TRAQ Scores and You

Pay no attention to that man behind the Banner.

Advertising is huge.

It can change the minds of millions, causing their wallets to follow.

It drives Google to provide free high-speed internet wherever they can, and to deploy high-flying internet-broadcasting blimps to wherever they can’t, just to ensure more eyeballs are on the Internet, where Google controls a good chunk of all ads served.

Advertising is the reason many artists don’t starve, either by making “creatives” for the ad industry, or by displaying ads adjacent to or within their works to earn revenue.

For many artists, this ad revenue is the difference between creating as a career, and creating as a part-time, unprofitable second job.


Falling off TRAQ

I help my wife with an ongoing, full-time project called The Devil’s Panties, a PG-13 comic about her adventures in life. Advertising is one of the things I have to monitor and continually tweak to make sure we can keep the lights on. All told, advertising accounts for about 1/3rd of our income.

Leading up to the 2013 holiday season I noticed a sudden drop in ad revenue. Almost overnight, our ad fill rates went from about 60% to about 10%. This was VERY bad news.

Very, VERY bad news.

The holidays are the best time for ad revenue, and that income helps my wife and I get through the dry months. I had to get to the bottom of this. While investigating, I picked up on a little tidbit that would lead me down a rabbit hole.

I discovered something called a “TRAQ Score”. The TRAQ score is a product by Integral Ad Science (formerly known as AdSafe Media). It’s a combination of several metrics that tells potential advertisers how “fit” your site is for their brand. Think of it as sort of like a credit score for your website. At some unknown interval, advertisers check this score to determine which ads are ok to serve on your site, or if it’s ok to serve ads at all. As near as I could decipher them, the metrics are:

  • ADT: Adult Content
  • ALC: Alcohol
  • DLM: Downloadable Material
  • DRG: Drugs
  • HAT: Hate Speech
  • OFF: Offensive Language
  • SAM: ???

These individual scores range from 1-1000, the higher the “safer”. What happened to us was that our ADT(Adult) content score dropped from 1000 to 450. For comparison, 450 is the same ADT score that xhamster.com has, which is a porn streaming website.

I couldn’t pinpoint what exactly changed on our site, so I tried to “clean” it up as much as I could. After self-censoring the site I waited, and waited, and still nothing. Our adult content score was still 450, and ad revenue was still crap.

I reached out to our ad provider and they offered to “whitelist” our domain, which I took to mean that they wouldn’t query our TRAQ score before deciding what ads are worthy of showing.

The holiday season was in full swing. Before they white listed us, we were getting about 0.30-0.45 CPM, with up to a 28% fill rate.

The day after they white listed us, we got up to $1.00 CPM with up to a 50% fill rate.

Our Holiday Journey

This was great news, and my wife and I both breathed a sigh of relief.

Unfortunately that relief wouldn’t last long.

Off the Cliff, Again.

Our ad revenue dipped again after the holiday season, much harder than the usual post-holiday drop. I reached out to our ad provider again, and it looked like whatever special arrangement was made to whitelist us was now over. They mentioned that the adult content score was too low, and that there wasn’t anything they could do. They reached out to Integral Ads(the creators of TRAQ) on our behalf, but it was entirely out of their hands.


Getting back on TRAQ

I decided to reach out to Integral Ads directly, to see if our score could be appealed. A few tweets, emails, and phone calls later, all I can gather is that our site is scoring low because the word “fuck” appears too frequently on the site. I asked for a re-score, and pointed out that pornographic websites score the same as our PG-13 comic, but a Business Development Associate at Integral Ads said the score is unlikely to improve as long as the language remains.

In other words, if I don’t completely censor the site, we’ll be blacklisted as if we were a porn site, and we will continue to see pitiful ad revenue. Were we any smaller, this could’ve spelled the end of The Devil’s Panties, and were we just starting, we might’ve ended up just another failed creative project.

Looking at their website though, I could infer some very bitter hope:

“Stay tuned for our exclusive app that will allow you to find your own site’s TRAQ Score — coming soon!”

I’m sure they’ll be offering TRAQ score consulting services shortly thereafter, per this quote:

“Break away from the pack; be a media quality TRAQ star. Contact us today!”

Or maybe you might need some “protection” for your TRAQ score…

That might sound cynical, but I again draw parallels to the credit industry.

Imagine that you kept being denied for credit, that you couldn’t buy a car or a house. Imagine if you couldn’t even rent an apartment despite having the means to pay for it, and imagine if you had no idea WHY. This was the reality before the Fair Credit Reporting Act was passed in 1970. The Act gave consumers the right to request their credit reports, and also gave them the right to petition errors in their credit histories. While it was a victory for consumers it was a win/win for the credit reporting bureaus as well, since they charged creditors for access to their data to help them determine who deserves credit, and they also charged consumers for the privilege of knowing why they were getting denied. It wasn’t until 2003 that this information was legally required to be freely available.

Imagine if none of those credit protections were in place, imagine that you couldn’t appeal or even KNOW your own credit score, and now you have an idea of the current situation with TRAQ scores. Your TRAQ score isn’t transparent, you can’t appeal it, it is directly affecting the ability of thousands of content creators to earn a living, and worst of all: none of those creators know that TRAQ scores even exist.

I do empathize with the desire of companies to not have their brands associated with some types of content, but certainly someone who watches “ADT” might still want a iPhone, people who do “DRG” probably still want Hot Pockets, and those who drink “ALC” might also want to go to Disney World. For their own sake, maybe it’s time the advertising industry matures past the point of “brand safety”, and moves towards targeting potential customers wherever they may tread with contextually appropriate material.

Brand safety aside, maybe there is a need for a tool to determine the content of a website to help target ads better on “racier” sites, but is TRAQ accurate enough to be that tool?


Finding your own TRAQ

Integral Ads at one point hinted that they’d eventually make TRAQ scores publicly available, however this mention has disappeared from their website. In the meantime, in the interest of transparency, I wish to share with you the public TRAQ Score API:

http://api.adsafeprotected.com/db/client/25124/absit?adsafe_url=YourDomainHere.com

Usage is pretty straight forward. You swap your domain in for YourDomainHere.com, and it returns a JSON response detailing the components of your overall TRAQ score:

{
“action”=>”failed”,
“bsc”=>
{
“adt”=>1000, // Adult Content
“alc”=>1000, // Alcohol
“dlm”=>1000, // Downloadable Material
“drg”=>1000, // Drugs
“hat”=>1000, // Hate Speech
“off”=>1000, // Offensive Language
“sam”=>1000 // ???
},
“iab1"=>[“iab_style”], //IAB category? (arts, style, hobbies)
“iab2"=>[],
“ttl”=>”2014-04-22T14:46-0400" // Time To Live (expiry date)
}

Here is what is currently returned for TheDevilsPanties.com:

{“action”:”failed”,”bsc”:{“adt”:450,”alc”:1000,”dlm”:1000,”drg”:1000,”hat”:1000,”off”:1000,”sam”:1000},”iab1":[“iab_style”],”iab2":[],”ttl”:”2014-04-22T14:46-0400"}

Here are the results for my oft-mentioned NSFW example xhamster.com:

{“action”:”failed”,”bsc”:{“adt”:450,”alc”:1000,”dlm”:1000,”drg”:1000,”hat”:1000,”off”:600,”sam”:1000},”iab1":[],”iab2":[],”ttl”:”2014-04-22T18:51-0400"}

The Not Safe For LIFE website rotten.com is clearly as safe as my wife’s comic:

{“action”:”failed”,”bsc”:{“adt”:450,”alc”:1000,”dlm”:1000,”drg”:1000,”hat”:1000,”off”:1000},”iab1":[],”iab2":[],”ttl”:”2014-04-22T19:47-0400"}

Maybe you should bypass your local dealer and hit up crayola.com for your next fix:

{“action”:”passed”,”bsc”:{“adt”:1000,”alc”:1000,”dlm”:1000,”drg”:875,”hat”:1000,”off”:1000,”sam”:1000},”iab1":[“iab_arts”],”iab2":[],”ttl”:”2014-04-23T16:10-0400"}

TRAQ also assures us that the whole of youtube.com is kosher:

{"action":"passed","bsc":{"adt":1000,"alc":1000,"dlm":1000,"drg":1000,"hat":1000,"off":1000,"sam":1000},"iab1":["iab_arts"],"iab2":[],"ttl":"2014-04-23T16:16-0400"}

The values returned are all over the map and sometimes have little bearing on the actual content(or the severity of the content) of the queried web site. To be denied ad revenue because of your site’s content is bad enough, but it seems you can be denied revenue due to a clerical mistake, an algorithmic error, or a difference of opinion, with no effective way to appeal.

As I mentioned early in this article, I don’t know what all the values returned mean, but the most chilling part of the API is the number “23124" in the API’s URL, which looks to be a client ID. I’ve been swapping random numbers into that part of the URL and based on my rate of success there are hundreds, if not THOUSANDS of advertising companies using TRAQ scores to determine who is worthy of hosting ads. If you have a website, you are probably being monitored by Integral Ads.

Plug in your domain to the TRAQ API and find out for yourself how you have been silently measured, and if you’ve been secretly blacklisted.

Find out if YOU are on TRAQ.

-obby


Addendum: Just as this article was about to be published, our adult content score was increased from 450 to 500. It appears that Integral Ads is open to adjusting scores based on input, however I still disagree with the notion that The Devil’s Panties is 50 points less “adult” than a site featuring ass to mouth videos.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.