Jason Calacanis Punches Comscore In The Face. Comscore Punches Back. | Hacker News Reader

Jason Calacanis Punches Comscore In The Face. Comscore Punches Back.

techcrunch.com

38 pointsmun41130 comments

At the end of the day, the real problem with using ComScore #'s when developing a media plan is unlike with all other forms of mass media (TV/Radio/Print), in Online the advertiser has direct feedback from the audience of their advertisements - hence you don't really need to use a 3rd party measurement service to help you determine if your campaign was on target/"worked".

A much better way to buy media online is sprinkle money around, see what performs best, allocate more money there, measure, test, repeat.

ComScore was designed to meet the need of old world media buyers working in online - those who can't wrap their heads around allocating media $ based on actual placement performance, rather than a 3rd party measurement company's estimate of what the audience might look like. There are A LOT of these people out there - hence ComScore has done OK for themselves. Unfortunately I agree that over the long term, their importance in the world of media buying will dwindle, and they will become more of a competitive research tool used by middle management and tech bloggers to re-enforce their viewpoints.

I'm not in the ad game, but what you just said doesn't sound right to me. I mean, if I see 10 ads for a new show on TV network X, even if I don't click on them they have value. Not everything is monetizable online, and those things that aren't will be hard or impossible to track.

The key point I'm making is as an advertiser, there is no way for me to know that you saw those 10 ads on TV - hence the need for 3rd party measurement. On the web, I have a direct line of sight to you - hence why introduce a 3rd party?

Nothing I read in the replies or MA's commentary actually countered the substance of Jason's argument.

Fred's line: "someone has to pay, so pay up." That sounds like a straight-up mobster.

Linda Abraham's line: "we measure people" (what about dogs? I wondered), "we have a slightly cheaper option of 5K". IOW, we can be flexible in what we hit you up for. Oh, and we have cameras in people's homes, so we know the difference between people, machines and cybernetic half-breeds.

MA's line: this person is right, someone needs to get paid, Jason's wrong. The judge and jury in one persona, but logic-free.

But all this doesn't address the core fact: that people using ComScore to evaluate a site (why can't these people pay?) can be greatly influenced by the sites paying ComScore for "auditing" and other ancillary services to make sure they "get counted correctly". 'Cos you know, if you don't pay up, some "accidents" might happen.

It’s unfortunate that you were picked on as a child. It must have been difficult to you. But you’re an adult now. If you want to debate, please do so with facts, not just blind fury.

That last line of reply was worth reading the whole article for.

Before this, I knew nothing about Comscore. That one line has really turned me off. If they want to win people over, insults are not the way to do it.

Yeah, never understand why companies go this way, when responding to social media. I assume it's an act of impulse by one of their employees, I would be amazed if it was actually discussed and cleared by their PR department. But it sure doesn't look professional at all, and I do expect professionalism from a company like Comscore.

Isn't that the whole point of social media? Be authentic and don't sound like a PR?

I, for one, give them a hell of a lot of credit for pushing back at Jason in this way. He's the one who brought up the subject, and it's fun to see people stand up for themselves. It's part of why Jason is such an entertaining writer himself. He's a big boy; he can take it.

I can understand that certain companies use social media to be more authentic and to not sound like a PR. But comScore targets enterprise clients. They should play the enterprise game, not the social media game (although the two aren't mutually exclusive, I expect them to sound professional at all times).

Correct, I'm a big boy and I can handle it. :-) No one should ever feel bad for me... I got a free roll on life. I'm playing with the houses money and my life is 1,000x better than I ever thought it would be.

Seriously, I've been fighting with these guys for a decade trying to get them to address the fact that their data is wrong and HIGHLY DAMAGING for startups looking to get ad revenue.

They ignored me for a decade, and this email was the FIRST time I've really been able to get them to address the issue.

I try to take these inside conversations and put them in the sunlight and part of that is writing with a certain, well, flair (at least 20 pieces of flair in fact).

You can be sure I'm laughing and smiling when I write these things, and more so when folks like Comscore or Kiretsu Forum take the bait.

Now that Comscore has engaged the debate other folks can move it forward. I think Comscore will do the right thing here and move to a freemium model like:

1. Free for companies with under 50M impressions 2. $1,000 a year for companies with 50m-250m impressions 3. $5,000 a year for companies over 250M impressions

If they did this they would build a funnel of clients and a ton of good will. Compare that plan to their current plan:

1. Pay us $5,000 for accurate stats or... 2. Don't pay us and we will use our old and admittedly inaccurate system for counting your stats--thus damaging your company and ability to get advertising.

best j

The problem with your freemium model is that there's no real incentive to jump up once you're over 50MM impressions. At over 50MM impressions, you have your own in-house ad sales force (or you should) and they can open the kimono to media buyers. I think they're kind of screwed in that they have to charge everyone and I can see the logic that a min/max model would spit out $5,000 as the magic number. Plus, the only people who care about impressions are people selling on a CPM (or similar) basis. Someone like, say, Dropbox or 37signals or even Consumer Reports doesn't care.

(Lastly, you've been around here long enough to know not to sign your comments/posts. If you're going to add extra stuff to the bottom, add some of that sweet, sweet LA sunshine.)

Everyone had some valid points and we're all a lot better for the discussion. But don't become another Mark Cuban. :)

Note that Calacanis was the one who started about the boy/kid incident, if you haven't read it.

The last line was petty. No need to sully a well-written response like that... offering to answer any further questions instead would leave me much more impressed.

You know the voice that the adults talk in on Charlie Brown? Anything written by Calacanis (and to a lesser extent, Arrington) makes a lot more sense if you read it in this voice. I used it for the entire article, and I feel like I have a much deeper understanding of what's going on.

As publishers, why should we be supporting any of these services, Quantcast included? Am I really going to have to put 12 different JS includes on my site and pass on my users' browsing history to these companies just on the slim chance that some advertiser is using their stats? I get enough complaints already about the Google Analytics tracking (and that's after putting it on the bottom of the page to minimize the chance of blocking the page, thus hurting my stats).

Everyone is talking about "pixels," but actually giving publishers blocking JS includes. This means that as a publisher, you're going to have to prioritize the user experience versus tracking, and also prioritize the tracking! If you put the Comscore JS under the Quantcast JS, then the slower the Quantcast JS returns, the lower your Comscore stats will be (because it gives the user more time to navigate away from the page). It seems like it would give an incentive to the analytics firm to serve the tracking JS as slowly as possible.

I wish we could just make portions of our Google Analytics data public and be done with it. I don't think the "pixel include" race is going to scale very well.

is there something im missing in regards to this whole public analytics game? Most of us use google analytics. Those are the most accurate stats. If we're interested in making stats public (or a certain subset) like we do with comscore/being quantified, why don't we just make google analytics stats public?

1) Take API for google analytics 2) Make app with api 3) Let site owners choose which stats to make public 4) Stats are public, problem solved.

This is a weekend project that requires funding in the amount of red bull.

*Please please correct me if I'm wrong. This isn't my area of expertise and I may be overlooking some very important things.

> Those are the most accurate stats

comScore would disagree--GA counts cookies not people. When I visit your site at home, on my phone and then at work I look like three people. Add in the fact that I use Chrome, Safari and FireFox regularly and I can look like 7 people. I dump cookies quite frequently too.

Not that comScore is perfect at what they do, but GA being a gold standard for accuracy is false.

okay cool, thanks for clearing that up.

Do you know how comscore does this? Just using statistics?

Who is going to take the lead in setting up an independent analysis for Compete.com vs Quantcast vs Comscore and see who has the most accurate data when compared to the raw web server data?

I think the test should compare 20 1k, 20 10k, and 20 100k sites. If anyone decides to take bets on it, I'd guess Compete has the best data and Comscore has the worst. Let's see the services have a throw down!

Given that Comscore argues that the raw web server data is flawed because of the same person using multiple machines, different browsers, deleting cookies, etc., that would be a pointless comparison. It isn't an independent analysis to take two different approaches to information-gathering and see who has the most accurate data when compared to the data gathered by one of the approaches.

The question isn't whether Compete or Quantcast or Comscore is closest to the raw server data. The question is whether the raw server data can be used to accurately judge the number of people who visit a website. If so, Comscore's data are most likely irrelevant because they are approaching the problem in a completely different way.

I couldn't agree more. The point of the contest wouldn't be to see who guessed right, but to see if any of the parties are so far off with their estimates that they should be embarrassed.

I'd be interested to hear what Google/Quantcast/Compete have to say about people vs cookies and if/how they tell the difference.

I'd like to know how ComScore can tell the difference between a TCP connection from a machine and one directly from a human.

"As such, I’m asking for complete and unconditional surrender. Make your tracking pixel program 100% free in the next 10 days or the boycott will continue."

Who's the bully here again?

I wonder if anyone with some internet prestige can ask MA if he is unbiased in this.

Ie, does he get his comscore access "for free" -- are his stats "accurate" without paying?

Don't talk shit about Jason. He can SEO any site into 5 million uniques. Any site.

Any site where the content has already been made by others, you mean.

Word.

Reply on news.ycombinator.com