In a previous post, we showed that ad color combinations shouldn’t necessarily be the same across the same page; some color combinations work excellently at the top of the page, for example, but are outperformed by different ones in the footer or sidebar. Testing is the only real way to arrive at the best format for each position on the page.

We’d like to challenge another piece of ad formatting conventional wisdom: matching your ads’ background and border color to the page background color will make it perform best. We ran a test on HubPages, and found certain shades of gray (the exact shades for each zone were determined by YieldBuild) outperformed white consistently. HubPages’s background color is white:

Going with your background color is probably a fairly safe rule of thumb, if you’re not equipped to test or use YieldBuild. But the figures above should amply demonstrate that relying on a rule of thumb means you’re leaving money on the table.

Paul Edmondson delivered a great presentation at the Web 2.0 Expo in San Francisco today. A copy of the presentation is available here.

We’ve covered a number of insights into online ad format optimization on the YieldBuild Blog, and we intend to share as much data and practical information on what works, and what doesn’t, when it comes to testing and formatting AdSense, YPN and other contextual ads on your site.

Paul Edmondson, YieldBuild’s CEO, has just finished a presentation at the Web 2.0 Expo in San Francisco, in which he explored what we call the Four Pillars of Successful Text Ad Optimization. These are the four core areas that you must focus on, in setting up your testing protocol, in order to maximize the revenue your online ads deliver for you.

Pillar One: Ad Size

Size matters for a couple of core reasons: ad units can vary widely in their performance on a page, and larger doesn’t necessarily outperform smaller. Also, ad networks vary in their ability to sell different ad sizes, so certain sizes might be much more popular and earn more money due to greater advertiser competition.

Pillar Two: Ad Format Attributes

Ad performance can often be lifted 3-12% just by tuning corners, borders, background colors and font colors. You have to test and retest your site to find out—an optimal combination for one zone on a page may not be optimal elsewhere.

Pillar Three: Ad Placement

When considering ad locations, three key strategies to think about are:1) From the beginning, design your templates to accommodate the most popular ad sizes.2) It is important to have every ad count. Additional ads sometimes reduce overall page revenue, so test before committing to additional ads on the page.3) Don’t underestimate below-the-fold ads. Ad units placed at the bottom of a page’s content is often one of the most underutilized high-performing ad units.

Pillar Four: Maximizing the Network

There are hundreds of ad networks to choose from, and understanding the best fit for your site is critical. For instance, certain networks have better advertiser coverage in certain verticals, even if they generally underperform when compared to the larger networks. Sites with little contextual information are often better served with traditional display ad networks. Understand how each ad network pays. While Google, Yahoo, and Microsoft can pay like clockwork, many of the smaller ad networks are less consistent as they are cash flow-sensitive. Make sure you’re comfortable with payment terms before starting with a new network.

In upcoming blog posts (we post every week here, so be sure to add us to your news reader in order to avoid missing any important insights), we will explore each of these pillars in-depth, sharing aggregate data from our large client base (we’ve served over 2 billion ads over the past 6 month private beta period).

ContextWeb and Online Text Exchange, the folks behind a study we reported on back in October that demonstrated context at the page level improved ad performance over contextual matching at the site level (e.g. an ads for computers will perform better on a news site article on computers than it does in a general article in a computer-themed site), have released the results of another study that show more or less the same thing.

This time, the test was on skin-care products instead of computers. But the results point to the same finding: context matters. Branded recognition from the ad enjoyed double-digit improvement when they were shown on a skincare page or on a skincare site, than on a page unrelated to skincare.

What’s important to note from both studies is that context will work equally well at the page level. This means that a publisher does not have to build an entire site around the ad’s vertical—a page on the topic will perform at least as well. (Although driving traffic is an entirely different matter…)

 Attention! Paul Edmondson, CEO of YieldBuild, will be speaking at the Web 2.0 Conference in San Francisco, Thursday, April 24, at 11 am: Maximizing Ad Revenue Through Format Optimization. Be sure to mark your calendars!

One of our customers, Dablink - a Brazilian online photo editing site (http://www.dablink.com.br/) - has been running the YieldBuild Publishers Solution for several months now, and recently had this to say about their experience:

“I must confess that I was quite skeptical at first with YieldBuild’s ability to increase the eCPM of my inventory. I was already running a quite high eCPM, and didn’t believe there could be any margin for further improvement. But since it didn’t cost me anything to give YieldBuild a try, I signed-up for the beta, and I was amazed by the performance. After just a few days, the earnings and the eCPM soared to levels I’d never seen before. Even better, I noticed that some ad placements on the bottom of my pages that weren’t frequently viewed simply disappeared. Through its constant retesting process, YieldBuild calculated that these ads were not performing well and removed them, thus leaving my page with only the ads that I really need.

“It’s very pleasant to have the task of watching your inventory optimized automatically, knowing that you are getting the most out of it without cluttering your page with useless ads.”

Attention! Paul Edmondson, CEO of YieldBuild, will be speaking at the Web 2.0 Conference in San Francisco, Thursday, April 24, at 11 am: Maximizing Ad Revenue Through Format Optimization. Be sure to mark your calendars!

Some people think your goal as a maximum revenue-seeking publisher is to squeeze the largest possible ad units into your page. The reasoning must go: the larger the ad unit, the more likely visitors will see them (and click on them).

The reasoning is perfectly sound, but sometimes testing reaches a different conclusion.

In this test on HubPages, the larger banner performed significantly better than the smaller one in the header region. However, both in the sidebar and in the footer, smaller ad units performed better.

Just goes to show you - unless you test, you won’t know which ad unit performs best within the ad regions of your pages.

Attention! Paul Edmondson, CEO of YieldBuild, will be speaking at the Web 2.0 Conference in San Francisco, Thursday, April 24, at 11 am: Maximizing Ad Revenue Through Format Optimization. Be sure to mark your calendars!