to have to obtain and operate complex advertising technology to optimize for cost per view (generally expressed as cost per 1000 views, or CPM) when their advertisers are bidding based on cost per click (CPC).

What are Run Of Site Ads and why should you be cautious of them?

You may have noticed that the above discussion of ad sizes didn’t mention run of site ads. That is because run of site ads are not a particular placement nor a particular size. What “run of site” means is essentially that your ad can appear anywhere on the publisher’s website. You don’t get to pick.

Think about that. If your ads can appear anywhere, then where are they appearing in reality? They are appearing in the ad inventory which no one else wanted to buy. Your ads can’t appear in the placements which were sold. They can only appear in the placements which were not sold. If your insertion order specifies run of site ads, you are getting the other advertisers’ leftovers.

That’s not to say that ROS ads are bad in all circumstances, nor that publisher-side ad salespeople who try to sell them are trying to trick you in any way. There is nothing malicious going on. In order to get value from ROS ads, you need to do your homework and negotiate accordingly.

How to get good value from ROS ads

Any worthwhile publisher will be able to provide averaged metrics for their various ad placements. If you look at their pricing and stats you may find something like this:

One good assumption is that if people aren’t clicking the ad, it means they’re not paying attention to it. There is no other reason why people would click one ad at a much higher rate than others. Averaged out over time, we cannot assume that the ads in those positions were simply better. Likewise, there would be no logical reason why the position of an ad alone would cause a person to be less likely to click on it aside from it not getting the person’s attention in the first place. This is why billboard banners have very high clickthrough rates (CTR): it’s the first thing you see at the top of the page. Publishers like to price large ads higher than smaller ads, but it’s not always the case that the larger ads have a higher CTR.

With that assumption, take the inventory offered and convert the CPM to CPC using the CTR. The math is simple: CPC = CPM / (1000 * CTR).

Here, we see those really “cheap” run of site ads are actually the most expensive on a per click basis, and the billboard banner is the cheapest! Again, even for more nebulous goals like brand awareness, we can only assume that CTR is a proxy for audience attentiveness. Without eye tracking or mouse pointer tracking data, which publishers are highly unlikely to provide, CTR is the best attentiveness proxy we have.

With this information, you can make the case to the publisher to drop the price of their ROS ads. They might do it. They might not. Most likely, they’ll meet you somewhere in the middle. By making a metrics-driven case to them, however, you’ll be more likely to get the best deal they are willing to offer. (ProTip: If you’re not picky when your ads run, go to a few publishers with a low-ball offer a week or so until end of the month. Most publishers sell ads on a monthly basis, and if they haven’t sold all their inventory, you’ll likely be able to pick it up at a cut rate. They get $0 for any inventory they don’t sell. Just be ready to move quickly.)

The other situation in which ROS ads are useful and can be a good value are when you want to buy up all the ad inventory. Perhaps a highly relevant publisher has a highly relevant feature and that all ads up to an audience you want to saturate. You can pitch a huge buy of ROS ads which will soak up the remaining inventory for the period of time when that feature is running, and potentially get good placements at the ROS price. Just make sure you know what you’re buying and the publisher isn’t trying to sell their best placements on the side.

Lessons

• Run of site ads aren’t all bad, but novice advertisers can end up blowing a bunch of money if they’re not careful.
• Regardless of placement, always be mindful of the metrics of the ads you’re buying.
• Even if your campaign goals are more attention-oriented than action-oriented, CPC is a good proxy for attentiveness.