PPC Fraud - what don't you want to pay for?

15 comments

Since I'm working on a bit of a ppc engine right now I'd like the community's feedback on clicks they feel they shouldn't have to pay for.

Here's some of the obvious ones:
- clicks from non-targetted countries
- clicks from you on keyphrases you advertise on
- clicks from your competitors on keyphrases they advertise on

Then we start getting into some grey areas, like anonymous proxies. Some folks use them legitimately, others use them to mask their click fraud attempts. Yet banning them entirely seems a bit adverse for the PPC provider.

What type of clicks are you unwilling to pay for in your PPC ad campaigns?

Comments

None

Quote:
I'd like the community's feedback on clicks they feel they shouldn't have to pay for.

I don't really want to pay for any click. :)

Why reinvent a broken wheel

Bid per hour vs per click. That way you don't need to waist your time building filters and algorithims that will end up not working.

Show users peek hours over a week so they can bid higher on high volume times.

Besides that don't get branded as the best source of unqualified asian traffic like findwhat has become LOL.

i agree with what seomike is

i agree with what seomike is saying.

i also think that any attempt that improving PPC requires increasing its complexity, which may prove to be an even greater problem than click fraud.

BPI

Turn it around completely and go bid per impression. A buck cpm and a 2% ctr comes out to a 5 cent cpc as a starting point. Looks the same, acts the same without the hassle. Fold in seomike's dayparting idea and it doesn't sound too half-baked.

PPC is broken but it can be fixed

The PPC model can be improved on without sacrificing the business model or making it more complex. It's all about the GUI and the advertiser's overall experience.

Although I do like the idea of a bid per hour model my experience working with engines tells me that the CPA and CPM models don't work for search ads. I've had the luxury of toying with those ideas in the past with beta advertisers and even though the math on those models looks good the numbers don't work out in the wash. From the engine side of things the reason the adoption of other models has not become more prevalent is because CPC kicks everything else's ass from a revenue perspective. There's just no way to slice it or dice it. From the advertiser/marketer perspective I agree with Jason. I don't want to pay for clicks either, but I'll ad, unless they are productive.

When it comes to proxies, all anon and/or abusive proxies should be blocked by default. Period. Those are usually the lowest quality of all clicks. If something cannot be verified the advertiser should not be charged for it.

Again, we go back to can click fraud be detected? Yes it can. It's really not hard to do if you can gather the right data. Can it be worked around, sure. Yet it is possible to build a system in such a way that if someone does work around it the system can spot that as well, red flag it, and then engineers can adjust the software accordingly, if necessary. Engines have a financial incentive not to stop click fraud. This is the only reason an industry wide practice has not been implemented to address the problem. There is far too much revenue on the line.

As an advertiser I do not want to pay for anything that I cannot track and/or verify. Unfortunately the system doesn't work that way right now with any PPC search provider. It will one of these days. You can count on that.

Bid per hour vs per click

How does that work then?

Duplicate Clicks - PP(unique)Click - PPUC

If a user goes back and for between the SERPs and my page a few times over the course of a period of time (say an hour or so while they comparision shop) then I should only have to pay for that unique visit ONCE. GOOG and YSM love to charge us as many times as possible. Which also makes it easier for our competitors to run up our tab.

Only making us pay ONCE per unique visitor to a unique domain per 24hrs sounds fair. So if the user clicks on three different ads of ours for three different terms, and go back and forth between the serps five times each AND are one of these crazy double-click on every link you see is how the web works (you know these guys? my dad is one) instead of charging us for 30 clicks, just charge us ONCE for that unique user.

That should be common

A good majority of PPC engines have a timeout setting for charging a click to an advertiser. Usually that is 1 click per unique IP in 24 hours. I know that is standard practice for most. If they are hitting you for more than that it would be wise to contact the engine and find out what their policy is. You may want to adjust your bidding to account for it. I do know in certain very heavy verticals that the time allowed between clicks per IP is lower. This is because the engine assumes that the same user is doing more looking or getting closer to a buying decision.

Rynert, the idea behind bid per hour is just that. You determine which hours each day that you want to have your ad appear and set an appropriate bid amount for your listing to be there. Each hour that the listing is active your advertiser account gets charged the appropriate amount based on the bid you have preset. It gives a bit of flexability to businesses because it enables "day parting" so you can display ads for only the hours your business in in operation (for example). You can also target certain demographics using day parting to help increase ROI. If you know that people ages 19-25 for example are online mostly during the evening hours and you run a music download site then you may only want ads to appear betweeen 4pm and 1am. Sketchy examples but I'm doing my best to simplify what I think seomike is talking about.

How does that help?

Apologies if I am being dim here, but putting asside 'day parting' being a good idea on it's own, I don't get how it helps avoid click fraud.

a) spend 1000 for a given hour, get 100 clicks and generate 2000 in sales.
b) spend 1000 for 100 clicks and generate 2000 in sales.

Both methods will have valid and fraudulent clicks, in both methods you spend n-amount and compare to n-sales (assuming sales was the goal)

If the point is that by choosing the specific hour with the most searches, therefore a higher proportion of legitmate clicks, you reduce the % of fraudulent clicks then surely the fruadsters will simply target those hours?

Not being dim at all

If I am getting what seomike is saying it's that you're only paying for traffic by the hour, not by the click. Therefore if a fraudster wants to defraud you they can certainly try but they will not get very far because the percentage of revenue they would recieve would by next to nothing. The publishers would have to be paid by the hour as well so there is no point is NOT showing the ad if it is going to be displayed anyway. It's a flat fee system. One fee for all the displays of the ad per hour regardless of where the ad is displayed. CPC and CPM impressions would not count under a pay per hour model (CPM counts would only be for stats) so the ad revenue dynamic is completely changed.

I like the idea of it but the problem with pay per hour is that you could never build a network of publishers around it. At least not a major network. A site would have to own their traffic (be the size of a MSN, Yahoo, or Google) just to make it pay off. The minimum hourly rate would have to be huge to draw publishers away from the PPC and CPM ad model and establish a network. It's actually a decent idea but one that would have to overcome a lot of hurdles I think.

mmm advertisers

I had not thought from the other side of the coin. Would it even be possible to run the 'by hour' concept outside of the search sites themselves. How you would pay advertisers?

Say I pay 100 for one keyword in a given hour. Google would give away, being generous, 20 of this. Would it aggregate all the clicks on all their adsense customers then divide the 20 pro-rata to those people?

This would end up with lots of people getting fractions of a penny. If they round up it will blow the 20, if they round down people will get nothing, if they leave as fractions it would be a nightmare to manage.

band aids

won't fix a faulty concept. While you may have to eschew billions and settle for millions (not so bad a thing ;) the true sponsor model will eventually win the day, MSN has the upper hand here in as much as they can lose money on search and still laugh all the way to the bank. The CPC killer will be the winner, G$ hits the skids in a hurry when someone steps up to disrupt the status quo.

You "own" your slot for a given period of reasonable time (not minutes or hours) and thats it. Of course the bidding process can be incorporated. Want Viagra for a month PLACE YOUR BID. Offer branding opps instead of stealth SERP spam and you have a model with longevity.

Fix a problem, make money, from acct. mgmt. to click fraud, CPC is really just a big problem begging a solution.

not a real solution IMHO

Want Viagra for a month PLACE YOUR BID.

that probably puts the term out of reach for many who buy it right now. including those who overspend briefly every so often.

Exactly

Seobook this is why I will always believe that CPC is a good model. It lets the little guy play ball with the big guy on a level playing field (for the most part) Yet, like Hardball said CPC is a "big problem begging a solution". Someone must come up with those solutions. Sooner, rather than later because at this point the entire model is in jeopardy.

at some level

CPC sounds so democratic and fair but reality is its the "little guy" who goes in with irrational bidding skewing results for those with a clue, CPC is running on the "sucker born every minute" wind, for every bozo that lost his shirt with irrational and UNPROFITABLE methods yesterday you have a new queue lined up to make the same mistakes, now you can make a profit from the mistakes of others for awhile but eventually it will dry up.

I'm not suggesting that the little guy gets locked out but you could save him from himself with a more rational spending model. Fraud and CPC don't help anyone, especially the "little guy", his budget is burned before he knows what hit him. The big guys have the upper hand, it takes months for the reports to get analyzed, combed, finagled, presented, approved, measured, re-budgeted blah blah, and at the end of the process (in itself a big expense) they call failed campaigns "branding".

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.