Jump to content

  • Log in with Facebook Log in with Twitter Log In with Google      Sign In   
  • Create Account

Subscribe to HRA Now!

 



Are you a Google Analytics enthusiast?

Share and download Custom Google Analytics Reports, dashboards and advanced segments--for FREE! 

 



 

 www.CustomReportSharing.com 

From the folks who brought you High Rankings!



Photo

David Vs Goliath Using Data Logs As The Slingshot


  • Please log in to reply
22 replies to this topic

#1 Orpheus Descending

Orpheus Descending

    HR 3

  • Active Members
  • PipPipPip
  • 90 posts
  • Location:Ranelagh, County Dublin, Ireland

Posted 14 May 2011 - 12:10 PM

I'm working with a very small client in the UK who was sold a SEM package by one of the big telecoms in the UK. The package cost over 20,000 and was guaranteed to drive 20,000 visits (the contract states visits not clicks) to the client site. I'm not here to complain about the horrendous way this company sells paid search. I need to understand how to track it.

The telecom has a client centre which shows very top line data based on Yahoo and Google and how many clicks were driven by each.

In March, after making changes to the keywords, ad groups and ad copy for this client, their traffic fell off a cliff. When I looked at their third party client centre, their traffic had gone from 200+ to 17 clicks per week. (This was due to properly targeting keywords that cost more but which are much more appropriate to the client's products). When I contacted the company, they said their data is always a few days behind and would be updated - and within 24 hours their traffic showed 120 more clicks over the past 5 days - but (of course) the GA stats hadn't changed. From what I could tell, they manipulated their data.

So I started to look deeper. As it turns out, GA is reporting only 11% of the Yahoo traffic and only 45% of the Google tracking this company says they've driven. When I contacted the company, they were adamant that their traffic is not picked up by GA in anyway. I can see that GA is picking up the traffic from this third party (and no they are not properly linked or tracked via GA, they won't allow that) and all of the clients sales records tie beautifully to GA (which means that if this company did actually serve those 20,000 visits absolutely none of them converted - GA has tracked 24,000 visits and has a conversion rate of 1.73% so it would be odd that all of the traffic driven by the telecoms through paid search has no conversions) but to get beyond the GA debate, I've gotten the Web site's Data Server Logs.

I'm a wee bit out of my depth here. For March 1, the data server logs show 5 unique IP addresses which came from Yahoo. This company is showing 19 Yahoo visits. GA is showing 2 Yahoo visits. There are; however, 16 visits from one IP address over a 24 hour period showing activity just after midnight, 2:15, 3:10, 6:19, 7:11, 8;06, 9:28, 9:42, 13:59, 14:29, 16:53, 17:37, 18:38, 19:14, 21:10, 22:00. On March 2, this same IP address comes up for 8 visits - the third party is showing 20 visits (of which I can only find 10 visits driven from Yahoo of any kind) and GA is showing 0 Yahoo visits.

Also, I'm not finding any third party tracking code in the URL string so I can't tell how this is routed through the third party servers.

I have a conference call on May 23 with the client and the third party and it would be great if I could present this information with some confidence. I'm also happy to take this offline as I want to be careful with this. Any thoughts or ideas would be very welcome.

#2 Jill

Jill

    Recovering SEO

  • Admin
  • 32,940 posts

Posted 14 May 2011 - 04:52 PM

I think you can present confidently that this company is ripping off your client.

#3 Orpheus Descending

Orpheus Descending

    HR 3

  • Active Members
  • PipPipPip
  • 90 posts
  • Location:Ranelagh, County Dublin, Ireland

Posted 15 May 2011 - 08:45 AM

QUOTE(Jill @ May 14 2011, 10:52 PM) View Post
I think you can present confidently that this company is ripping off your client.


First of all, it's been a very long time since I've been here, so 'hello,' Jill. It's nice to 'see' you.

I've done a bit more digging and there are 3 IP Addresses that repeat every day for Yahoo traffic. One of those IP addresses is a Yahoo! IP in Sunnyvale (this is a UK account). Can I safely say that when my client wasn't getting the traffic BT guaranteed, that BT somehow set up an automated 'something' to make it look like this client was getting traffic? [After the first year, BT hadn't gotten the guaranteed visits for this client so they set up for another year. We are now at the end of the second year.] Is it possible that the changes I made to the account, trip this mechanism up and that's why the traffic dropped off the cliff?

I haven't found the same issue with Google yet, but there is a problem with that traffic too when comparing BT stats to GA's. What I do find interesting is that Google visits require about 30 lines of data retrieval whereas the Yahoo ones are only about 4 lines of data for each. Can that mean the whatever mechanism this is hits the paid link and then aborts the function before the page is fully loaded each time? Is that also why GA isn't recording the fraudulent clicks? Or does GA detect the fraud (like AdWords) and remove the stats? GA is picking up actual visits to the site from Yahoo - but it looks like that traffic is only organic. I'm assuming the data server accurately specifies the medium as paid or organic?

Since the next step is legal action, I'd also like to know if there are any law firms out there that deal with this kind of case? Considering the complexity of the data and the nature of the allegations, I'd like to find someone who knows about the world of online tracking. Or would the next step be to hire an expert to take a second look at the server logs or would the Web developer be able to help us look at the data logs? I suppose I can't quite believe that this is happening so I'm assuming I've overlooked something or that I don't know what to look for and am therefore finding issues that aren't real.

This is a very small business in the UK that is about to go under because of this and I'm helping them pro-bono because they've been getting screwed for the last 2 years and it should not be this way. Any help I can get for them will be greatly appreciated by me and by them.

#4 rolf

rolf

    HR 6

  • Active Members
  • PipPipPipPipPipPip
  • 675 posts
  • Location:Suffolk UK

Posted 16 May 2011 - 03:08 AM

Without looking into the ins and outs of the situation, it sees more than a little suspect to me. If all is as it appears, and you do seem to have a handle on how it is, then I would definitely be advising the client to part company with the 'service provider'.

I think the client also has grounds to sue for a LOT of money, however, first you would need to convince the client of this and get them to understand how the scam has worked. Then they would need to engage a lawyer who also understands this stuff, which is not going to be cheap. Then they need to go to court and bank on the idea that a judge etc. will understand what is being alleged, which will not be a quick or easy process, especially if the 'service provided' decides to throw money at fighting, rather than throwing money at your client until they go away. Really, a big and expensive gamble if they are already struggling.

The long and short of it, IMHO, is that your client has been screwed and unfortunately doing anything other than walking away is probably throwing good money after bad. Sorry if that's a bit brutal or negative, just trying to be practical.

#5 Orpheus Descending

Orpheus Descending

    HR 3

  • Active Members
  • PipPipPip
  • 90 posts
  • Location:Ranelagh, County Dublin, Ireland

Posted 16 May 2011 - 04:18 AM

QUOTE(rolf @ May 16 2011, 09:08 AM) View Post
Without looking into the ins and outs of the situation, it sees more than a little suspect to me. If all is as it appears, and you do seem to have a handle on how it is, then I would definitely be advising the client to part company with the 'service provider'.

I think the client also has grounds to sue for a LOT of money, however, first you would need to convince the client of this and get them to understand how the scam has worked. Then they would need to engage a lawyer who also understands this stuff, which is not going to be cheap. Then they need to go to court and bank on the idea that a judge etc. will understand what is being alleged, which will not be a quick or easy process, especially if the 'service provided' decides to throw money at fighting, rather than throwing money at your client until they go away. Really, a big and expensive gamble if they are already struggling.

The long and short of it, IMHO, is that your client has been screwed and unfortunately doing anything other than walking away is probably throwing good money after bad. Sorry if that's a bit brutal or negative, just trying to be practical.


Thanks, Rolf. I absolutely agree with you and have just been on to the client about it. Our conference call was moved up to this afternoon and my guy is up for the challenge. There are internet governing bodies here, too, which might be the more prudent way to go. If these guys are doing it to one, why not all? For some reason, I'm really shaken by this and I do understand the risks.

I just want to know that what I'm seeing is real. Could there be a salient piece of data I'm overlooking that explains all of this? If anyone reading this can think of something I might have over looked, I would be grateful to hear your thoughts.

#6 rolf

rolf

    HR 6

  • Active Members
  • PipPipPipPipPipPip
  • 675 posts
  • Location:Suffolk UK

Posted 16 May 2011 - 04:46 AM

QUOTE
There are internet governing bodies here, too, which might be the more prudent way to go. If these guys are doing it to one, why not all?


I agree, they are probably doing it to a lot of people, so if you think you can bring it to light then that is definitely a worthy thing.

Which governing bodies do you have in mind?

QUOTE
I just want to know that what I'm seeing is real. Could there be a salient piece of data I'm overlooking that explains all of this?


I know what you mean, nobody wants to be made a fool of, but from what you have said there is a genuine discrepancy and therefore a legitimate cause for concern. As such I think you have every right to demand clarity and accountability.

There is no reason I can see why the raw server logs, GA reports and their own data should not tally in at least a percentage based broad strokes fashion, and if their data is the odd one out then the onus is on them to explain why.

#7 Orpheus Descending

Orpheus Descending

    HR 3

  • Active Members
  • PipPipPip
  • 90 posts
  • Location:Ranelagh, County Dublin, Ireland

Posted 16 May 2011 - 05:34 AM

QUOTE(rolf @ May 16 2011, 10:46 AM) View Post
There is no reason I can see why the raw server logs, GA reports and their own data should not tally in at least a percentage based broad strokes fashion, and if their data is the odd one out then the onus is on them to explain why.


If I'm only looking at the odd data out, then GA is wrong. The problem is that the data server logs that match BT's client centre show the same 3 IP's repeatedly day after day. One IP in particular has 16 visits in one day and along with the other 2 repeat IP's which make up the 19 visits that BT are counting. GA only counts 2 visits from Yahoo! that day. The next day the same IP address is counted for 10 of the 20 visits counted by BT. The day after that the same IP is counted for 10 of the 13 visits - all the same IP address. Would there be any reason why Yahoo! would route multiple users through a single IP address? And would that come up on the data server logs from the same browser and the same referral string time after time after time and could that still somehow be legitimate traffic?

The client is looking into filing a complaint with the SFO (Serious Fraud Office). What makes this a little more tricky is that BT outsources to Webvisible in the States. Not sure how all of the various jurisdictions will work.



#8 rolf

rolf

    HR 6

  • Active Members
  • PipPipPipPipPipPip
  • 675 posts
  • Location:Suffolk UK

Posted 16 May 2011 - 06:55 AM

QUOTE
Would there be any reason why Yahoo! would route multiple users through a single IP address?


Not that I can think of

QUOTE
And would that come up on the data server logs from the same browser and the same referral string time after time after time and could that still somehow be legitimate traffic?


Not as far as I can see

Something definitely seems fishy to me fish.png

#9 Orpheus Descending

Orpheus Descending

    HR 3

  • Active Members
  • PipPipPip
  • 90 posts
  • Location:Ranelagh, County Dublin, Ireland

Posted 16 May 2011 - 07:03 AM

QUOTE(rolf @ May 16 2011, 12:55 PM) View Post
Not that I can think of
Not as far as I can see

Something definitely seems fishy to me fish.png


Thank you, very much.

#10 scouseflip

scouseflip

    HR 4

  • Active Members
  • PipPipPipPip
  • 201 posts
  • Location:North West, England

Posted 16 May 2011 - 08:24 AM

Saw this thread on PC Advisor - ht tp://www.pcadvisor.co.uk/forums/25/business/350251/bt-web-clicks/ - sounds like your client is not the only unhappy customer... but at least one of them got their money back.

#11 Orpheus Descending

Orpheus Descending

    HR 3

  • Active Members
  • PipPipPip
  • 90 posts
  • Location:Ranelagh, County Dublin, Ireland

Posted 16 May 2011 - 08:59 AM

QUOTE(scouseflip @ May 16 2011, 02:24 PM) View Post
Saw this thread on PC Advisor - ht tp://www.pcadvisor.co.uk/forums/25/business/350251/bt-web-clicks/ - sounds like your client is not the only unhappy customer... but at least one of them got their money back.


Thank you, very much. OffCom might be the ticket and it's reassuring to know this guy is not the only one (terrible but reassuring).

#12 Michael Martinez

Michael Martinez

    HR 10

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 5,094 posts
  • Location:Georgia

Posted 16 May 2011 - 12:43 PM

QUOTE(Orpheus Descending @ May 14 2011, 10:10 AM) View Post
I'm a wee bit out of my depth here. For March 1, the data server logs show 5 unique IP addresses which came from Yahoo. This company is showing 19 Yahoo visits. GA is showing 2 Yahoo visits. There are; however, 16 visits from one IP address over a 24 hour period showing activity just after midnight, 2:15, 3:10, 6:19, 7:11, 8;06, 9:28, 9:42, 13:59, 14:29, 16:53, 17:37, 18:38, 19:14, 21:10, 22:00. On March 2, this same IP address comes up for 8 visits - the third party is showing 20 visits (of which I can only find 10 visits driven from Yahoo of any kind) and GA is showing 0 Yahoo visits.



QUOTE(Orpheus Descending @ May 15 2011, 06:45 AM) View Post
I've done a bit more digging and there are 3 IP Addresses that repeat every day for Yahoo traffic. One of those IP addresses is a Yahoo! IP in Sunnyvale (this is a UK account). Can I safely say that when my client wasn't getting the traffic BT guaranteed, that BT somehow set up an automated 'something' to make it look like this client was getting traffic? [After the first year, BT hadn't gotten the guaranteed visits for this client so they set up for another year. We are now at the end of the second year.] Is it possible that the changes I made to the account, trip this mechanism up and that's why the traffic dropped off the cliff?


You're chasing phantoms.

Yahoo! IP addresses won't have anything to do with traffic that the Telecom is or is not sending to your client.

You have to look at the referral entries in the raw log data -- and you'll be lucky if that data is being captured. Many Web hosting ISPs set up Apache to NOT capture referral data by default. Don't ask me why. I just know I have had to turn it on more often than it has been there. Maybe that's just a string of bad luck for me.

Based on what you describe here (and it could be that we are using terminology differently), you have not been looking at the right data to check the Telecom's claims.

#13 Jill

Jill

    Recovering SEO

  • Admin
  • 32,940 posts

Posted 16 May 2011 - 03:02 PM

Might they be counting spider visits? (GA doesn't) and they shouldn't be either.

#14 Orpheus Descending

Orpheus Descending

    HR 3

  • Active Members
  • PipPipPip
  • 90 posts
  • Location:Ranelagh, County Dublin, Ireland

Posted 16 May 2011 - 04:11 PM

QUOTE
Might they be counting spider visits? (GA doesn't) and they shouldn't be either.


Jill, I was certain they were not counting bots or spiders of any kind. I could see those quite clearly in the server logs and I tracked them and Google/Bing/Baidu, etc were not showing up in BT's results.

But I just looked up the referral URL in the string I keep seeing repeated and it looks like it is a Yahoo! crawler.

Here is one line of data which has been altered to remove identifiable data:

2011-03-19 02:46:13 XXXXXXXXXXX 11.111.111.111 GET /links.asp - 80 - 22.222.22.222 HTTP/1.0 Mozilla/5.0+(compatible;+Yahoo!+Slurp;+http://help.yahoo.com/help/us/ysearch/slurp) - - 200 0 2953 362 562

Is this a Yahoo! issue? I'm not sure what this means - but I'm dizzy from looking at data for the last 10 hours.

#15 Michael Martinez

Michael Martinez

    HR 10

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 5,094 posts
  • Location:Georgia

Posted 16 May 2011 - 05:38 PM

QUOTE(Orpheus Descending @ May 16 2011, 02:11 PM) View Post
Here is one line of data which has been altered to remove identifiable data:

2011-03-19 02:46:13 XXXXXXXXXXX 11.111.111.111 GET /links.asp - 80 - 22.222.22.222 HTTP/1.0 Mozilla/5.0+(compatible;+Yahoo!+Slurp;+http://help.yahoo.com/help/us/ysearch/slurp) - - 200 0 2953 362 562

Is this a Yahoo! issue? I'm not sure what this means - but I'm dizzy from looking at data for the last 10 hours.


That's just Yahoo! crawling the site. They are still running their own search services for a few overseas/non-English markets and for Yahoo! Site Explorer.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

SPAM FREE FORUM!
 
If you are just registering to spam,
don't bother. You will be wasting your
time as your spam will never see the
light of day!