Webmaster Forum Rules | Posting Guide | Contact Us | Testimonials | Contributing Geek Program | Advertise on Geek/Talk
Welcome to the GeekTalk Webmaster Discussion Forums from GeekVillage.com

Click Here To Register. It's Free!

Go Back   geek/talk: Signature-free discourse for serious web publishers > YOUR REVENUE: Making Money On The Internet > Making Money with CPC and/or CPM Programs
User Name
Password
Register FAQ Calendar Search Today's Posts Mark Forums Read

Reply
 
Thread Tools Display Modes
Old 02-27-2000, 11:32 AM   #1
Michael Epperson
Registered User
 
Join Date: Feb 2000
Location: Chicago, IL
Posts: 7
Angry Burst Stats Always 40% Lower Than Log Stats

My site has Burst rich-media "cache-busting" adcode on every page. Yet my daily Burst "Total Page Views" stats are always 40%-50% lower than the total page view stats reported by Webtrends. I'm not talking about the number of non-paying default impressions; I'm talking about the Burst "total page view" stat which, it seems to me, should be very close to the total page views reported in my logfile, since there's adcode on every single page.

Does anyone know why this might be? It's driving me crazy. On the Burst ad stats page, one will find the following in small print: "This program is being served by a 3rd party other than BURST! Due to such
reasons as caching, ad-serving technology differences, Internet sluggishness..etc., each firms numbers will differ from others in the industry. We have found that click numbers are usually quite close, but impression values may differ as much as 40%!"

That 40% figure appears to be accurate in my case. Anyone else seeing similar results? If this is the reason for the discrepancy, the implication is that 40% of my page views result in blank ad banners, because of "internet sluggishness," etc?

Perhaps it's a caching problem, and the cache-busting code is ineffective? The only other explanation would be that 40-50% of my visitors have their graphics disabled, which is unlikely.

Anyone have any ideas?

Thanks very much,

Michael Epperson
Michael Epperson is offline   Reply With Quote
Old 02-27-2000, 09:11 PM   #2
xor
Registered User
 
Join Date: Feb 2000
Posts: 29
Question

What are your page view settings on web trends? I know that with webtrends you can set different files to count towards page views. For example you could count all htm, .html, .shtml as page views. Perhaps a .gif or .cgi got put in there by accident.

If you have cache busting code on every page you shouldn't see that big a difference. Perhaps 10% but not much more.

I'm pretty sure that the third party 40% difference thing is only taken into effect after campaigns are audited and burst results are compared with third party results.
xor is offline   Reply With Quote
Old 02-27-2000, 10:06 PM   #3
gmiller
Registered User
 
Join Date: Jan 2000
Location: McLemoresville, TN 38235
Posts: 138
Cool

Some people use text-only browsers like Lynx. Others use graphical browsers but turn off graphics files. Add those two factors to poor performance on the Internet, and the worst case scenario could easily be 40%, as BURST! indicates.
gmiller is offline   Reply With Quote
Old 02-28-2000, 05:23 AM   #4
fairhousing
Member.
 
Join Date: Apr 1999
Location: birmingham, alabama
Posts: 1,544
Talking

that sounds like the problem to me, if not interesting.

------------------

fairhousing is offline   Reply With Quote
Old 02-28-2000, 09:21 AM   #5
Michael Epperson
Registered User
 
Join Date: Feb 2000
Location: Chicago, IL
Posts: 7
Talking

Thanks for the input everyone; I really appreciate it. Like I said, this had been driving me nuts.

XOR, thanks for the tip: I double-checked the Webtrends settings, and configured the program to recognize only html pages, since that's all I have running on my site--no htm, shtml, cgi, etc. It didn't seem to make any difference. Log stats for yesterday, ~2200 page views; Burst stats for yesterday, ~1100 total page views (again, both paying and default). Argh!!!

Gmiller: I had considered the graphics-disabled visitors possibility, and Burst offers the same caviat in their FAQ; I just couldn't believe it! As you can probably tell, I'm kind of new to the webmaster scene, so I don't have the experience to know whether or not it's typical that 30% or 40% of visitors will have graphics disabled or use Lynx. It seems like a really high number to me, but that's a totally uninformed opinion.

Are any other Burst people seeing this same level of variance between log stats and Burst stats? What about other companies--Flycast, Advertising.com, Datais, etc...

Thanks again for all the input!

Michael
Michael Epperson is offline   Reply With Quote
Old 02-28-2000, 01:03 PM   #6
MagicMan
Registered User
 
Join Date: Feb 2000
Location: Seattle, WA
Posts: 124
Exclamation

Michael,
Although we don't use BURST, we too are having a similiar problems with comparison of stats. ADSDAQ stats are 70% different from our raw stats, while Advertising.com stats are about 40% different. I can understand a 10% variance, but this is beyond comprehension. In reading Advertising.com's FAQ's, we found that each page has to have a unique identifier on it. We have over 1600 pages on our site, so doing a simple search and replace would not work since each page has to have this identifier. They have a random generator code that is supposed to generate new code for each page, but we have found it didn't work properly, so we are having to do each page, one by one. This has brought the variance down, and everyday it gets smaller. ADSDAQ on the other hand I have no clue, and they keep telling me they are looking into it. The only reason we have kept ADSDAQ is the CTR has been outstanding(until the end of month, as their inventory has dwindled to the same 5 advertisers) at over 5%. Advertising.com's CTR is currently around 2.3%. SO, having said all that, Advertising.com numbers are getting into line, and we will have the code problem solved by tomorrow, and look forward to a strong March. On the Adsdaq front, at this point we are not sure which way to go. We are going to miss the bonus plateau of 500,000 page views by less than 70,000, which would have doubled the payout to us. Just seems very strange that it has worked out this way this month.

Les

MagicMan is offline   Reply With Quote
Old 02-28-2000, 05:42 PM   #7
KP
Registered User
 
Join Date: Nov 1999
Location: USA
Posts: 33
Cool

Michael, just FYI, Burst consistently shows my pageviews at less than 40% of what they should be. I use Flycast, which defaults to Burst. My Flycast stats will say 70,000 defaulted to Burst for that day, but Burst says it only got 40,000 or so.

I check my site often and do not see blank banners. So I know that the default is getting through and not being lost in the http redirect. I've never questioned Burst about it - I guess I figured it wouldn't do much good.

KP is offline   Reply With Quote
Old 02-28-2000, 05:43 PM   #8
KP
Registered User
 
Join Date: Nov 1999
Location: USA
Posts: 33
Cool

I forgot to add that I do use cache busting on all my ads.

KP is offline   Reply With Quote
Old 02-28-2000, 06:43 PM   #9
Michael Epperson
Registered User
 
Join Date: Feb 2000
Location: Chicago, IL
Posts: 7
Cool

Well, given the discussion so far, it seems that these kinds of discrepancies are par for the course for Burst--and, according to Les, for other ad programs as well. And I agree with you, Les--a 10% variance would be acceptable, but 40% and over... It's more than disappointing.

I have written to Burst support MANY times about this, providing them with detailed numbers and several days' worth of stats. Here are some excerpts from the responses I've received:

"We have had this inquiry numerous times, and are always eventually able to isolate the variable causing the discrepancy.

"As you are aware, we only count the incidence of our banners appearing. We
do not count page views... Often times, many trackers will record the incidences of other graphics on a page.

"If there was a time where either our servers did not serve ads or another
link in the delivery chain temporarily failed, then your site's pages would
load while we were not recording banner impressions...

"I can say, with full confidence, that we have never had an incidence where
we found our servers to be incorrectly tracking banner impressions.The cause of discrepancy is likely because your server logs are counting impressions of items other than banners.

"We do not track page views to your site. We only track the incidence of ad
impressions we serve.

"I suggest you contact Webtrends to find out exactly what is counted in their
"Total Page Views" figure.

"It is very important to us that you are satisfied with our service and that
we earn your trust."

The fact is, I know that my logfile stats are accurate; I have correlated the Webtrends page view count with the raw logfile, and Webtrends is configured to only register html and shtml files as page views. So I know that's not the problem. As for visitors with graphics disabled--maybe, but not 40% and over. I just don't buy it. That leaves me with the "internet sluggishness" line, meaning that 40%-50% of the time, my pages are loading blank banners. I check my site often throughout the day and, like KP, have not seen evidence of so many unloaded banners.

Perhaps the "cache-busting" code just isn't working; that's all I'm left with. Either that, or the stats are "fixed" and we're delivering more impressions than we're getting paid for. I don't know about the rest of you, but I'm not inclined to embrace that conclusion just yet.

I do have to wonder, though: Providing accurate page view/ad impression stats SHOULDN'T be that difficult, should it? I can understand an occasional hiccup causing a 10% discrepancy, but 40% - 50% EVERY DAY? Forget the advertised CPM rates. Divide by two if you want the real rates.

Realistically, though, there's not much I can do other than discontinue serving ads because I'm not making what I SHOULD be making--and as a result, make nothing at all!

At the very least Burst should move its "40% fine print" to the ad select page--either that, or adjust the advertised CPM rates to more accurately and honestly reflect the ACTUAL rates, because they're clearly NOT what they're advertised as being. And the worst thing is that it just isn't clear what the true source of all these lost page views is. 40% - 50% daily??? It's like an Internet Bermuda Triangle.

Very frustrating.
Michael Epperson is offline   Reply With Quote
Old 02-28-2000, 10:58 PM   #10
gallery
Registered User
 
Join Date: Aug 1999
Location: Belgium
Posts: 691
Cool

Hi,

I also do not believe that 40 % of visitors have disabled graphics loading in their browser.
I know also that there are software that only disable ad banners.
I know quite a lot of people on the net and none of them are disabling grahics.

I am using imagenetworks as CPM advertiser.
I usually get 70,000 impressions for their banners daily.
Yesterday, I got only 45,000 which is strange given the number of visitors I got that day.
The problem should come from imagenetwork...

------------------
www.pics4fun.com : a huge collection of funny pictures, cartoons and comics.
webmaster@pics4fun.com
gallery is offline   Reply With Quote
Old 02-29-2000, 12:25 AM   #11
Arie
Registered User
 
Join Date: Nov 1999
Location: Malta, Europe
Posts: 292
Cool

I use banner rotation software, that enables me to use IP tracking, so that it will only count banner views from unique IP's in a 12 hour period.

If I use that option, I count a little less banners than Burst does......



------------------
Regards,

Arie Slob,
InfiniSource, Inc.
http://www.infinisource.com
Internet & Windows Resources
Arie is offline   Reply With Quote
Old 02-29-2000, 12:45 AM   #12
xor
Registered User
 
Join Date: Feb 2000
Posts: 29
Cool

I'd also try and use a couple other stat programs and see what they report for page views. Then take all your findings to Burst and ask them what the problem is.
xor is offline   Reply With Quote
Old 02-29-2000, 02:15 AM   #13
MagicMan
Registered User
 
Join Date: Feb 2000
Location: Seattle, WA
Posts: 124
Exclamation

Michael,
We have gone as far as hiring an Independent Auditor to audit both our stats and our server stats. The audit actually proved that our server stats were lower then what is being reported to us. We provided the company with that data, and still have recieved no response. The thing I find interesting in the reply to you is about their site being down and views not counted. If they know their site is down, at the end of the month adjust your numbers to reflect the outage time, ie divide your views by total # of hours in the month and get your average per hour, and then multiply that by the down time. It's not a perfect science, but would get everyone a lot closer to the truer numbers.
Les
MagicMan is offline   Reply With Quote
Old 02-29-2000, 03:44 AM   #14
consent
I am a Contributing Geek. Are You?
 
Join Date: Nov 1999
Posts: 249
Unhappy

I would think it might also be the fact that a banner takes time to load (perhaps this possibility has been covered by the term "internet sluggishness").

If you have a page of non-graphics, your text links will load quickly, and a visitor (especially a regular visitor, who knows where he wants to click) will immediately explore a link before the banner even appears.
consent is offline   Reply With Quote
Old 02-29-2000, 11:02 PM   #15
gmiller
Registered User
 
Join Date: Jan 2000
Location: McLemoresville, TN 38235
Posts: 138
Cool

It seems like everyone is looking for one single problem that causes you to lose huge numbers of impressions. I don't think that's how it is. As I said before, I think some people are using text-based browsers (I do it fairly often, the speed difference between Lynx and Netscape or MSIE is incredible), disabling graphics, using banner blockers, losing the banner ad due to redirect times, getting broken images because of technical problems somewhere between the ad server and their own computers, etc.

There are so many things that can go wrong that it's not very surprising to get low numbers. I doubt any of the major ad networks are cheating, and BURST! makes available a list of every IP address to which it has served ads each day. Every time I've successfully viewed an ad and then checked the Analog stats the next day, my ad view was accounted for.
gmiller is offline   Reply With Quote
Reply

Bookmarks

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
I used to like to check my Burst stats TJ Making Money with CPC and/or CPM Programs 5 12-16-2000 04:17 AM
funky burst stats??? serrow Making Money with CPC and/or CPM Programs 4 02-04-2000 09:41 AM
stats down for burst and fly serrow Making Money with CPC and/or CPM Programs 3 01-02-2000 03:03 PM
trouble with web sponsor stats log on fairhousing Making Money with CPC and/or CPM Programs 5 10-28-1999 01:39 PM
Is it just me or is everyone else's cybergold stats much lower?? Ericy Making Money with CPC and/or CPM Programs 5 04-18-1999 08:19 AM

Please support our advertisers. They ensure our survival.

All times are GMT -5. The time now is 01:13 PM.


GeekVillage.com is copyright © 1998-2015 Curiosity Cave - Science gifts for clever kids. All rights reserved.
Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.