Then they sell extra options for advertisers that allow, for example, measuring the overlap between two or more sites (so, if your aim is to reach max number of uniques, it wont make sense to run the campaign on two river whose traffic overlaps significantly) - with the accuracy of one user. The system produces almost 1:1 data to Google Analytics. Sorta expect to see these elements brought into Compete in future (although I suspect getting river to pay might be impossible as companies arent used to it in States).

-Dan

Compete is wayyyy off.

How is the new gig working out for you? I will be in your neck of the woods next week.

Cheers.

Once you factor out cookie deletion, Compete.com has almost identical counts to quantcast on a quantified site!

I think you need to put this in perspective. Nielsen has struggled for decades to accurately measure TV viewership.

For simplicity lets say there are 1,000 channels available on TV. For anyone that watches a TV show the chances they only viewed the show in one location at a single point in time is close to 100%.

Now lets consider the web… 10M+ unique domains (vs. 1,000 TV stations)… People access the same website from multiple locations (home, work, starbucks, mobile) at multiple time during the same day/week/month. To top it off there is a significant population of users constantly clearing their cookies. As such, there is not a single measurement technique, local or panel based, that will ever be RIGHT. All we will have is data and those with the best data, with the most consistent measuring methodologies will be the most useful.

Quantcast was designed to be an advertising network. When you ‘quantify’ yourself you are simply joining their network. Local JS tracking is good for local reporting, but tells you nothing of who your users are, what they did before they got to your site and what they did after they left your site.

True competitive intelligence is only afforded by panel based services and although panels can only provide estimates, as long as the estimates are based on good data and within the confines of a consistent methodology there is substantial value to be derived… Trust me… just ask one of Compete’s 100+ consulting based clients… or just ask TNS.

Nobody from Compete is commenting on examples there that show how off Compete is. I’ll paste those comments here:

March 1st, 2008 at 4:13 am

This is a smart move. I tried a number of sites and keywords, but find some results surprising and suspicious.

For instance, look at keyword “simpy”:
http://searchanalytics.compete.....tion/simpy

This gives 14,879 avg. monthly search referrals for simpy.com domain.
I run simpy.com and I’m looking at its referral stats right now and see that this number is about an order of magnitude off the actual number.

Here is another one. Technorati happens to be Quantified Publisher:
http://www.quantcast.com/technorati.com
From that we can see TR gets 13-14 monthly PVs.

But this is what Compete SA gives:
http://searchanalytics.compete.....technorati
Keyword “technorati” is 3.71% of its search referrals, that being 797,969 avg. monthly search referrals.
100 / 3.71 = about 26

Thus, the total number of search referrals is 798,000 * 26 = 20,748,000

So this is saying TR gets almost 21M PVs from search referrals alone. But in fact, since Quantcast measures TR traffic directly (i.e. no estimates, that *is* the real traffic), we can see that the PVs are at 13-14M, which means Compete SA numbers are about 50% off!

Can someone from Compete comment on this? Am I making some wrong assumptions anywhere?
I like the service, I like the numbers, but if they are indeed as off as these two examples show, then…

March 1st, 2008 at 4:19 am

Oh, and in my Technorati example in my comment above I forgot to emphasize:

Compete shows almost 21M monthly *search referrals.*
Quantcast shows 13-14M monthly *PVs*. (i.e. not *just* search referrals)

This makes Compete SA numbers even more off (not all PVs are search referrals - only the initial PV that lead the searcher to a site from a search engine is counted as a search referral, I presume).

Big note: this is not a Compete SA vs. Quantcast comparison. I am using Quantcast here only because in this case Quantcast gives the actual traffic numbers (i.e. not estimated), because Technorati happens to have the little Quantcast beacon.

March 1st, 2008 at 4:29 am

And here is question. I’ll use Technorati as an example, why not:
http://searchanalytics.compete.....norati.com

Note this in the Overview box:

Domain / Category technorati.com
Time frame 2007-11-29 to 2008-02-27
Available results 7,945 terms

Is this saying that, in the last month, 7,945 distinct terms (search queries, really, right?) were used by people who went to technorati.com after doing a search (on google, yahoo, ask, live…)?

If my interpretation is correct, then isn’t this number *way* *way* *way* too low? I looked at a few other domains and the numbers looked way too low, too. Am I not interpreting these numbers correctly?

Thanks.

Thanks for Quantifying with us. We had a small error with the data update just this morning. This will be fixed shortly. The timing is certainly unfortunate.

Chris
Quantcast Corp.
chris@quantcast.com

Uh huh.

How on earth can Compete become the “Gold Standard” when they can’t even measure properly? Isn’t that sorta the whole game?

comScore isn’t fabulous, either, but they tend to be a whole heck of a lot closer to the mark than Compete.

Compete’s stats are way off base, often as bad as Alexa’s. Just pick some of your favorite sites and compare to comScore, or Quantcast if they’re Quantified, or some other metric that’s reasonably verifiable (Google Analytics, etc).

In my case, Compete says 1M, and Quantcast (we’re Quantified) says 3.5M. 350% difference, and we’re hardly unique. Go see for yourself.

I wish someone would actually fix the analytics problem already. It’s not even core to our business (no ads, so we don’t really care how many uniques we get), but it’s still stupid to see these wildly varying numbers. Maybe Quantcast will do it, but I somehow doubt everyone will get Quantified….

http://www.editgrid.com/tnc/te.....?savebar=0

Alexa is ok, we tried to do the same with quantcast but no luck so far.

Of course, free web analytics isn’t Compete’s bread and butter. They want to sell consulting and services so this acquisition should help them. Wonder what it means for their US team?

Quantcast is the only third party that consistently measures my traffic accurately. If you are operating in a niche vertical, or a geographically targeted vertical, panel based measurement will always skew higher or lower for you. One of my sites reports 100% higher than actuals on panel-based sites because the panel is overweighting my audience.

Since Alexa and Compete do not offer this service, comparing uniques is not a valid comparison for size or market penetration.

Just my 2c.

Compete, the Boston-based web analytics and online market research firm startup founded in 2000, was just acquired by Taylor Nelson Sofres plc (TNS), a leading market information and insight company for $75 Million. Compete leverages clickstream data t…

An earlier comment cited comScore as “the gold standard”. Although I agree that the market tends to default to comScore, comScore’s data sources are thin, unreliable and extremely limited in reporting the breadth of the web. The IAB recently called for an audit of comScore and others due to the inconsistency of traffic reporting, recruiting tactics and other issues. Measuring the web is a difficult business and Compete has emerged as a viable alternative. Adding the resources and reach of TNS will only improve and expand the Compete offering. I look forward to the day that Compete is considered the “Gold Standard”.