About Us

Bulletpoint StarImulus® is a technology focused design + interactive agency.

In addition to the services we provide our clients we also have several products in the works. Our office is always filled with chatter and this blog is an outlet for some of our creative energy, rants and ideas.

Search

Previous Posts

Imulus Twitters

Featured Project

Featured Project visit website Client: Rocky Mountain Institute

Latest Quick Links

Aug25

How Accurate Are Quantcast Estimates?

Based on the few examples I’ve looked at, not very!

Site A: Enterprise Class / Fortune 1000 Web Site
This comparison shows Quantcast ESTIMATES in green, versus the ACTUALS in WebTrends. Both measures track “VISITS”. This is important because the default for Quantcast is “PEOPLE”. I still haven’t figured out the difference between the two. The variation between the two sets of data is significant and the traffic pattern seems inverted.

Site B: Regional Business Information Site
This comparison is between Quantcast ACTUALS in green and Google Analytics ACTUALS in orange. Again, we are tracking Google Unique Visitors against Quantcast Visits. Another important note is that when actual data is pulled in Quantcast you’ll see separate tracking for US vs Global. Since Google doesn’t distinguish, we’ll focus on the global on both.


Site C: National Ecommerce Site
In this example Quantcast faired to be a bit more accurate. I compared Quantcast People (ESTIMATED) with Google Absolute Unique Visitors and the results were a bit more accurate.


Trend Overview
These graphs below were interesting to me because of the traffic discrepancy despite both sites utilizing tracking codes. Both of these graphs are pulled from Site B. Google’s Max is 206 VISITORS per day, whereas Quantcast is showing about 330 VISITORS.

Google Analytics

Quantcast Visits

Summary
I think Quantcast is an excellent tool; however you should proceed with caution when using the analytics obtained from both it’s estimates and actuals. It is important to note that both measures are benchmarking Quantcast against WebTrends and Google Analytics; however both of these tools have been used for many years and are fairly trusted. I’m not sure I’m ready to fully trust in Quantcast. But I’m very open to differing opinions on this.

posted in: Google, marketing, web development

This post was published on Monday, August 25, 2008 at 6:30 am

Leave a comment


Comments

1

Chris G

August 25, 2008 at 1:08 pm

I’m very glad to see an example of real versus estimated for these kinds of services.

Last night I posted something similar on Google Trends for Web Sites, and there was nothing accurate about GTfWS’ data at all.

http://gotanalytics.blogspot.com

2

George

August 25, 2008 at 1:45 pm

Hello Chris

Thanks for the link, that is a good post and a nice supplement to this Quantcast data. Oh how I wish our findings were different. It would be excellent to fully believe in this sort of information!

Thanks
George

3

Wendy Ricci

August 27, 2008 at 10:36 pm

George - I’ve noticed this discrepancy with Quantcast as well. However, I think it’s easy to miss the point of a tool like this. It is not meant to replace web tracking tools like WebTrends, Omniture, etc. Sure, marketers and web developers often look to this type of tool for traffic data, but it is meant for media planners (like me!) to help guide advertising decisions.

Now if we could get any two web tracking tools to provide the same traffic info for one site…that would be a miracle!! Thanks for the post.

4

George

August 27, 2008 at 11:26 pm

You are totally correct Wendy.
I agree and have used these tools before for projections and to better understand competitors yet they are pretty far from accurate. Assuming I was planning on buying media on Site B based on Quantcast’s numbers… I’d be seriously disappointed when that ad actually ran. Or in the case of Site A, if I ran an ad based on impressions my impression rate would be much quicker then I forecasted.

You are right, it sure would be great if they were dead on accurate… I guess I assumed that Quantcast was more accurate then I found it to be.

5

Adam G

August 27, 2008 at 11:49 pm

First off, full disclosure - I’m the CMO at Quantcast. It’s great to see discussion about our data, and the questions/comments above. Couple quick points for clarification (I’m happy to go into more depth if anyone is interested - just email me).

The traffic and audience data we provide for sites depends on whether they have “quantified” their property and added our tracking tags. If they have, we supply both:

a) unique cookie counts (which will typically align with web analytics services like GA, Web Trends, and Omniture - if tags for both services are implemented in similar fashion), and

b) people based counts (which are more aligned to the types of audience estimates you get from Comscore and Nielsen

It’s important to note that sites that don’t have our tracking tags implemented (non-quantified sites) have audience data estimated based solely on our 1.5 million person US panel. These audience estimates are *people* estimates, not unique cookie estimates (which do not represent unique people, or even unique machine counts).

There is a substantial difference between unique counts provided by web analytics services like GA, Omniture, and Web Trends (which are cookie counts) and people counts provided by panel services. You can’t/shouldn’t compare them. People counts adjust for cookie deletion, non-acceptance, multiple machine use, and multiple person use per machine. You can read more about this difference at: http://www.quantcast.com/docs/.....e-overview

Without knowing the exact sites you use in the examples above, I’m unable to provide a definitive explanation for the discrepancies you note. But here is a best-guess for some of the posted examples:

For Site A, my suspicion is that the site you are comparing is not a quantified site. As such, you are comparing people data from Quantcast with cookie data from Webtrends. Cookie data will generally always be inflated relative to unique, unduplicated people counts (per the white paper cited above).

With Site B, I’d want to know specifically what data points you are comparing. If it is “unique visitors” from GA, and “visits” from Quantcast, the higher counts from Quantcast make sense - as visits are not unique, while unique visitor counts are. Some unique visitors who are only counted once by GA, and who visit the site multiple times, will be counted multiple times by Quantcast.

With regards to the traffic trending charts - the key question is whether GA tags and Quantcast tags are implemented in similar fashion. Many times, they are not, and this causes discrepancy.

One of the challenges for the industry is recognizing that “uniques” can be many things - people, cookies, machines, etc. It’s critical that you understand the methodology of the measurement service you are using, and others that you may be comparing with.

6

George

August 28, 2008 at 7:26 am

Hello Adam

Thank you so much for clearing up some of the points here. I can take this chat offline and send you the actual sites if you would like to do more research on how we’ve implemented the tags. I’m glad to see the Quantcast is active in explaining it’s differences and thank you for the “Cookie Corrected Audience Data” doc, I couldn’t find that on my own.

Thanks again
George

7

Imulus Insights | Stop “Trying” to Be Social

November 13, 2008 at 8:45 am

[...] an example, I wrote a blog post about Quantcast a few months back. Within 2 days of the post, Adam, the CMO of Quantcast left me a comment. That, [...]