Forum Moderators: open
Any suggestions as to what would be the top forms of abuse/fraud.
I also wonder if any of them monitor time spent on a site. For example people just clicking and immediately exiting your site
Affiliate programs for webmasters
which can be subject to manipulation.
No time limits on a site visit. For
example a competitor can click on your ad and
immediately leave your site costing you a bid.
Unqualfied PPC traffic floating around your site with no
intention / ability to purchase anything ie. children.
Programs and techniques to trick the fraud detection
sytems of PPC's ie. manipulation of the IP address.
any others? are these valid? thanks
Kanoodle has pretty good fraud program. I was noticing that they were not charging me for about 40% of the click throughs they were sending to my site and here is what they had to say about that:
"An uncharged click is a click that was interpreted by our fraud protection system as being of low quality. We do allow the user to visit your site but we do not deduct the associated bid price from your account balance."
Obviously I was surprised and impressed with that. Where they hook you though is by giving you a $5 start up account for free and then once your account runs low, The smallest amount you can buy is $50 to continue your listing. But for 5 cents per click, I suppose that's not bad.
AND As far as Affiliates go... I run a small PPC SE and the webmasters who sign up as affilates put our search tool on their site and when their visitors use it, the webmaster is credited towards his listing on our system. There's no "real payments in dollars" (only free advertising) to him, so I'm not quite sure how that is a bad thing, unless I'm missing something.
-robert
[edited by: rcjordan at 11:31 pm (utc) on April 28, 2003]
[edit reason] sorry, no specific references to sites. [/edit]
Here's a "dark side" for ya. Looksmart charges their advertisers .15 cpc (minimum) and for that much money, I was not about to find out if they have a fraud protection plan (which I doubt) or if the visitors were targeted.
Here's another. I don't know if this falls under PPC or paid inclusion, but here's a "fast one" Inktomi will pull on you if you're not careful...
What they do is somehow manage to show a few of your URL's indexed in their engine (which you did not submit to them) Nor did you pay for, but if you're smart, you won't try to add any more either - like I did BECAUSE:
One day I noticed that they had about 6 of my 40 pages indexed. I was surprised because I had never submitted those to them, so I thought I would be smart and at least PAY FOR the index page to be "included" which mysteriously was not there, - only some of the subpages were.
So anyway, I paid my little $39 to inlude the home page and about 3 days latter they added it and about 3 more days, they dropped the other 6 that were already in there. (NICE!)
When I complained, they gave me a lame excuse, and when I came here to gripe about it (which I should have checked first) - I noticed that I was not alone... And that really made me mad that they can get away with it.
-robert
But what bothers me is that most "joe six-pack" web surfers have no idea that the results they are looking at - had to "pay" to be there for them to see (which they probably wouldn't care if they did know) but I wish these people would not use the PPC and Paid Inclusion engines because they would find BETTER and MORE results using a non paid SE.
For example, Joe Six-pack goes to PPC number 1 and he searches for "beer in idaho" but he only gets 4 results because if he would have gone to webcrawler, he would have got 400,000 results.
So how do we educate the web surfer so they know to have better luck using "free" engines? This would obviously solve the problem wouldn't it?
-robert
Some comments:
No time limits on a site visit.
I'm sure we have all gone to a search engine, clicked on a result and then clicked straight back because we didn't like the site - this isn't fraud.
How do you suggest they monitor it?
Just looking at server logs you can get a good measure of how long visitors stay on your site. Even if it's not 100% accurate, as a comparison against non-PPC methods it's stll valid.
robenochs, I don't disagree with you about pay for inclusion, but that is a different subject for a different thread (or one of the many existing ones?).
For example, Joe Six-pack goes to PPC number 1 and he searches for "beer in idaho" but he only gets 4 results because if he would have gone to webcrawler, he would have got 400,000 results.
It depends what Joe Six-pack is looking for. If he wants to buy beer in Idaho he's better off with PPCs. If he wants to just read about it then webcrawler will suit him fine ;)
How do you suggest they monitor it?
Just looking at server logs you can get a good measure of how long visitors stay on your site. Even if it's not 100% accurate, as a comparison against non-PPC methods it's stll valid.robenochs, I don't disagree with you about pay for inclusion, but that is a different subject for a different thread (or one of the many existing ones?).
So the questions remains: How do you suggest PPCs monitor time on a site?
How do you suggest PPCs monitor time on a site?
Why should they? They sent you the visitor and it's up to you what you do wuith them. They don't have to because it's nothing to do with them. If you have an awful site that visitors leave straight away, that isn't the PPCs fault. It's up to you to monitor exit times and the time people stay on the site. And then if you don't like the time spent or ROI from Overture, for example, then don't use it :)
The biggest fraud we have run into has been partner sites that feature certain headings or listings outside the regular search system. This becomes very obvious when you suddenly get 40 clicks on a term in one 24 hour period that only has had 3 in the previous 30 days.
As I write this, I'm wholesale deleting terms from our ah-ha.com account. Bad just got a whole lot worse, don't know why I bother with them for the occasional diamond in the rough.
I check my HitBox logs and not one Visitor shows up from any of the FindWhat partner sites nor does that search term show up in the "Search Words" category. I then do a simple test and click into my site using the same keyword with a FindWhat partner site and it gets recorded into HitBox within seconds. The only statistic I see is the number of Bookmarks skyrockets up.
Lastly, I can't believe that 500+ individuals actually scrolled down 40-50 place rankings in a FindWhat partner listing and then clicked into my site. it's beyond belief.
Wonder if the State's Attorney Generals Office would find this an interesting case?
We establish a separate landing page for each PPC. On that page we put 2 tracking scripts, both java, both at the bottom of the page. What's interesting is that although a weblog may record a visitor, sometimes they are not on the site long enough to trigger the javascripts. We interpret this as autobot visits. Once the javascript is triggered, the next test is to see if they actually proceed into the site at least one more page. If they do then that's good, the traffic is good, and if we fail to get a sale from them it's our problem. If they do not proceed to at least one more page we consider them bounced. It is very interesting to compare the bounce rates from campaigns. Some of the ppc engines mentioned here have bounce rates in excess of 90%. The average is about 30% and anything below that we consider good targeted traffic.
The final step is to actually look at the ROI per ppc engine. We use a special piece of software that does just that... takes the number of vistors from each page and then gives us the percentage that actually wound up in a sale.
We have now determined that there are very few of the popular PPC engines that deliver truly targeted traffic. Some with good customer service have terrible bounce rates, while some industry specific search directories are probably some of the best sources of traffic. Do yourselves a favor and invest a lot of time and some money in tracking these campigns carefully. You could wind up spending thousands of dollars and get zero return unless you can truly measure the effective of a campaign.
Although, we consider some of the same information, we tend to look at ROI first and then the "bounce rate".
Also, I'd like to add that while some PPC are almost a complete waste of time with only high bounce rates, the bounce rate for some PPC will vary with the Keyword used. So if we decide to target some new keywords, then we start testing all over again with the various PPCs. :)
For a second I thought there was a world wide rush for my product, but alas no such luck. There was no measurable increase in sales or inquiry activity.
: (
I compared the activity for the keyword to others I use because there is a pretty predictable ratio in click volume (for comparable rankings) across many search engines and again this ratio was out of whack by a factor of 15 at least.
I did send them an e-mail and am awaiting a response. In the meantime I have reduced to bid to .01 as i am now a bit gun-shy to say the least.
Any comments?
7 search - 4730 visits - 62.03% bounced
findwhat - 3445 visits - 47.31% bounced
ah-ha - 1149 visits - 81.64% bounced
kanoodle - 1060 visits - 81.70% bounced
Xuppa (Bay 9) - 842 visits - 88.0% bounced
goclick - 795 visits - 75.6% bounced
ispcheck - 324 visits - 19.75% bounced
overture - 297 visits - 31.65% bounced
epilot - 257 visits - 92.61% bounced
ppcforhosts - 257 visits - 28.02% bounced
searchfeed - 254 visits - 62.20% bounced
hampstertrap - 70 visits - 82.86% bounced
linkpassport - 59 visits - 91.53% bounced
looksmart - 43 visits - 65.12% bounced
kazzaz - 36 visits - 100% bounced
hithopper - 28 visits - 85.71% bounced
towersearch - 16 visits - 68.75% bounced
Epilot and towersearch claimed the largest number of clicks that never actually showed up in our stats. Although epilot talked a good line and promised to reimburse us and work with us to determine the bad traffic source, we never did receive a credit and in fact they subsequently tried to upsell us. Also, some of these campaigns burned through our money really fast. We did complain to goclick and received absolutely no satisfaction from them. Others, we just cancelled the account.
Peter
Director of Sales
24-7 Webs
ROI is the only true measure that decides the success or failure of a PPC.
Put simply - we spent £X and made/lost £Y.
No two PPC's are the same, the manner in which they show results is different, the advertiser profile will be different, the demographics of what sort of users you find your ads in front of will be different, the pricing models are different, the volumes are different.
But I go back to it, I spend X and make Y. As bid managers we even encourage clients to include what we charge them in X and in some cases even take the associated costs in delivering the product/service off of Y to get a truly representative ROI, and that keeps it all realistic.
PPC's will investigate the issue of fraud. Unfortunately, some of the metrics mentioned in this thread tend to indicate poorly written creatives and/or landing pages not reflective of the search and to try to place the blame of the PPC's for the relevancy rules they have is not fair.
I've seen many people complain about the fact the editorial people rejected ads for relevancy, if they were just in it for the money, then surely they'd take anyone's money for any keyword right?
Fraud does exist, and everyone needs to be diligent in working together to stamp it out.
Provided they go that extra one page, we can assume resposibilty for everything else beyond that. As I said in an earler post, we also track the ROI per PPC engine, so we have a pretty good idea who sends quality and who doesn't.
I should add, that relying completely upon server logs means you may be looking at autobot hits instead of genuine traffic. That's why we use 4 different measurement tools.
Take from the stats for what you will. We didn't say anyone was committing fraud. We just gave you the stats which are hard to argue with. We base our decisions on whether to spend large sums of money with a variety of advertisers based on our own models, these stats being a large part of that model.
Peter
Director of Sales
24-7 Webs
I wasn't being critical of your stats, indeed I applaud your ability to track to that level, it's just that the anomalies have to put a +/- factor in which can skew results a bit.
We have campaigns running on some PPC's (at clients request not ours), where they get 1000 clicks and no sales, whereas on first tier providers we'd expect to get around 20 or so sales. We don't need to be that granular to know that this isn't right.
I understand your methodology and it's good as long as the site purpose is to make sales. We run sites for a few clients whose purpose is not to make online sales, but still use PPC for branding and traffic. In these cases there is no measurable ROI that can be clearly defined by sales to clicks ratios. That is part of the reason we use our system of tracking and analyzing and subsequently saying yes or no to PPC providers. In so doing we identified those providers we prefer to do business with and those we don't.
Re People Spaces June 21 post: "Some of the ppc engines mentioned here have bounce rates in excess of 90%. The average is about 30% and anything below that we consider good targeted traffic."
Somewhere, about 6 months ago or so, there was a similar topic and a response from one of the "senior" members of this club (WebmasterWorld.com ) who owns a company promoting web advertising, who stated flatly that he had been tracking this problem of PPC fraud for some years and claimed he could prove 30% of all PPC traffic on average was fraudulent.
After all of the above is recorded and digested, I think we should all realize that probably that guy was right: about 30% or more of all traffic from PPCs is probably fraudulent. Those with the most sophisticated tech knowledge, and tech tools, may be able to sort out the fraud from the real, and challenge a PPC, but most of us are just guessing.
Incidentally, in all of the above, unless I missed it, I did not see any comment about fraud from Google Ad Words ...
As for comments on some previous posts: Ah-ha and Kanoodle.., and in general all the cheapest PPCs: I tried them a year ago and spent major bucks for nothing. I don't trust either of them at all, and no others at that level, no matter how cheap their keywords are.
As for Overture: I have had problems with them, with "spiked' results over a 7 day period, then back to normal, then "spiked" again, back to normal, and spiked again for about a two month period last year. I pointed out the anomoly in traffic for my product for that time of year -- it was totally out of bounds. I did not get a satisfactory answer from Overture.
Someone in a previous post here said the PPCs will investigate, and correct fraudulent use, but most of the posts here indicate they weasel out it, or just flat do not admit to the problem, from their investigation. That was my experience with Overture. I still use them, and also use FindWhat, but I use them very carefully, and I control the spend rate with them carefulluy.
I go back to that person who some months ago who said he could prove 30% fraud rate. (Wish I had saved that post somewhere), but I think that is probably what all of us have to contend with. Except, again, no citation here of Google Ad Words fraud, in all the posts mentioned, unless I missed it.....?
Are you working for them or somehting? They are the worset or one of the worest anyway.
<<<again, no citation here of Google Ad Words fraud>>
And now Google is using affiliates so will have to wait and see. Affilites like said here before is a problem in this business
Kanoodle has pretty good fraud program. I was noticing that they were not charging me for about 40% of the click throughs they were sending to my site
I'll re-word what I tried to say in this thread [webmasterworld.com]:
From my point of view, as a distribution partner, Kanoodle does filter out more traffic than my other 7 PPC partners. i.e. clicks are sent to the advertiser but they are not charged for it. Same site, same traffic but on average Kanoodle filters out 1 in 3 clicks whereas Findwhat, for example, filters out 1 in 7.
Anyway...... from the other side of the fence Kanoodle is one of the most strict when it comes to working out what constitutes a valid click.
Just because Kanoodle rejects 1 in 3 clicks doesn't mean they have better fraud tracking than Findwhat. . . it could just mean Findwhat just has higher quality traffic to begin with, therefore they don't need to reject as many clicks as Kanoodle.
it could just mean Findwhat just has higher quality traffic to begin with
Traffic I send to Kanoodle and Findwhat advertisers comes from the same site, so traffic quality is exactly the same. Same site, same traffic but Kanoodle filters out more.
