| 7:46 pm on Apr 29, 2005 (gmt 0)|
Did I say it was fully launched?
No, I said "being deployed", I saw it was still in testing.
Did I say double the money?
What I wrote was "2x bonus of free clicks" - I didn't say money.
If your technology is as good as it's claiming the least you could do if someone suffers abuse as a show of good will, assuming your software is really up to snuff is give them some free clicks considering the customers they probably lost when their ad budget was suddenly drained.
If the technology is solid enuogh you may rarely ever have to do it, if it's not....
| 8:14 pm on Apr 29, 2005 (gmt 0)|
Never mind, it was just a suggestion to lure advertisers, not start a war of words.
| 8:46 pm on Apr 29, 2005 (gmt 0)|
No Incredibill I understand what you mean but in PPC clicks cost money. Giving 2x the clicks is also 2x the cost. Sorry, if I confused you there but I wanted to clear that up for you.
If our technology is as good as we think it is then we really won't need the guarantee. It's there for assurance. We're not expecting it to be used all that often unless someone finds a way to beat our system. As with all technology that is possible. However we've biult the system to spot those leaks so that we can kill them quickly.
We'll fix a problem if one is spotted and then re-launch the campaign once we can assure the advertiser the problem is solved - and not before then. Why? If it's not fixed then our arse in the hot seat because the onus is on us to do it. That is an advertiser that will never return becuase they lose all faith in our ability to perform as a company if we can't fix the issue.
That is something I will not stand idly by and watch happen.
If you need more incentive to try us out when the program is ready all I can say tell you is this; would you rather keep advertising with companies that cannot protect your ads from abuse or ones working hard to do so? This is a choice every marketer needs to make. If indeed we have the right solution, which I am confident we do, why would you advertise with any other middle market search network?
| 9:19 pm on Apr 29, 2005 (gmt 0)|
"BTW, if Google or Yahoo had technology like this and they haven't implemented it that means that they were purposely profiting from click fraud. Why have a product and not use it? Ahhh.... revenue! "
Joe - How long did you work at Kanoodle?
| 9:41 pm on Apr 29, 2005 (gmt 0)|
|Sorry, if I confused you there but I wanted to clear that up for you. |
I'm rarely confused, one of the least confused people you'll ever meet, which is amazing considering I just passed the competitive trials and am now a member of the U.S. Olympic Drinking Team.
I understand what your costs are in a CPC program, run them direct myself for some Fortune 500s even, and I've given away a few clicks to make customers happy before. You don't even want to know why I did it because it will make your head spin worse than ClickFraud.
However, I'm going to tell you because spinning heads and spewing pea soup amuse me greatly.
Here's the short synopsis of the problem which you may have already addressed, and it has a simple solution. Based on some stats I saw in my own CPC efforts this problem could cost advertisers up to 10% (or more) of their budget in some cases and it's not even click fraud.
A) Visitor is on my web site, sees ad and clicks ads.
B) Click is counted via my CPC tracking script and redirect made to advertisers web site
C) Visitor never gets to advertisers site due to various internet routing issues and gets a timeout being redirected. Advertiser log files showed fewer referrals than my stats show clicked on the link which varied from less than 1% to much as 10% on some days when something nasty happened on the net.
When I first encountered this problem I simply gave the customer the number of clicks they paid for, so if they paid $1 CPC for 1,000 clicks I let the campaign continue to run (although my side showed more than 1,000 clicks) until their web stats showed the actual referrer count reached the 1,000 click target.
After that I installed a tracker in the advertisers site and the click is only counted/charged when the tracker on the destination page responds. That means not just 1,000 clicks, but 1,000 verified visits to the advertisers site are counted.
You would be amazed how much it improved their conversion rate with verified visits.
Also, advertisers don't have to panic and shut down their advertising when their site goes offline, another perk for the advertisers as the clicks just aren't counted.
I think it's a huge problem and a dirty little industry secret, but I won't tell anyone if you don't :)
| 9:59 pm on Apr 29, 2005 (gmt 0)|
I was at Kanoodle nearly two years. I am not going to discuss any details of my knowlwdge on Kanoodle. It's not that I am not allowed - it would not be professional of me to do so. If they want to answer questions let them join in. I don't work there anymore so I cannot answer for anything they do, have done, or will do in the future.
Incredibill, There's always been a problem in traffic stats from one site to another. It's normal really. I believe this should never be more than a 10% discrepancy. Still it's up to the advertiser to make sure their server responds right. We can't be responsible for that.
I like the idea of what you are proposing though. It's worth some looking into. The problem is in implementation. Like I said before getting advertisers to install an ROI tracking code is hard enough.
They know it's a benefit to them but yet they still fail to use it. What can I tell you? I think I see a solution though although it would take a lot of doing. I'll think on it with our develpoment team some more.
| 10:18 pm on Apr 29, 2005 (gmt 0)|
incrediBILL, thanks for posting your description of click discrepancies. This is yet another reason why I think CPC is a poor business model, click fraud aside. The Internet is a best-effort, not a guaranteed transactional service, and thus there will be discrepancies. The degree of discrepancy will vary depending upon the user base, their connectivity to the publisher and advertiser's sites, etc.
Some of you may be interested in a Usenix paper I found that uses digital signature and hashing techniques to communicate the referred traffic data from the advertiser to the publisher.
| 10:46 pm on Apr 29, 2005 (gmt 0)|
|Incredibill, There's always been a problem in traffic stats from one site to another. It's normal really. I believe this should never be more than a 10% discrepancy. Still it's up to the advertiser to make sure their server responds right. We can't be responsible for that. |
Problem is, some advertisers have no idea how to do this. Even some of their hosting companies don't know how to do this, or if they do, they will charge the advertiser extra.
Not to throw a wet blanket over Internet advertising, but I think if there is to be a real, robust future to it, some minimal acceptance criteria will have to be met by both advertisers and publishers with regards to analyzing traffic seen at both. No, it won't be perfect; it won't eliminate click fraud or discrepancies. But it's something that both advertisers and publishers will both be able to abide by, because it will have been vetted by communications technology professionals. The question is, how willing will advertisers and publishers be to pay for this technology? It isn't as easy as just putting your site on the web and filling out a short signup form.
| 11:06 pm on Apr 29, 2005 (gmt 0)|
|I believe this should never be more than a 10% discrepancy |
Believe what you will but when a HUGE router or some connecting optical gear blows at Verizon or Level3 and a site is down 8-16 hours while a tech drives cross country to retreive a $100k part (they dont stock a lot of redundant $100k parts it seems) your site is off 100% until they restore service and the ads are going nowhere when clicked.
I have 3 customers that are huge CPC users that have pager notification when they see server access is down, then they call someone to disable their ad campaigns until they get the notification that service is restored.
If the ads are running, CPC or CPM, they are just wasted until service is restored in the current paradigm.
I've also had customers complain that they can't connect to a perfectly fine running server (no service alarms at all) night after night and it wasn't the destination network at fault. It turned out to tbe broadband provider doing maintenance so anyone in their network wouldn't be able to complete the connection to the advertisers server during that period.
Maybe the simpler (but incomplete) solution is just to put all your advertisers URLs in a site status monitoring system and ping them every 5 minutes. When your monitoring system sees a site go up/down it relays this information to the ad server which would enable/disable the ads for that site. It's incomplete though as it could just be YOUR server that can't access the other site, would need redunancy from multiple geographical locations to all compare notes and agree it was down.
Don't stare too hard at the network or a packet will drop.
| 11:32 pm on Apr 29, 2005 (gmt 0)|
Gregbo, I think you're bringing up a valid point. However the beginning of this PDF assumes that "click shaving" is done on purpose. I can state that in all my experience with ad networks, all the people I know, etc... that I have never even heard of this practice happening on a PPC search network.
We're talking about two separate issues here. Click Fraud as done by malicious publishers and click shaving done by the the ad network itself. If my assumption is correct the PPC networks don't really need click shaving because there is already enough click fraud to go around for everyone. I can say with a good amount of confidence that click shaving is not a practice I have ever seen in the PPC search market.
That being said, it is a practice I saw taking place back in the days of PPC banner ads. Being that the paper is from back in 1998 it was a problem back then. In today's ad market I am not so sure I can be convinced of that.
"The Internet is a best-effort, not a guaranteed transactional service, and thus there will be discrepancies."
Well said. There are a lot of different factors involved and I intend to read the document in full over the weekend to see if I can spawn any ideas from it.
"Problem is, some advertisers have no idea how to do this. Even some of their hosting companies don't know how to do this, or if they do, they will charge the advertiser extra."
Also well said. As someone with a web hosting background I know that data can get lost in transit no mater how fast servers are - it happens. Packets get dropped sometimes and even the fastest servers cannot avoid that.
From the ad network's perspective once the click occurs the advertiser gets charged. I can guarantee you that the ad network never misses a click because that is revenue in their bank. So for the most part you can assume accurate click counts from the ad networks.
Here's where the problem comes in.
Delivery of that click should occur as long as the advertiser's site is up on the other end (sans the idea of click shaving like you mentioned) and there is no network congestion taking place choking up the pipe, hardware failures, etc.
The tracking of clicks can be delayed by any number of factors. The most common of which is a delay in the response time from 3rd party site counters and/or tracking pixels.
Slow page loads ans slow databases are an issue too. This can be avoided to some extent by making sure your site is not being hosted on crowded servers. Check our your site's IP address on Whois.sc and find out how many sites are on your server. If it's virtual/shared hosting with more than 75-100 sites move your site. I never had more than 100 domains on a server as a web host unless it was a dedicated machine for one user. Some web hosts have over 1,000 sites on a single server. That's really bad overcrowding in my book.
Lastly, tracking software like Analog Web Stats, Webalizer, or Urchin are DB dependant. It's a better option but still has issues. If the DB is slow to respond then stats will be off. This is aggrivated even more if the visitor closes the window quickly. The average user spends 11 seconds on a web page. If the DB is slow that will tell you why even more tracking problems exist.
And there are even more issues involved in tracking that can go on and on.
Maybe there is an answer to the problem of click delivery. However the real problem right now is click fraud. Stats I'll deal with. I for one want to be sure absloutely sure the visitors I am sent from a search engine are REAL. As long as I have that assurance then stats may well be the next issue that needs to be addressed by both the advertiser and the ad networks. It's an issue that should not be ignored but I don't think it's the most pressing one we are facing at the moment.
| 2:49 am on Apr 30, 2005 (gmt 0)|
I was trying to point out that the issue of click fraud and discrepancies is part of a larger issue which is basically the limits of what can be accomplished over the Internet today with a cost per [some action on a web site] model.
I appreciate your feedback. Getting these underlying technical issues out in the open is important, IMO, because an understanding of them is vital if Internet advertising is going to succeed. The better the understanding of the problems, the better the solutions will be.
| 1:59 am on May 3, 2005 (gmt 0)|
|This is aggrivated even more if the visitor closes the window quickly. The average user spends 11 seconds on a web page. If the DB is slow that will tell you why even more tracking problems exist. |
If you put the tracking bug at the top of the page instead of at the bottom your chances of counting the tracking first are much higher.
The technology exists to implement a full duplex advertiser tracking system and not drop a single valid transaction. If the DB is slow then you have an implementation issue, that's a technology problem which doesn't mean this isn't possible. If your programmers are worth the salt used to season a steak they will understand what a FIFO QUEUE is and simply buffer up requests until the tracking engine can process them as the desired result is to instantly respond to the web page sending the tracking request in all cases so the advertisers web site gets instantaneous responses.
| 10:01 pm on May 10, 2005 (gmt 0)|
I have been testing Blowsearch for a client of mine and while they may be controlling click fraud their distribution still is very poor for my industry. Other 2nd tier cpc programs have converted better than Blowsearch for us.
Just my 2 cents that click fraud is not the only issue with programs. At the end of the day it's all about conversion, and with Blowsearch conversions have been less than even Kanoodle, FindWhat, Epilot (and I certainly don't endorse these companies).
I was hoping Blowsearch would live up to its' hype...so far it has not.
| 2:48 pm on May 11, 2005 (gmt 0)|
Traffic on middle market players is generally less to begin with. There's plenty of reasons for which I have openly discussed before. The quality of traffic is more important than the quanitiy. The middle market - is the middle market- because for the time being it is a supplement to whatever campaigns you may be running elsewhere.
Keep in mind alot of factors influence the amount of traffic you recieve. One is the ketwords selected. Are they top terms with good traffic volume? Another is are you bidding properly? The more you bid the greater your chances are of getting traffic, especially on niche targeted terms, from the network.
Lastly are you using our ROI tracking tool and working with you assigned account rep (yes you have one) to improve the campaign?
Please keep in mind that Click Defender is also still not live across our entire network. We are still in testing at this time.
| 3:38 am on May 12, 2005 (gmt 0)|
"One is the ketwords selected. Are they top terms with good traffic volume? Another is are you bidding properly? The more you bid the greater your chances are of getting traffic, especially on niche targeted terms, from the network.
Lastly are you using our ROI tracking tool and working with you assigned account rep (yes you have one) to improve the campaign?"
When I posted that my conversion with Blowsearch were even less than on Kanoodle, Epilot and other 2nd tier CPC's, Yes I was bidding on the same top keywords in my industry. The test was exactly the same for all programs; that's what a test is. The same keywords, position and ultimate conversions. Blowsearch was as bad as any CPC program I've seen.
I'm not trying to 'ruin' your reputation here, but many people on this forum know that I know what I'm talking about. This is objective information about Blowsearch, not an axe I have to grind with them.
Always request a 100% free trial with ANY CPC program. You should never pay to 'try' a new program that needs YOUR money. I didn't lose a penny trying Blowsearch, just didn't make any either.
| 1:32 pm on May 12, 2005 (gmt 0)|
I don't understand how that could be possible seeing as how we do not offer free trials and we have not for quite a long time now. How long ago was this "test" you ran?
Again, I'll point to the fact that there are a lot of things that affect conversions. One of which is the site the traffic is being sent to, the titles and description used, etc.... etc....
If you'd like to PM me your information I'd be happy to discuss with you the details and get to the bottom of why it didn't work for you. The bottom line could well be that we don't have the kind of traffic that works for you. However I the thing I don't like to see is claims being made without the opportunity for our company to fully work with you to see if we can improve it.
Did you work with your Account Rep to see if we could correct any issues for you?
| 4:21 pm on May 12, 2005 (gmt 0)|
My blowsearch test is ongoing, but I have been disappointed to see where the traffic appears to be coming from. I haven't seen any referals from the sites they feature on their site, but plenty from the "xyzsearch.cc" type sites that none of us have ever or will ever use.
I'm sure if I started bidding google $ for terms I'd see some better sites, but if I was doing that why not just use google?
| 4:58 pm on May 12, 2005 (gmt 0)|
I set my Blowsearch bids to the #1 position for the terms I tested so I guess I tested their BEST distribution partners. But those traffic partners did not convert either.
In the end, the saying 'you get what you pay for' could not be any more appropriate when it comes to CPC programs. Sure Google and Yahoo are very expensive, but they convert. And at the end of the day that is all that matters. Sure we'd all love to find programs where the cpc prices for top spots are $0.35, but there isn't one out there I have found and I've tried them all.
| 5:56 pm on May 12, 2005 (gmt 0)|
Boston see my PM in reply to yours.
Gamb, you said "I haven't seen any referals from the sites they feature on their site, but plenty from the "xyzsearch.cc" type sites that none of us have ever or will ever use." That's what the traffic source selection is for. If you're tracking ROI using our ROI tracking (which you can use in conjunction with your own) you will be able to deselect any traffic sources you don't want. The .cc also connotates that we allow traffic from foreign countrie to click your links.
The default traffic settings for BlowSearch allow for traffic from North America (USA and Canada) and Western Europe. You can also select the regional targeting you want. For most of my own campaigns on BlowSearch I choose North America only.
Remember these are things that are in your control. If it's not working for you then we need to be looking at it with your Rep (hell even I'll look at it for you) but I need the account details.
The fact that the traffic didn't work for you does not mean it will not work for someone else. Campaigns will work better for some than others. We have advertisers who see good ROI and we have others who might even see none.
We've worked with advertisers who flat out said our traffic sucked and they would never come back - who now say that they are getting good ROI - after we worked on their accounts and tweaked their campaigns.
Poor performance is not necessarialy a traffic issue. It can also be poor site design, a bad promotion, bad price points, inneffective titles and descriptions, poor qualification, etc.
What does that say for BlowSearch? It says we're the same as every other network out there when it comes to the issues our advertiser have. I've seen campaigns fail miserably on Google and Yahoo too and I am sure you have as well.
It's about working together to solve the issues so that a campaign does do well. It takes time and effort on both sides to do that. I don't know the details behind Boston's or Gamb's accounts but I'll be more than happy to step in for any advertiser at any time. I don't give a darn how busy I am at that moment either.
If you guys want my help all you have to do is ask. However my input may involve changes in the creative on your end, among other things. Middle market campaigns cannot be run (in many cases) the same as a 1st tier campaign.
| 6:00 pm on May 12, 2005 (gmt 0)|
hehe this is funny... i can write such thing in 30 minutes. and that should be an important news?
| 8:21 pm on May 12, 2005 (gmt 0)|
I think the fact that you guys are making all of this effort says a lot. I also thought the phone call that came in after we signed up was a nice touch - BUT it all means nothing if you partner with poor sites. I understand I can remove them, but I have a feeling you are going about this backwards - just start with good partners and you don't have to worry.
I do worry about the quality of the traffic and will keep an eye on things and keep tweaking my campaigns. I'll keep everyone posted.
| 8:55 pm on May 12, 2005 (gmt 0)|
Thanks Gamb. For me it's all about the service end of our business.
And I won't say you're wrong about quality sources of traffic. We have actually eliminated over 20+ partners over the last couple weeks and we are replacing them with other sources.
The nice thing coming about Click Defender is that regardless of who sends us traffic we're only sending what we know is real and valuable to our advertisers. Once it's out of beta and we're done testing it will be a complete solution.
At that point the onus is on the advertiser (with our help of course) to convert the traffic we're sending them. If we can verify and guarantee that all the traffic we send is from real human users then there are no more excuses, no more guessing, and no more blaming the (at least) our engine for poor quality partners, traffic, etc. If it's a human user then and they clicked your link then they were either, browsing, shopping, or buying.
That is where REAL ROI in search will be able to finally be calculated.
BTW, we're also working with TrackingROI and Clicklab to verify what Click Defender can do. We have heard the need for 3rd party verification and we're incorporating it into our business model.
We're getting there. I'll keep everyone up to date.
| 2:21 am on May 13, 2005 (gmt 0)|
|If it's a human user then and they clicked your link then they were either, browsing, shopping, or buying. |
From the standpoint of a disgruntled advertiser (insufficient conversions), a frequent clicker is just as bad as a fraudulent clicker.
| 1:39 pm on May 13, 2005 (gmt 0)|
It sounds like your expecting one conversion to equal one sale. That's usually never the case. As a matter of fact a lot of recent research shows that users don't usually make online buying decisions for weeks.
Staying on that point, the most you'll get charged for from the same IP is 1 click per day. Repeated clicks on the same listing more than once in 24 hours are delivered but not billed.
See the study done by DoubleClick and ComScore. It's on the main page of the Doubleclick site under "What Impact Does Search Have on Purchase?". It's a PDF and a really interesting read.
With this study you can reasonably assess that you need to figue in repeat visits into the cost of your PPC advertising campaign in order to calculate the ROI properly. You need to know (on average) how many times the same user will come back to your site before they buy?
Hopefully you're tracking ROI over a long period of time. Being able to track repeat visitors is extremely important to developing a good snapshot of your true ROI, among other things that should be figured in.
| 8:50 pm on May 13, 2005 (gmt 0)|
|It sounds like your expecting one conversion to equal one sale. That's usually never the case. As a matter of fact a lot of recent research shows that users don't usually make online buying decisions for weeks. |
Of course. But this doesn't invalidate my point. If someone's not getting the type of conversions they want, as compared with what they get at other SEs, etc., I wouldn't expect them to continue advertising on the low-conversion medium.
|Staying on that point, the most you'll get charged for from the same IP is 1 click per day. Repeated clicks on the same listing more than once in 24 hours are delivered but not billed. |
Leaving aside the matter of people who have dynamic IPs and/or their access is mediated via multiple proxies (each having separate IPs), in the end, a non-converting clicker might as well be a fraudulent clicker, as far as the goal the advertiser had in mind (to eventually make some money).
| 2:14 am on May 15, 2005 (gmt 0)|
If someone's not getting the type of conversions they want, as compared with what they get at other SE's.... then they need to fix the campaign.
We allow the advertiser to control the source of origin for their traffic. That means that if the advertiser is not tracking ROI and the campaign is performing poorly they they need to look at the info we're providing and adjust their campaign accordingly. If a particular source does not work you shut them down leaving the ones that do convert alone. That way you do not have to continue to try and convert on a low converting medium.
I think you know what I mean here. Still, where I differ is this, if BlowSearch can get to the point where we are absolutely sure the quality of traffic we're sending is a) verifiably real users and; b) the advertiser can control the parts of their campaign that don't work..... the onus falls on the advertiser to fix the problem with their site, campaign, etc. There's no excuses anymore.
Remember that no two networks are the same. We all have some different traffic sources and some that overlap. What works well for you with one network may not work on another. Comparing the performance of one network to another is usually a futile effort. It may take different ad copy and porromotional material altogether to create a successful campaign. Throwing a Google campaign into a search advertising account is a sure loser in my book. It might do okay but it could do a hell of a lot better if you took advantage of the extra ad copy space available in search ads.
Every engine has certain ways they market themselves. One engine may get more targeted searches for Travel than say, batteries. So with that in mind traffic selection allows you to keep what does work even though it may reduce overall click volume. Besides it's not about clicks it's about conversions.
"in the end, a non-converting clicker might as well be a fraudulent clicker" This is where we disagree completely. There is inherent value to every click you get (a long as it is a valid click). If someone is not buying there is a reason for that. I read Jeff and Bryan Eisenberg's site a lot and they make very valid points in regard to this.
If someone clicks that means they had interest because it was your ad that was SUPPOSED to pre-qualify them before clicking. If they wander through and leave it means that they lost interest, aren't buying right now, or bought somewhere else. Regardless of that if your site made an impact on them then they may come back later to buy because they remember your brand.
Maximizing every click for conversion is what search marketing is all about. If users are clicking and not buying there's something up with your copy or your site that needs looking into. As long as you're tracking sources and ROI over a long period of time to see that.
| This 86 message thread spans 3 pages: < < 86 ( 1 2  ) |