Google AdSense "Policy Violations" Reports Appear Inaccurate

[csdude55]

I regularly get "policy violation" notices from Google about something someone has posted on my site that I think are a REAL stretch.

For example, in my classifieds section I have a category for "apparel". Someone downloaded a picture from Kohls.com of a mannequin wearing the one-piece swimsuit she was selling, and it was flagged for "sexual content". A picture of a mannequin. From a clothing big box retailer. Wearing a one-piece. Sexual. OK.

Then today, what Adsense banner do I see on my site? Something from a lingerie website, with 10 models in VERY skimpy lingerie... zoomed in on the boobs, then it zooms out to zoom in on another one.

So the mannequin in a one piece is bad, but full boobs on display is OK?

Later, I see another ad for men's underwear, showing models with HUGE bulges to display their underwear.

But again, a mannequin in a one-piece? Oh god, no!

Business is tough enough, Google, I really don't need you breathing down my neck every day on content that would be G-rated anywhere else. Your playground your rules, I get it. But still.

[tangor]

@csdude55...

Follow the money. :)

[pubpolicycomms]

Rather than a "policy violation", Sexual content is a 'restriction', which means advertiser demand will likely be less for this type of content: [support.google.com...] In this case, you were simply given a heads up that you will likely receive less monetization for that category of content, as advertisers have shown less willingness to appear alongside that content.

As for ads appearing on your site, you have a number of controls available to you to block categories, sites, or products. If the content of a particular ad is not suitable for your site, I suggest you utilize these controls: [support.google.com...]

Additionally, if an ad is particularly offensive, or unsuitable, you can always report the ad, and a team will look at it and determine if there are policy violations for that ad.

Hope this is helpful.

[NickMNS]

I totally get your frustration, how is that the AI can flag the fake mannequin but it unable to flag the fakes in the ad.

[NickMNS]

@pubpolicycomms

In all seriousness,

Additionally, if an ad is particularly offensive, or unsuitable, you can always report the ad

While I don't doubt this, as a publishers we have a lot of work on our plates, creating new content, maintaining websites, etc. I used to spend time each day using the ad-review center to check my ads, blocking all the spam. But this exercise was taking hours, and was utterly useless, as blocked ads would re-appear the next day with only minor modifications. Supposed features in ad review, such as find similar ads were useless, they couldn't even pair identical ads. Google has endless resource to create AI powered bots to police our content but it seems as though they spend very little effort and resources policing their own ads.

Your take away from this thread should be that Google needs to do more, not tell us publishers that we are not doing enough.

[csdude55]

The issue for me, @pubpolicycomms, is that this was a regular occurrence before the policy change in... September, I think? So at least a few times a month I was being warned that my site was in danger of being suspended by Google due to violations that weren't really violations.

Of course, I would remove the offending picture or ad because Google pays the bills. But then I would lose site users because it looked like I was the one being obtuse, and they could easily post it on Facebook without any problem.

Now, there's no real understanding of what Google means when they send out a policy violation. Yes, it says that ads will just be restricted and it's OK, but does that affect the value of ads on the rest of my site? How many of these restricted-ad-pages can I have before Google suspends my site? It's all very vague, so we have no choice but to continue removing pages with violations and just hope that it's OK.

[csdude55]

Your take away from this thread should be that Google needs to do more, not tell us publishers that we are not doing enough.

Amen, brother.

[pubpolicycomms]

"Google needs to do more". Okay, tell me what else we can do to help. Constructive ideas and feedback are welcome.

To recap what we've done recently: We took what was once several policies, and changed them to "restrictions" to benefit you. Restrictions are designed to be an improvement to your experience, through the alignment of various product policies. Google is NOT limiting your monetization on these pages. Advertisers are choosing which content to bid on, and we are simply letting you know that these pages, with content that we've listed in the Help Center link I provided are under limited bidding. Additionally, since these are not policy violations, you don't need to make any changes to your pages, and you are not under any threshold for further action.

We've also provided more controls to publishers and advertisers to put control into your hands. We've more than doubled the number of categories that advertisers, and publishers can use to restrict types of ads, or content that is acceptable to those that participate in our network.

[engine]

Thanks pubpolicycomms, we appreciate you being there to answer the questions.

If any one has specifics, let's hear them.

My suggestion would be for Google to consider paying publishers that aren't going to reach the threshold. Do it as a gesture in these tough times.

[pubpolicycomms]

I'll submit that feedback. Not my team, but it's good feedback in general. Thank you.

[csdude55]

@pubpolicycomms,

My main request would be clarification in the official documentation.

One thing that I noticed was that after the policy change, I had a few policy violation reports that appeared to be optional. I took the ads off of the page, but also saw that my RPM went down by about 30%. Coincidence? Maybe. But it sure felt like my whole site was being punished... maybe listed as a site with sexual content, as opposed to a single page with presumed sexual content (eg, a mannequin).

A clear explanation in the documentation would be legally binding, so I, at least, would definitely appreciate knowing exactly what is and is not going to happen.

[pubpolicycomms]

Based on your example, this would be a 'restriction', not a policy 'violation', and therefore the lower advertiser demand would account for that RPM issue. With restrictions we do place this text in the description on the Help Center [support.google.com...] "If your content is labeled with an inventory restriction, fewer advertising sources will be eligible to bid on it. In some cases this will mean that no advertising sources are bidding on your inventory and no ads will appear on your content."

It does seem weird that would would have a reduction across all of your site, for only one page of publisher restrictions, unless there's more to this. You wouldn't be as you describe "punished" for one page of content that isn't violating policy.

Your point is well taken, and we are always working on trying to make our documentation and notifications more publisher friendly, as it only benefits us to have good publishers working with us.

Thank you for the feedback.

[NickMNS]

I appreciate the changes to policy violation, although thankfully it has never impacted me directly.

We've also provided more controls to publishers and advertisers to put control into your hands. We've more than doubled the number of categories that advertisers, and publishers can use to restrict types of ads, or content that is acceptable to those that participate in our network.

Fine, you've added more categories, but blocking of categories is poorly enforced. My add review is full of ads for dating sites, and tarrot card reading despite the fact that those categories are blocked. Moreover, the optimization suggestions I get I nearly every month are constantly suggesting that I unblock the categories that are blocked because of the high levels of spam.

What substantial controls have been added?

Here is a concrete example, a sampling of some of the many spammy ads that cannot be blocked.
[imgur.com...]

At first glance these ads appear identical, despite this when one clicks on "Find Related Ads" the system is unable to match them up. The implication is that one needs to click through 1000s of ads in order to select and block each one individually. You might say, "they are all from the same domain, just block the domain." Yes great point! but this isn't always the case, one often see similar ads from different domains. In fact if you look carefully you will notice that url's vary ever so slightly, no doubt to fool what ever controls are in place. Back to blocking the domain, to block the domain one needs to leave the "ad review center" section and go to the "all-sites" section then enter the domain and then return to the "ad review center". Leaving and returning to ad-review takes time, 3 ads on page take long to load 18 ads takes an eternity.

Now here is the kicker...
I explained above that sometime there are many similar ads from one domain but other times the similar ads are from different domains, and this is a real problem for a human reviewer. If one comes across one ad, there is no way to know if the domain will repeat or not. So now is it worth leaving ad review to report the domain? Probably not. So let's block the ad and keep going, a few pages down, here it the ad again. Now is it the same domain or not, "qgrip" vs "qgrips", I don't know I've just seen 50 ads since the last time I saw that one. "Just click back", NO! back doesn't take you back in ad-review it takes you back to the home page. No no no! click back arrow! Nope, that doesn't work either there is nothing to go back to since the ads were marked as reviewed. Filter for blocked ads, yes finally! Now we load 18 or so more ads that we already saw and hoped we'd never see again, we've now noted the domain and now head back to the un-reviewed ads. 18 more ads to load, but now the order has changed. "Where is the ad?" and on and on. It is a nightmare.

I am done, I have stopped reviewing ads because it is an exercise in futility, that will make a person mad.

I have been doing this since 2013 and AdSense has changed a lot, the shapes of the button have gone from square to round and the menus have moved from the top to the side.

Over the years, I have taken part in focus groups, sent countless support tickets regarding blocking issues, reported tens of thousands of ads featuring, nudity, hate speech, fraudulent claims and on and on. But still the basic functionalities that would allow publishers to truly and meaningfully block ads have never been added or changed.

In fact the most effective tool at reducing spam that was added was the ad-balancer, and I seriously doubt that the inventor of that feature ever thought that it would be used for that.

How to fix the problem:
Fix the "find related ads" feature such that it actually finds similar ads not just identical ads. Similar based on a wide set of features, including domain name, text content, image, account.

Why fix the problem:
These are not normal times, showing my users ads about removing earwax may not be desirable but it is not as bad as showing ads for fake Covid 19 test kit ads and fake N95 mask ads. Here is a sample.
[imgur.com...]

We as publishers can only do so much, Google needs to do more.

[NickMNS]

A soon as I finished writing the post above I went back to the ad-review and blocked one of the ear-wax ads and it automatically blocked many of the ads (50 or so). Which is new, so thank you this is a great improvement, but it only captured ads that were very similar, that used the same image. Several of the other ads in my imgur link were not flagged. So it is much better but in my view still needs work. But why doesn't "Find Related ads work"?

[CommandDork]

It's certainly not a perfect system by any means.

I regularly get flagged for "sexual content" on a page that details flying boats of World War 2.

Yep. Flying boats. WW2.

I could understand submarines or some other kind of innuendo but not flying boats.