The final digital ad environment problem I want to discuss is perhaps the most shocking. We took a look at our most recent ratings for the sites we used for our original human scoring process almost a year ago. One in five of those sites now returns a rating of insufficient data.

That may not mean much to you, but to us insufficient data means the site has errors, redirects, has almost no content or simply doesn’t exist, with the last option being the most common. After looking at the sites in more detail, we quickly learned that over 20% of them had, in fact, disappeared, turned into ad farms or been stripped of all content.

This was the first time we had ever seen such data on site turnover. Based on these numbers, 5% of sites bought as part of a quarterly plan could no longer be in existence by the end of that period. We have also seen some instances where black- and white-lists have become completely outdated and useless due to this rapid turnover.

This disappearing act may sound better than some of the alternative issues that could damage a media plan, but it is just one example of the larger problem. Online inventory is dynamic and constantly evolving. The process of acquiring quality media is tough and requires heavy maintenance. Having an automated process to vet inventory is simply a necessity.

Here are some examples of the types of sites we see classified as insufficient data.

Once a site is taken down or neglected, domains are often bought up by people looking to capitalize by selling the domain back to the original owner or by serving ads to leftover and incidental traffic. These “ad farms” or “parked domains” are nothing but sponsored links and/or information on buying/renting the domain. No original content at all.

Other sites become completely inaccessible. They could have errors, moved without providing a redirect or were simply taken down. Whatever the case, there is no longer anything to be seen. Not even ads.

And a few more examples for those of you who can stomach it.

For over a year now Trust Metrics has done its best to bring quality publishing environments to online advertisers. Over that time we have tirelessly tried to define what we mean when we talk about quality. In fact, we pretty much never shut up about it.

Today, instead of just talking about it, I want to share a few examples of the type of low quality sites we find all too often in scale inventories. These sites are what you might expect when stepping outside of a typical brand’s comfort zone (0-40 on TM scale). The lowest of them contain safety issues, but the majority are just poorly managed amateur content designed strictly to collect a few ad dollars. They contain very little content, have infrequent (if any) updates and extremely poor and outdated implementation.

We were shocked to find that OVER HALF of the sites in our study fell in the limited or poor rating categories (blacklist). Here’s a taste of the low quality sites we saw in our study. This is just a sample of what we have seen throughout the past year. Once again I ask: do you see any value in these impressions?

I am no longer a part of the human scoring process, but I continue to work with the computer-generated rating reports my scoring helped create. Now I get to view websites with the benefit of already having model outputs in hand. Working on the back end has given me an even better understanding of our product and what to look for when evaluating the Internet. Perhaps more importantly, it gives me access to sortable data and end of year totals.

It has been a busy January reporting some of the things we learned in our first year of ratings. We have been featured in MediaPost’s Metrics Insider and DigiDay. However, I want to take a little more time explaining the data we collected, because I think this data reflects where our industry stands and where it is headed.

This is a the first post of a three part series detailing and providing examples of the state of the online inventory. The study was based on two samples of 50k+ sites from our 500k+ database. The database is comprised of site lists we have collected from various sources throughout the year and sites we have added for scoring and other purposes.

Safety and Non-English

The good news: Safety, which up to this point has been the driving force behind a large portion of ad tech, is less of a problem than I imagined (having seen the things I’ve seen). Only about 9% of sites had safety issues surrounding profanity, pornography, hate speech or violence. Even better, there are countless technologies designed to avoid safety concerns that have made avoidance of this content common, simple and automatic. In other words, what little unsafe content continues to exist on the in ad inventories easily avoidable through the use well-known technology.

The bad: My focus today is on the one quality issue that outweighs all safety issues combined. Non-English sites made up just under 10% of the sites we looked at – more than the four most prominent and concerning safety issues combined. That includes German, French, Spanish, Serbian, Hindi, and Chinese speaking sites, to name a few.

Non-English sites certainly aren’t as toxic as the content flagged for safety reasons. They probably won’t damage a brand, but it’s hard to think of a common scenario where an English ad on a Hindi site would result in a positive action. It’s just pure waste.

This is one of the many issues that underlines the often ignored issue of quality – not safety. Safety is the loud, obnoxious brand killer that is impossible to miss. Quality is death by a thousand (million) missed impressions, because every impression on a non-English site is a missed opportunity to be in an effective environment.

Take a look for yourself. How much value do you see in these impressions?

New York, NY – January 13, 2012 – Trust Metrics (www.trustmetrics.com) – the industry’s leading provider of digital measurement standards for publisher transparency, effectiveness and quality – today announced the release of a new survey of hundreds of thousands of websites that shows the amount of non-English sites outpaces the number of sites labeled as “unsafe” – containing either pornography, or violence, or profanity.

According to Trust Metrics, the result of the analysis proves that a fundamental transformation needs to take place across the ad industry: proactively finding quality, optimal ad environments before the ‘buy’ is a more cogent and relevant strategy for media planners than simply reactively avoiding sites considered ‘bad’ or ‘unsafe’.

Between December 12-16, 2011, Trust Metrics conducted a comprehensive analysis of websites in its database, as well as the database of an independent ad exchange inventory.

The study revealed that non-English sites – a quality issue for buyers – amounted to approximately 10 percent of all sites surveyed.

Sites collectively labeled as “unsafe” didn’t even break double-digits in the survey: websites characterized by profanity amounted to 3 percent of sites surveyed, while hate speech registered at 2.5 percent, violence at 2.5 percent, and pornographic sites at only 1 percent.

Trust Metrics used the Interactive Advertising Bureau’s (IAB) Quality Assurance Guidelines definitions of safety to categorize the domains analyzed. The Trust Metrics study also measured sites that directed users to other domains or delivered dead-links.

“Issues related to safety in our industry tend to siphon away a lot of the oxygen surrounding concern and innovation,” said Andrew Lerner, CEO of Trust Metrics. “Our new survey debunks this urban ad legend by demonstrating that it’s quality issues, such as the huge amount of non-English sites cluttering the inventory landscape, that still should matter the most. In 2012, safety alone should no longer be the guiding benchmark of any successful ad campaign for any advertiser. Principles such as ‘safety’ or ‘context’ are merely subsets of a much larger, over- arching challenge which must be embraced by our industry — the pressing need to identify and pursue those quality ad environments that deliver optimal value to brands.”

In addition to categorizing the results of its survey into issue-areas, Trust Metrics deployed its proprietary advertising scoring technology – using more than 16,000 data points – to assess and rate the overall year-end quality of domains in the databases.

When Trust Metrics applied its scoring labels to the sites that were analyzed, the biggest category was “Limited” with just under 50 percent of sites falling into that category. The number of domains surveyed that were rated “Limited” or “Poor” jumped to about 58 percent. Websites that are scored “Limited” or “Poor”, according to Trust Metrics, generally do not attain minimum quality and/or safety characteristics, or typically fail multiple quality and safety measurements.

“Our analysis was a huge eye-opener for anyone involved in digital media campaigns,” Lerner added. “The fact that a vast majority of sites – almost six in ten – were rated as low to poor quality means our industry has a huge challenge on our hands, and it’s likely worse than any media campaign planner could have imagined a year ago at this time. The Trust Metrics analysis is a wake-up call for all brands interested in maximizing the value and ROI of ad campaigns. Our study shows that the right quality ad environments exist, but they’re harder to find and require the right tools and the right approach. For 2012, the existing models must simply be blown apart, so that the mission is no longer blindly dominated by an avoidance of the negative, but finding and optimizing positive quality domains that deliver the best ROI.”

About Trust Metrics Ratings

Trust Metrics created ratings to reflect the overall quality or value of a site or webpage in a single number advertisers and advertising aggregators (such as networks) can act on quickly and effectively. Trust Metrics ratings enable advertisers to identify the best places for their ads to engage their target audience while enabling publishers, networks, and platforms to get fair value for creating high quality content. The Standard Quality Rating is reported on a scale of 0-100, with 100 being highest quality.

About Trust Metrics

Trust Metrics (www.trustmetrics.com) provides brands and their agencies with the capability to identify and score high value advertising environments at scale. By deploying advanced algorithmic techniques and machine-based learning to analyze sites for their value as advertising environments, Trust Metrics dramatically improves the buying and selling of media by making it possible to vet billions of impressions in real time and in advance of a media buy. Trust Metrics is based in New York City.

Media Contact,

Nicholas Graham,
ngraham@trustmetrics.com

And still questions linger. The latest accusation, via Trust Metrics: thanks to exchanges, U.S. advertisers are showing up on tons of non-English language websites. Trust Metrics pegs non-English sites making up about 10 percent of exchanges. In fact, non-English sites in exchanges now outnumber those deemed “unsafe” for reasons like carrying profanity or risque content. One often-feared ad exchange issue that isn’t much of one, according to Trust Metrics: porn, which only occupied 1 percent of sites.

“The brands aren’t participating in a real way yet because they’re afraid,” said Trust Metrics CEO Andrew Lerner. “They’re right to be afraid. There’s good apples and bad apples out there.”

It’s best to take these kind of stats with some perspective. Trust Metrics is in the business of guarding against quality issues for advertisers buying via exchanges. It would therefore be in its interest to play up the issue, much like click-fraud vendors were regularly sounding the alarm that huge chunks of paid search clicks were worthless.

Lerner allows it is in his interest to play up the problem his company is out trying to solve. Yet he stands by what the data his company sees shows. The solution isn’t to avoid exchanges, he said, but for buyers to take more safeguards.

The problem points to a paradox of the programmatic buying world. On the one hand it is about finding pinpoint audiences. Yet on the other it is about scale. Sometimes finding those narrow audiences at scale means going very far and wide. It’s no mistake, Lerner noted, that Forrester ranks demand-side platforms on the number of impressions they can ingest. With that many impressions in play, many will turn out on sites that are worthless because they’re stuffed with ads, not in English or have other quality problems, he said.

Read more: http://www.digiday.com/data/are-non-english-sites-an-exchange-problem/

This year we’ve looked at more than 500,000 unique domains — some of them hundreds of times, over and over. This is called taking inventory of the inventory. We’ve discovered some eye-openers about the state of the inventory as 2011 ends, and 2012 begins. Here are our more important and relevant findings. Call them “2011 Tail-end Takeaways.”

First, there’s a lot more to be concerned about than safety. About 10 percent of all domains that sell ad impressions in scale buying environments (networks, SSP’s, and exchanges) are actually non-English language sites. In fact, more domains are non-English than the number of domains identified as pornographic, or containing violence hate speech, or significant amounts of profanity, combined. That should be a game-changing finding for media buyers and planners as we head into 2012.

What that means, in part, is that the ad verification posse in our industry must extract itself from what appears to be a rather outdated and overhyped fixation on a single issue silo, and instead expand its horizons to other quality issues equally toxic to media buyers and sellers.

Second, this isn’t your father’s media environment. The ad inventory supply at the end of 2011 is marked by constant, dynamic change and evolution. About 21% of the sites that we looked at a year ago in late 2010 actually no longer exist. That means tens of thousands of sites that may have been placed on whitelists at this time last year by either agencies or networks and supply-side providers are gone. There is no content at those domains, or they are ad farms. Never before have so many marketers bought so much inventory in environments so poor and degraded that they may not be around in six months — or less. Another warning for 2012.

Third, there’s a serious quality issue out there, folks. That’s a problem for both brands and direct marketers. According to our ratings, about six out of 10 sites (58%) available in RTB and large networks are sub-standard environments for advertisers because they don’t adhere to minimum publishing or editorial principles. Think about that one for a second. If there are a million or so domains that sell ads, most brands won’t want to be on 600,000 of them. That’s an ad-ops “Eureka” moment. If marketers wouldn’t want to buy inventory on “phantom” sites, they should be equally concerned about sites whose overall quality matures into negative — even harmful — quality territory.

Call these numbers what you will; a slap in the face to the industry, or a wake-up call. But I don’t think they should be seen as all bad. First of all, the inverse of the numbers about bad sites is true: there are tens of thousands of sites out there most advertisers would want to be on, as well as a couple hundred thousand sites that are average — not so great, not so bad, but good for reach. Second, it turns out that “good” sites drive performance: click-throughs, CPA, and brand metrics.

That’s good news for buyers and sellers willing to look into publishing environments and incorporate “media intelligence.” Yes, that would be yet another step in a process that’s already complicated.

Yet it’s also a clear opportunity for improved performance. Understanding where each impression is served and applying that knowledge will be one of the biggest drivers of highly successful campaigns in 2012.

That depends if you’re willing to take the time to dig in and see what really happened over 2011. Take inventory of your inventory.

Read more: http://www.mediapost.com/publications/article/165304/2011-ad-inventory-takeaways.html

This year kicked off with a lot of noise regarding ranking quality of online content, as Google faced a barrage of media criticism over its search results becoming overpopulated with SEO-gamed links from so-called “content farms.” The bigger question behind the brouhaha was simply: “Can you quantitate quality?” Judging by its ongoing Panda algorithm update, it seems Google thinks the answer is yes (and arguably it’s backed up that answer).

But Big G isn’t the only one of that opinion. Technology companies such as Trust Metrics, which was recently tapped to apply its digital measurement standards to targeting platform [x+1]’s online ad inventory, have developed intense formulas for ranking of website content in terms of inventory transparency and, yes, quality. In effect, Trust Metrics aims to broach the next level of verification – serving as a content matchmaker between online advertisers and publishers.

CEO Andrew Lerner was kind enough to take some time to explain how Trust Metrics calculates its quality rating, including ad-related mesurements, and how publishers can take advantage of it.

Clue us in to what’s so unique about Trust Metrics quality score — can you detail how Trust Metrics calculates quality, or is it so complex that it will blow our little minds?

It’s all mind-blowing to me, so read with caution…

Trust Metric’s Quality Rating captures the overall media value of a web site or web page in a single score (or rating as we like to call it). It’s a quick overview that’s accurate, actionable, and drives performance – we’ve seen CPA drop by three times and click-through rates double across many campaigns.

We don’t look at users. We don’t use cookies. We don’t look at the advertising creative (yet). We simply look at the content and how it’s presented. One mind-blowing thing is how important the media itself is for both brands and direct marketers, when most of what the industry talks about is “buying audience.”

To calculate a rating, Trust Metrics collects data that no one else has (as far as we know) – about 16,000 unique data features from a single site with our crawlers. That’s up from the 75 or so we started out with. And it’s not just a big group of obvious features relating to context or safety concerns alone (we tried that and it didn’t work very well). There are just as many quality concerns online as there are unsafe environments – ad clutter, page level user generated content, and ad farms are good examples of the kinds of issues we look at and report out in.

Then we do the math – applying the statistics – that lets us use all of that data to generate our ratings. That is kind of mind-blowing, actually.

So while the process is very complex, all we’re really doing is the same thing media buyers have been doing from the first days of mass media. And we believe that the ad industry is in sore need in many ways of getting “back to basics.” In the end, the output is a simple, interpretable and actionable 0-100 score.

Why do you consider this score better than other quality rankings? It’s OK to gloat if you back it up…

We collect more data on more important features than the competition – 16,000 to be exact – but we don’t stop there. What other companies call quality rankings are still just safety ratings by another name or raw feature counts like the number of ads on a page (not a consistently relevant number). Trust Metrics is the only ad innovator to truly incorporate the quality of website construction and content in our ratings. When we say “quality” ratings we mean it. And that’s including, but absolutely not limited to, safety and content.

What are some of the ad-related measurements included in your overall quality ratings? Why do you consider these essential? Which ones should publishers be paying more attention to?

There is a recent obsession with above/below the fold when it comes to ad placement, and we think it hamstrings the industry, not to mention the premium brands that invest hard-fought for dollars in it. There are certainly differences in performance that make those numbers important, but both locations have their benefits. The Trust Metrics standard rating considers not just location related to the fold and number of ads, but location of ads in relation to consumable content and ad size. Sites that have no ad space also receive a drastic reduction in rating due to their evaluation as poor (or non-existent) advertising environments.

We also recently released a specific Ad Clutter rating – separate from our Quality Rating. That’s basically an old-fashioned analysis of ad to edit ratio along with positioning on a page or across a site.

Do you consider yourself an ad verification company? If not, how do you differ?

Trust Metrics is about delivering any relevant intelligence about the media itself. That information does two things: it drives performance and it gives a planner what they need – whether that’s a look into the quality of the media or a specific issue like “is this site an ad farm.” That’s not verification companies do.

Where we are like those companies is that we also look at safety. But unlike them, we include that information at no additional charge as part of our standard product.

We think of Trust Metrics as the next evolutionary step up from verification as advertisers attempt to better understand the value of digital media. Verification is certainly necessary, but it’s no longer sufficient. While they address safety issues (necessary), the verification products don’t proactively find the right place for advertisers message because they simply don’t track the data we track and don’t report on the issues we report on.

How do you help publishers get “fair value” for producing high-quality content?

Our ratings are useful for both sides of the digital ad landscape. We ensure that advertisers don’t over-spend on low-quality content, and in turn, Trust Metrics ensures that publishers – particularly in scale environments – are properly rewarded for their efforts to provide worthwhile content. This usually means that available impressions are sold through at much higher rates because our clients on the buy side chose to buy on high-quality environments and our partners on the sell side want those sites sold first because they know performance will improve. And finally, networks that take the time to compile high-quality inventory can demand a premium.

Can you explain how pubs can use quality classifications? How do you ensure transparency?

Trust Metrics has done some work with publishers, but to date most of our focus has involved networks. We’re excited about our plans to launch a publisher report product next year that will offer publishers the ability to both see their ratings receive information on the areas in which they can improve their sites.

How do networks use your ratings to vet incoming applicants to networks or exchanges? Is this customizable?

How they do it is really pretty simple. Scale sellers like exchanges or networks use our ratings as a benchmark to compare the media value of sites and vet for other issues like safety and non-English sites. If, say, a site applies to be in a network that works with us, then that network will ask us for a rating: via API, our web interface or managed services. We’ll send back anywhere from 5 to 15 distinct ratings for that site and a network will decide if they want to let that site into the network at all or how to value a site. Networks also use our ratings to manage and track existing inventory.

And yes, definitely customizable – pretty much everything we do is. Some sites might be great for a certain segment, say fashion, but not so much overall. For example, there are plenty of luxury-oriented blogs that we might give a more moderate or limited Standard Quality Rating. (That means they are probably only “ok” at best for many kinds of advertisers.) But those sites would be perfect for a fashion advertiser. So Trust Metrics created unique custom ratings to figure that out.

Trust Metrics identifies media that helps build brand identities — how can publishers use this to their advantage?

There are several ways this information can be useful to publishers. First, sales teams have a better understanding of what is important to their potential clients. Knowing the type of environment a brand finds desirable and has found success in can help drive future sales that are beneficial to everyone. Also, knowing the type of brands to seek out can save time wasted on brands that may not find that publishing environment beneficial to them. Finally, inventory can become more valuable as brands recognize a particular publisher as a great source of impressions for their brand.

Problem

• Six Flags was planning a push to increase park attendance with an online marketing campaign. Their agency, Universal McCann, wanted to maximize reach and ROI in order to best use utilize their resources while maintaining the Six Flags family-friendly image.
• A large list of sites from a popular ad network was acquired and in need of automated analysis. Safety was an obvious concern, but efficiency is what drew them to Trust Metrics.

Solution

• Universal McCann supplied Trust Metrics with a list of recommended sites from the ad network’s inventory.
• Trust Metrics returned customized quality and safety ratings that provided insight and transparency for the sites on from the list. Thresholds were set and unwanted sites were eliminated for both safety and quality concerns.

Results

• The campaign was a huge success and online ticket purchases tripled in the following months.
• Costs per impression and acquisition were down and click-through rates were up thanks to the elimination of wasteful spending on low quality sites.

Details

In our first year of quality ratings we have been surprised by how much online inventory is – at best – worthless to advertisers. Even worse is the amount of environments that can be damaging to a brand’s image. We will be releasing some year-end totals in the coming weeks that might really scare some brand advertisers out there.

I recently came across an article detailing the importance of brand equity. Brand equity has never been easy to measure, but the research referenced in the article estimates that brand loyalists account for 50% to 70% of brand sales, with advertising and promotion impacting the other 30% to 50%.

With at least half of sales resulting from brand image alone, protecting that brand is as important as ever. Conversely, chasing the relatively small group of “swing voters” all across the seedy underbelly of the Internet seems like a poor investment.

A few temporary sales could be more than cancelled out by alienating a single loyalist who makes repeat purchases without benefit of sales and promotions. Don’t get caught targeting a potential customer in a bad environment. Those impressions will be – at best – worthless, and have the potential of damaging brand equity for the future.

Check out all the details on the importance of brand equity here.