Ad Churning and Irrelevant Site Wake Up Call

I have nothing but respect for Shawn and the senior members of this co-op. That said, this network needs to adapt to the ad-churning and unrelated site penalties that are underway. I understand Shawn is working on categorization - great. But the churning is obviously a huge factor in Goo's recent patent disclosure.

I know that this is an advertising network. If so, doesn't it make sense to mimic real advertising? To do so, all you need to do is think - "If I were paying for this ad, what would I want?"

If your site sells swing sets in Iowa, would you pay for an ad on a digital camera site written in Yugoslavian? No.

Would you want your ads to all be static, or all churn with every click? Probably a mix of both - static ads would give a type of branding and allow customers to have a second chance to come back to your ad, and churned ads would allow for a wider spread.

A few things that are relatively basic technically can only help everyone involved, and will help prevent penalization by engines:

- Allow members to choose the categories and languages for their inbound and outbound links.
- Allow at least some ads to be static for 10, 30 or 100+ days.
- Allow members to nix other site links if not wanted.

These are things you would do organically if you were purchasing ads. Purely as a bonus, Goo's current and near-term algo would love it. Instead, if things stay as they are, everyone who has inbound and outbound churned ads to completely unrelated sites will be penalized (read that as everyone in this co-op). The extent of the penalization would be to completely factor out the churned ads, or even put a negative slant to the sites involved.

 

I understand the reasoning behind wanting related ads and I give you that but you are basing your whole post on this "ad-churning".

Remember that it isn't a "patent disclosure" and it wasn't actually filed by Google. Its fine to argue if anything mentioned in the patent application will be reality but to say it is being used and then basing the rest of your post on that "fact" does not make a good post.

Link Vault, Webbys new service, provides semi static links but even he has shown no interest in related ads. At this point I think there is a better chance of them being implemented here at some point.

I have no issues with those proposing static links but it isn't because its to appear "natural" because the truth is you get more advertising for your buck by having them rotate. This way your weight is fully utilized. I don't want my weight getting used up being buried on a deep page with no visitors just so that it can be static.

There is more to this argument then just a knee jerk reaction to a patent application.

 

I agree. The ad-churning thing is a big problem, in my mind-- it has been since the beginning of the coop, but now I've read numerous confirmations that Google already limits the effectiveness of churned links(Shawn said this in another post, so he's fully aware).

You are correct about weight distribution, though-- I thought that the other day when I was thinking of a technical solution. I want static/semi-static links, but I don't want them on deep pages with PR0 or something when I'm giving PR0 sites links from my PR6 homepage or something... I guess it's probably technically impossible to do this in a fair way when you weight an entire site based on the PR of the main page(assuming all pages in the site: total are proportional, which is rarely the case)... Anyone-- any bright ideas as to what we could do technically? Shawn, could you weigh in on your thoughts when thinking about this initially and still since? Thanks.


--
Derek

 

YFS - Thanks for the well-phrased and calm response.

Stating my case as absolute fact was not my intention. Instead, I think it's most important for all of us to consider what is down the road. If so, most everyone would agree that now, or in 3 months, or 10 years, the successful engines will value natural, organic, related sites. Let's get there ahead of them. I mean, how many sites would you now have wanted to submit to Yahoo before they became fee-based? Make the inevitable changes now and reap the rewards for years.

Whether it's from the "disclosure" or not, the successful engines *will* penalize churned ads enough to negate any importance from the link, possibly worse.

And, you are right - there's a bigger spread from the rotated ads. That's why I suggest only making some of the ads static. And they should be static for various amounts of time, others should be rotated regularly.

Again, just think naturally - what would I want if I were paying for the ads? Do that and you'll succeed with the ad network and with the engines.

 

This will absolutely be the future and I do think the related part may be partially here although will be perfected over time. The original concern over having realted ads (which sounds great) is that the reality is the number of ads and sites in the system. This network is new and it was just a month or two ago when there was no way there was enough members to make that work. Especially in some of the categories. I am hoping this has changed as it has grown exponentialy. With referal ID's of 12908, it seems its around the corner.

As far as rotating some links and others static, since you brough it up, how would this actually be done and be assured to be fair? I'm not looking for code but instead trying to expand this thread to discuss the reality of it. It nice to say "Some statics, some rotating" but the reality is, how exactly would it work. You can't check the PR of every subpage nor if its even indexed so without knowing exact numbers, how do you apply weight fairly in that system? The server load if you did that would be tremendous.

And how would banner ads be incorporated? Would you then need weight bonus's for those to be fair?

Answers like "Who cares we would all benefit" won't be accepted as that is how these discussions usually turn.

Cheers

 

Thanks, YFS.

With my personal network, I had the same trouble grasping the potentially complex weight issue. Then it occurred to me that randomization is a perfect way to keep everything fair and balanced.

I'd say, stay away from micro-managing any such changes, so that if a PR6 is XXX, then it will get YYY, but only if QQQ etc.

Instead, do a straight random draw. When it's a link's turn to be laid down, it randomly gets one of these: full-rotation like it is now, rotation after 2 days, rotation after 10 days, rotation after 40 days, or rotation after 100 days.

I am by no means capable of writing such script, but believe it would be basic to assign each link to one of these categories randomly. Further, the member could "slide" the randomization by choosing the % chance they want for each category. So, if you want longer link times, you'd weight the 40 and 100 day category with, say 80%. But if you wanted more coverage and more rotation, you'd put 80% towards full-rotation and 2-days.

As for the categorization - I don't see the immediate capability of 50 categories. But, language could and should be the first thing. Visitors, and eventually engines won't see the relevance of a Polish language ad on my site. I'd say put the foreign language ads in another category right away. Then, have 3 main categories, then 5, then 10 and 50 as the network becomes able support it.

 

Can you explain that a bit better? My issue is say you have 100% towards static. The system says your weight allows you to be placed on 10 PR8 Sites (or some makeup of that). Right now a PR8 is basically the whole site when you may actually go on a PR0 subpage.

How to you get the fairness to be spot on (or even close) for that?

This would especially affect those with smaller weights which make up the majority of the network. With ads changing with every visitor on every page, statistically its more fair.

 

YFS -

I do not know the particulars of how the weighting system works, but if I understand you right:

Say the member's Master Site had a weight of 10,000. Currently, the 10,000 is spread out however DP does it. My proposal would *not* be to change the whole thing and how it is now done. I wouldn't want to micro-manage the PR of everything in and out.

Instead, Shawn has said it would be basic to make links static. So, it should be basic to have 5 categories of links:

A- Full rotation
B- 2-day rotation
C- 10-day rotstion
D- 40 day rotation
E- 100 day rotation

The member would choose what % of their 10,000 weight they want assigned to each category.

The members inbound links would be valued as before, but each link would be randomly assigned to one of the categories above (based of the member's preference for amount of rotation). So, some links would stick and others would rotate. The PR destination, etc. would be irrevevant - it would be just as it is now for each category.

 

The problem with that, is your weight allocation can be stuck for 100 days on a PR0 page(or more likely pages) instead of being rotated between PR0, PR5, PR8 pages, etc. Right now, with the instant change, the weight allocation onto these low PR0 pages isn't as big a deal, 'cause it doesn't last.

On the other hand, the possibility of getting a 100-day link on a PR8 might out-weigh the amount a PR0 stick would hurt. 100 days would probably even let Google pass PR(maybe even for a toolbar update!)... I still think there's a better system that can be determined, tho...


--
Derek

 

Derek -

I think randomization is the key to everything. That way everything equals out over time. More static links would just spread out the time a bit. But, still provide equal opportunity for all.

There would statistically be the same chance as there is now of one of your links getting on a PR6 or a PR0. It is just that the amount of time the links spend on each site would vary.

 

Are you a software engineer? Randomization is rarely a good solution, from my experience. Randomization has to be used as little as possible. When you're distributing links such as this, some people bring substantially more value to the network than others-- they should NOT benefit equally-- they should each benefit according to how much they give. If we were all equal, randomization might work. We're not.


--
Derek

 

You're right, it's not all equal, which is why I always thought the weight system was in place. Those with higher weights get their ads shown more often, showing their 'higher value' to the ad network.

 

Derek -

"Are you a software engineer? Randomization is rarely a good solution,"

No, I'm not an engineer.

I completely disagree with the randomization comment. I mean, how are the links given out now - randomly, right? I was not aware of any type of targeting.

My suggestion is to merely duplicate how things are currently done, but with some of the links sticking longer.

 

No, things are not random. They are handed out, according to Shawn, based on their weight in the network.


--
Derek

 

Shawn talked about this issue :
http://forums.digitalpoint.com/showthread.php?p=145119#post145118

 

Thanks, Wendy. Good to know Shawn's spoken on this issue before(I missed it originally). I guess that's the bottom line-- the coop is an 'advertising network', even though I get no measureable traffic from it(112k total weight)... I guess the smart thing to do, is to diversify our strategies to include a 'link exchange network' versus an advertising network. Since we're not being an 'advertising network', this shouldn't violate Shawn's TOS to use both...


--
Derek

 

it was easy to miss, it was in a long thread

hope it helped!

 

Thanks, Wendy. I did see that, but honestly can't understand why the network would stick itself with something when it is a certainty that G is going to eventually discount all the inbound churned ad links. Adding some variety - some semi-static links - will cure the problem, AND help advertising.

 

It's a free service that is "advertising" not a "link exchange". Be happy that your ads will give you some backlinks even if they are discounted in Google. It doesn't take a lot of effort to run the ads and it will help you rank well in Yahoo and MSN. Work on link exchanges on your own if you want to do well in Google.

 

I am not entirely convinced google visits the same page more than twice on some of my sites so the coop is having little affect on google.

Also google has sandboxed me since december and I am seeing no affect of the coop.

BUT other site engines are obviously faring better with the coop

Do we really want to change our theme to google tune ? I for one do not want to dance to it's tune. There are many posts here that describe irrevalant results in the SERPs, outdated results and multiple google algorithms used for the SERP ranking. With so many detrimental effects on googles you may want to consider...

Google is slipping in the search market. Google knows this and is expanding to other avenues as its competition catches up.

I am not entirely convinced that a simple publication of the google algorithm will devastate everything the coop has accomplished when you look at the big picture (all search engines combined) - Of course I realize if google is your only source of income you have to begin to worry !