Forum Moderators: DixonJones
We use a JavaScript tag-based web analytics solution for visitor tracking AND we also get campaign reporting from our online agency, who uses DART. In theory, the number of click-throughs on our ad banners which drive traffic to our homepage should be roughly the same as the campaign ID referrals for that specific banner which are tracked by our web analytics solution. In reality, the differences in the two reports is quite significant - in some cases greater than 50%!
Has anyone else experienced this issue?
1. Is it likely that people really are clicking on the banners, but perhaps they are closing the browser before our homepage has fully loaded?
2. Is it possible that the campaign ID information in the URL is being lost in transit, and therefore not picked up by our JavaScript tags?
3. Is it possible that all of this traffic really is arriving at our site, but is not being picked up properly by the JavaScript tags for some reason?
4. Could it be a dial-up vs. broadband download time issue?
Any help or guidance on this issue would be greatly appreciated. Happy to provide more information if needed.
Thanks,
mph88888
Welcome to Webmaster World!
Interesting dilemma you pose. A very common one, with many possible answers and reasons why.
I'll have a stab at answering a few from my experience.
Your question 1. the answer is probably unlikely, although some of the ad serving technology can be hit and miss
Often only one set of parameters is passed through, and as such a large number can be lost. This is typically the case where there are numerous snippets of affiliate tracking code. Once the first tracker snags the visitor it doesn't make the information available for others.
Other possible anomalies. Page tagging tends to handle proxy visitors and direct type ins better than some analytics tools, or it might be robots with session ID's counting one visit in one place, but each visit in another.
I'd have a wild guess here but I bet the traffic data from the agency is a lot more than the analytics......
There will always be discrepancies when you try to compare two sets of seperately acquired data. My advice would be to stick with one set and then use that for all your comparisons, at least there will be consistency in the preparation and you can work out your conversion metrics on that basis.
If the sample is significant and your agency is wanting to charge you more, then you are quite within your rights to ask for some clarification on the arrival of the figures. That beign said most of the ad server providers have audited techniques, so it may be you have old pages indexed that haven't been tagged.
I can't be sure that the problem at the DART end still is an issue today. But I'll still go try to look up the old emails and see if I can remember what we found out.
1) DART and other like systems (ATLAS DMT, Bluestreak) track clicks on ad banners using their image pixel tag. These are clicks on the ads only - with NO guarantee that the user actually arrives at the destination site.
2) These ad banner clicks open up a new browser with the destination website, which several users likely close before the page has fully loaded. In this case, the Javascript tag will not execute and no page view will be recorded.
3) In the case of dial-up users, it's quite likely that the landing page of the destination site (usually the homepage) has not fully loaded before the user continues to another page in the site. In this case, the Javascript tag may not execute.
4) The location of the Javascript tag is important as well. If the tag is at the top of the page, it's got a much higher likelihood of recording the page view as it will execute immediately. If it's at the bottom of the page (which is quite common), the page view may not be recorded, particularly for slow connection speeds or heavy homepages where users get impatient and click to another page.
Any of the above make sense?
mph88888
Have you compared stats on the number of views of the destination page comparing stats from the your JS tag to stats from your logs? If your theory works, then the log stats should resemble DART's and be a lot higher than your JS numbers.
Any of the above make sense?
Makes perfect sense.
They do say that 82% of statistics are wrong......
I'd suggest that your best shot is to try to work with one set of figures that give you the closest feeling to what you know to be true.
>> I believe it is embedded in the ad banner - an invisible 1x1 pixel tag. It gets called immediately upon a visitor clicking on the banner.
Have you compared stats on the number of views of the destination page comparing stats from the your JS tag to stats from your logs? If your theory works, then the log stats should resemble DART's and be a lot higher than your JS numbers.
>> Good suggestion, however the accuracy of web log file data is another issue altogether that I'd rather avoid for now!
