No Conflict, No Interest

That's a John Doerr quote that I often use. And this post reeks of conflict of interest. So I'll disclose that right upfront. I am an investor in both Indeed and comScore.

TechCrunch has a post up about Indeed's competitor Simply Hired's use of WhenU to drive traffic to it's service. Until recently, Indeed had about three times the audience of Simply Hired. In the past couple months, Simply Hired's numbers have been on a sharp increase. Now there are a bunch of services in the jobs space on the web. So it's not necessarily a good thing to focus on just Simply Hired and Indeed. There are a lot of places to go to find jobs on the web. But I personally think the job search services like Indeed are the best way to find a new job.

What's most interesting to me about this post is not the question of whether popups/popunders are a good audience acquisition strategy (it is not). It's whether UVs seen by virtue of popups/popunders should count in a service's UV count.

It seems that Compete, Alexa, Hitwise, and possibly other third party tracking services count that kind of traffic. And it seems that comScore doesn't. If that is in fact true, then it's one more reason why comScore's data is better.

Comments

[Stat Head]

I'll disclose right upfront - I am very intimate with Compete, although I am not an investor :)

Fred,

I disagree, I think it is good that Compete, Alexa, etc captures that traffic.

As a marketer (or competitor) I want to know what the marketing team at Simply Hired, etc is doing, and find out if their latest campaign is working for them.

Take a look at the "engagement" metrics on Compete.com for simplyhired. They're all down *dramatically*.

From that, now I know I probably do not want to follow in SimplyHired's footsteps. Right now Compete.com does not tell us where the bump in traffic came from, but I think that will come soon.

If some of these other services had filtered the traffic, I would have never known.

As a marketer, I'm much more concerned with how I can drive quality traffic to my site vs. comparing myself against someone on a pure UV basis all the time.

Only when I can start improving, that can I start competing on UVs :)

Hope to meet you finally at Gnomedex :)

ps. I believe whenu serves most popunders from their own servers -- so it should not bump simplyhired's traffic, unless someone clicked on the advert and actually went to the site. My assumption may be incorrect.

[Martin Edic]

Our experience with Alexa is that it doesn't have relevance- their results can easily be skewed and gamed because their installed base is simply too small. The only reason, IMHO, that they still exist is that they're the only free way to get some idea of competitive stats.
The use of pop-unders to build apparent traffic is probably viewed as a clever guerilla tactic internally at SimplyHired- which is an incredibly stupid approach. Indeed has a great user-experience. If I went there and found pop-unders I'd never go back. I'd hope that building real user-traffic rather than inflated site stats would have priority from a business POV....

[Andrew]

compete.com does not count those as page views.

[No Conflict?]

Hitwise does a good job of filtering-out pop-up and pop-under traffic and because they report data on a daily basis, their stuff is more actionable than comScore.

[Ben Edelman]

Fred, I share your sense that the big-picture questions are most interesting. And make no mistake about it: This issue is bigger than SimplyHired.

In "How Spyware-Driven Forced Visits Inflate Web Site Traffic Counts" (May 2007) ( http://www.benedelman.org/news/050707-1.html ) , I showed half a dozen specific examples. Frankly it wouldn't be hard to document a dozen more. (If only I had the time!)

In my testing and in my observations, companies in intensely-competitive traffic-based businesses seem to be turning to forced visits with growing regularity. I consider that unfortunate -- in part because it pressures competitors to do the same, and crucially because the net effects include 1) more popups and popunders (which reasonable users dislike for their harms to productivity), and 2) more money flowing to software users reasonably disfavor (e.g. for nonconsensual or misleading installations). So Fred and I reach the same bottom-line conclusion -- that forced visits aren't a good "audience acquisition strategy."

As to Fred's final question, whether these visits should count towards unique visitor counts, I think the answer there too ought to be no. As most folks use UV counts, their purpose is not so much to explain what happened in the recent past ("how many people came last month?"), but to predict what will reasonably happen in the future. ("How popular is the site? How many quality users will come next month and forever more in the future?") Now, bona fide past visits are a good prediction of bona fide future visits: If some number of users came last month, willingly and intentionally, it's reasonable to think about as many will come next month. But as to forced visits, the inference is much less clear. How many forced-visit users come next month just depends on how much forced visit traffic the site purchases. The answer to that question says nothing about the site's long-term traffic trends or its true popularity with users.

To my eye, then, investors ought rightly focus on indications of a site's real popularity -- of the users who **choose** to go there, who **want** to go there. Investors can't learn much from counts of forced visits that are directly within the site's control, and that are at best weakly correlated with a site's true popularity. Though it's not easy for traffic measuring services to separate the forced visits from the actual visits, the actual visits are the numbers I'd want to watch if I were trying to predict a site's future standing.

[Robin Wolaner]

I had always heard that no conflict, no interest was a Dick Kramlich quote.