YieldBuild’s guiding goal is to extract the highest revenue possible from each page it is delivered on. It really is indifferent to the conventional wisdom with respect to what colors, styles and placement ends up achieving that maximum revenue—it is just after what works.

Imagine our surprise when given the freedom to display ads all over a page, YieldBuild has very often allowed certain ad zones to wither away, resulting in fewer than the maximum number of ad spots on a page.

Why?

We’re not sure. It could be that having too many ads occludes your content and puts off visitors. Or clicks are spread more thinly among ads with a lower average CPC. But these are just guesses—what we are certain of is that if you want to maximize revenue, you don’t necessarily want to maximize the number of ads on a page.

Since the picture is decidedly mixed (sometimes you do want to max out those ad zones), YieldBuild users rest a bit easy, because YieldBuild automatically makes that determination for each page’s unique preferences.

But if you’re the DIY type, be sure to try reducing the number of ad zones on different page types occasionally to see if it has a positive, measurable impact on the page’s revenue generation. You might be as surprised as we were.

If everyone responded to ads the exact same way, then you’d see remarkably little variation in the formats and approaches that sites use to expose their visitors to their advertisers’ messages. You could format your ads to look like everyone else’s and stop worrying about it.

However, it’s been clear from YieldBuild’s own testing that each site’s body of visitors has its own preferences and ad sensitivities, and that is often due to the range of different responses among individual visitors. Where visitors came from to visit a site can often give a glimpse to what ad formats they respond to, and which they ignore.

Think about it. Visitors coming from a boisterous social community site with loud, visual ads (think MySpace) might completely ignore subtly-formatted ads. Likewise, visitors from a text-heavy informational site with muted colors (like Craigslist or Digg) might be put off by obtrusive, wildly-contrasting colors. It all depends on what their eyeballs have become accustomed to, and each visitor’s browsing history paints a different story.

Ideally—and this is, forgive the shameless self-promotion, something YieldBuild already does—you’d correlate incoming traffic source with visitor response to an array of text ad layouts, and then build your ad layout engine to respond appropriately as the mix of traffic sources to a page changes.

If this is beyond the capabilities of what you can do on your site and you’d still like to go it alone, you might try to identify the 3-5 most prominent incoming traffic sources to your site, and test different layouts on pages where just one of these traffic sources predominates. Then, you can make an educated determination of the optimum mix of layouts for pages where incoming traffic is drawn from a number of sources. It might take a lot of work, but as they say in the retail business, knowing your customer is the key.

It would make ad format optimization a lot easier if all you had to do was come up with a set of best-performing ad layouts for a page, and be done with it. If only it were that easy!

The reasons it unfortunately doesn’t work this way are two-fold:

1) A page’s traffic sources typically change over time. Each traffic source’s visitors have varying responses to different ad formats and layouts, depending on what site they had visited before. As the balance of visitors tilts in a new direction, the page’s ad layouts should respond accordingly.

2) For sites that rely on repeat visitors, ad blindness can set in, and without some variation, repeat users will unconsciously ignore ads (even those that worked like gangbusters the previous month), reducing clickthrough. Continually coming up with new, best-performing ad layouts is an undeniable need if your goal is to maintain or increase clickthrough rates.

This might sound more daunting than it is. It does require, however, continuously generating new layout variants, testing to see which perform best, and adding those to the collection of layouts cycled on the page. It also, naturally, requires putting a few former great layouts out to pasture. But the payoff is worth the effort.

This insight is what drove us to build this functionality into YieldBuild – i.e. YieldBuild continually creates and simultaneously tests new layouts, while disposing of layouts that no longer do the trick.

The goal of time-consuming and laborious ad format optimization should be to find the holy grail of ad formats, one whose color, placement and format magically achieve the optimum clickthrough rate and revenue for the page. Right?

Wrong.

What must be one of the strongest testaments to the adage “the whole is greater than the sum of its parts” in this business, YieldBuild has consistently found that a set of different “good” layouts beats a single “great” layout in overall page performance. Although, head to head, any one of these “good” layouts would underperform the best-performing layout, a system where all the “good” layouts cycle on the page actually bests the supposed best layout.

What this means in practice is that once your optimization testing has yielded an array of high-performing layouts, you’ll do yourself a favor by filtering them all into your ad formatting system, rather that performing even more optimization testing to find the supposed standard-bearer.

Keep in mind, though, that the set of high-performing layouts will continue to change over time, so you’ll need to generate and cycle in new tested layouts, and continue to retire old layouts that cease being able to perform up to par.

YieldBuild does all of this automatically, but if for whatever reason you’d like to implement your own optimization scheme, please keep this insight in mind.

It stands to reason that ads that naturally integrate into the general style of a page are more likely to get clicked on, since they suggest they flow with the content of the page.

And often this is the case. On many pages, we’ve been amazed to see YieldBuild slowly but surely, from among the millions of style and placement permutations possible, arrive at a format the blends beautifully into the style of the page, from color, placement and even border style.

But not always.

Quite often, we’ve actually seen the contrary – a style completely incongruous with its surroundings outperforms the blended style. On a background of pastel colors, a bright red ad will turn out to be the best performing. A dull gray ad can surprisingly do better than all other variants on a brightly-colored page.

We here at YieldBuild have seen it happen enough times to know that there is no hard-and-fast rule with respect to format blending and performance. Each page should be optimized by itself, according to its unique traffic blend, to determine whether eye-dazzling, subtle, or a blend provides the greatest receptivity among visitors.

Tip: Try introducing a subtle/blended approach for one ad zone where you’ve been using bright/contrasting colors (or the other way around – one bright ad where the rest of the page has been using blended colors) and see what kind of impact the change has. As a second step, iterate towards more-subtle or more-contrasting colors to hone in on an optimum color/format. Finally, be sure to shake things up a bit, every 100,000 PVs or so, to see how your audience responds to changes.

Bigger is better, right?

In the world of text ads, simple sayings like that rarely hold true. In this specific case, they really don’t.

Despite the commonly-held perception that bigger ads are more likely to capture the attention of a viewer and entice him/her to click, our experience using YieldBuild across a number of very different sites paints a complex picture. On some pages, bigger ads perform better. On others, more, smaller ads perform better. Seemingly without rhyme or reason, different visitors respond differently to differently-sized ads on different pages.

Wow, sounds like something the Mad Hatter would say. But it’s true. Some visitors might find larger ads too aggressive and will avoid them. Others need something that stands out from the page’s text to warrant attention. This sort of guesswork is interesting, but thankfully YieldBuild doesn’t require us to really understand the whys of visitors’ behavior with respect to text ads. It just consistently knows the hows of ad formatting and optimization and what it takes to achieve higher clickthroughs and revenue.

If you decide to go it alone, just keep in mind that the best ad size on one page of your site won’t necessarily be the best everywhere—make sure to optimize each page or set of pages by their format/layout and traffic type, and be sure to reoptimize when traffic patterns change.