Ad Quality Report: Direct Sales Has A Big Edge On Viewability

Integral LogoDespite the benefits of RTB, ads purchased through direct sales still have the highest quality when it comes to viewability and potential risks, according to data from Integral Ad Science.

In its 2012 Semiannual Review for Q3 and Q4 2012, Integral, previously known as AdSafe, showcased data related to its new TRAQ Score, a quality media score designed to help advertisers understand the environment where their impressions lie.

“A lot of what Integral has done to date is help companies figure out environments they don’t want to be in; a defensive strategy,” Integral Ad Science CEO Scott Knoll told AdExchanger. “The whole idea of the quality metric and the new quality score is to help the advertiser go on the offensive.”

TRAQ Scores look at the environment of the ad placement, such as whether the site is professionally written and what else is on the page, including the number of other ads; the exposure of the ad, such as how likely the ad will be seen and if there is a possibility of click fraud; and the share of view, including the size of the ad and what else is in the window.

Looking at the four types of ads — direct sales, exchanges, networks, and hybrids — direct sales results in the highest average TRAQ scores from Integral, while exchanges saw the lowest average, though not much behind hybrids and networks.

Integral also broke down its findings, looking at some specific risks associated with online advertising. When it comes to viewability, direct-sold ads ads, again, were seen for the longest amount of time. Of direct ads, 56% were in view for more than one second, 37.8% were seen for more than five seconds, and 22.7% were in view for more than 15 seconds. Exchange ads saw the lowest viewability, as 37.6% were in view for more than one second, 26.1% were seen for more than five seconds, and 15.7% for more than 15 seconds.

By Time Frame

Overall, Integral Ad Science found that the share of inventory associated with risky content decreased between the third and fourth quarters of 2012, from 19.8% to 17.4%.

In the report, Integral noted, “In the third quarter, publishers carried the lowest proportion of high-risk inventory at 2.1%, while networks carried the highest at 5.1%. However, in the fourth quarter, publishers’ high-risk content increased to 3.6% and networks carried the lowest proportion of high-risk inventory at 3.2%, though they still carried the highest proportion during the two quarters combined.”

Looking at the specific categories of risk, Integral tracked the percentage of ads that showed up on illegal download sites, adult content sites, and sites with offensive language, drug or alcohol references, and hate speech. The specific categories fluctuated between the third and fourth quarters, based on whether publishers, exchanges, or networks served the ads.


Knoll addressed the sometimes large differences between Q3 and Q4: “A lot of it depends on the inventory that’s available at that time. It’s fairly irregular; there’s not really a regular pattern.”

Despite the constant changes, Integral and Knoll hope that the new TRAQ score can help marketers take a more active role in their advertising and media buys.

“Once we started to move to programmatic buying, the automation is focusing on the person or the audience,” Knoll said. “In that environment, where you are focusing purely on the person and where they are, those people may be on sites that you don’t want your brand associated it. It turns out to be a great dumping ground for bad inventory that someone can’t sell directly, but in a blind pool, they can get money for it.”

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Alejandro Correa

    while these results are not entirely surprising, I wonder if they are skewed by the fact that it is harder to gather the data for ads bought through exchanges (because of the iframe daisy-chains).

  2. Joe Garis

    I don’t believe this study was done right. For it to be done right the study should be isolated to a group of 10 sites (the constant). These sites should all sell direct ads and some ads on exchanges or networks. Then buy ads from those 10 sites direct and then buy from those 10 sites on an exchange and network. From there look at whether there was a quality difference on those 10 sites between direct/network/exchange. The way this study was done there is no way exchanges and networks would not produce lower quality. In the real world us media buyers can use and exchange or network and hand pick which sites we want to bid and buy from. You can even use audience targeting, and specify a whitelisted set of sites this audience should be shown ads on. This article makes it sounds like buying on exchanges is a purely machine gun approach where you buy an audience and your brand gets screwed. That’s up to you, the media buyer, to control that. If you can’t control that, get a better DSP. So while this study would discourage someone new from exchange/network buying, the reality is you can control a campaign, choose the sites and get quality with a lower price. Sorry, but this study was a waste. I could have told you what you found and saved you lots of time and money. Redo the study as I stated and that will be very informative.

  3. Good question Alejandro, this analysis is incomplete without a breakout of unknown impressions vs. known for each source. Without this data I would not form any type of conclusion with what is presented.

  4. Ben Trenda

    Hey Joe, you have a good point that there is variability among sites/ publishers. But even if they had done the study as an “apples to apples” comparison (don’t know if they did or not), the result of the study should be the same.

    If you understand how inventory gets into RTB, you understand that the publisher’s server serves the direct sold creatives first nearly 100% of the time. Very few exceptions. This means the best inventory doesn’t make it into an RTB environment, and therefore, it doesn’t matter how good your DSP is: the ads bought direct will still be more visible, earlier in the user session, and on better pages/ placements. One DSP may be able to help you filter better than the next one, but the DSP doesn’t have a magic wand that will make the best inventory available in RTB when the publisher doesn’t hand it over to an SSP or exchange.

    Just saying.

    • alejandro correa

      I understand that the inventory that publishers value most highly is the last inventory that goes out into exchanges. With that said, I am skeptical that pub’s perception of value is driven by data: I don’t think the pub has a lot of visibility into what’s actually working for the advertiser. So, when pubs assign value to inventory that’s “early in the session” or whatever, they’re guessing that it’s “better” but don’t actually know. This is supposition on my part…would love your comment on this. The only way I could see pubs getting data behind this is CTR, and I don’t think that’s a great indicator of performance.

      Personally, I think that the quality of the content on the page, and the production value of the website, is a much more reliable indicator of the impact the ad will have than something like “session depth” or “time on page.” In other words, the impression sold direct and the impression sold via RTB has the same actual value, even though the range of what’s actually paid is very different.