Why Has Traffic to Auto Sites Crashed?

The following chart (via Compete.com) of top Internet domains ranked by traffic gains/losses in 2007 is interesting. Other than porn and video, there is not much commonality among the gainers, but it is intriguing that three of the biggest traffic losers in 2007 were auto related: autobytel.com, motortrend.com, and carsdirect.com.

Why would that be? My initial reaction was that there must be something wrong with the data. Compete.com has its critics, and it wouldn’t be surprise to discover it was awry here. Are subdomains being missed? Changed URLs? Read the comments on the Compete.com blog post and you’ll see some people already trying to untangle the numbers.

My second reaction was to muse whether this could have to do with fewer Americans buying new cars. Fewer new cars purchased = less traffic required to car purchasing sites, right? Possibly, but there is no way that you’d see, say, a 5% drop in car purchasing turn into a 90% traffic drop.

The correct answer, I’m guessing, is that it’s all some funky Compete.com data error, but I wonder what. Anyone else care to mull what’s going on here?


  1. Something that Compete does not track is sub domains so websites like Yahoo Autos may have taken away a lot of traffic from these sites. I have noticed a huge push with Yahoo advertising its Yahoo Autos all over the web. People may just be sticking to their networks when they search for cars.

  2. Quantcast gives similar declines – althought it only shows data for the last 6 months.

  3. Google algo change? They zapped the comparison shopping sites a little while ago. Auto arbitrage next?
    Just a guess, you’d need before/after charts from Hitwise on inbound fraction of traffic from Google to the sites to confirm.

  4. aaron wall says:

    Consumer auto spending is probably off somewhat, as one piece of that.
    Brand advertising by end dealers, in response to above, is probably up sharply, which may drive more direct traffic to the end brand sites and dealerships.
    Some of these sites also syndicate their content to newspapers and other authoritative websites, which might have got whacked by duplicate content filters.
    If you search for the core brands… like honda, for example, the google sitelinks drive down all the organic listings that come after the official site.
    Universal search has also snagged a chunk of the automotive industry. Search for honda civic and find
    – wikipedia listing
    – youtube listing
    – google news listing
    Couple those three with
    – the honda.com homepage
    – world.honda
    – automobiles.honda double listing
    – and a honda ad
    that is 8 listings right there.
    Plus they compete with yahoo autos, msn autos, small enthusiast sites, and photos that are put in the search results (search for pontiac g5). Kinda a rough vertical to be a generic overall site in if your content is similar to everyone else’s content.

  5. Paul – When writing this post I noticed the auto site decline as well.
    I think you can attribute this decline to a couple things:
    1) Sluggish auto sales in 2007 (less people researching cars)
    2) A saturated market of 3rd party research sites: from what I can tell Autobytel and Carsdirect compete with about 20 other sites. Autobytel doesn’t seem to offer any sort of competitive advantage from any of these competitors. Additionally, people just might not be ready for Carsdirect.com’s “online car buying” experience.
    I would also break motortrend.com out from the other two. While it offers automotive research, it’s also a media company competing with (growing) automotive blogs.
    They’ve also pushed their dealer lead-gen material front and center (at the cost of burying their auto news content) so the site behaves similarly to a really weak 3rd party research site, which in the already overcrowded (and now sluggish) market, won’t cut it.

  6. My anthropologist $.02. You will find a high correlation with fuel price spikes and peak driving periods. Gas price spikes induce anxiety which is soothed by a visit to the local and familiar autosite which is covering the issue. …and now back to the unwinding of the CDS market and looming systemic risk.

  7. Autobytel launched myride.com as its core site in October and is no doubt using its marketing ammo to drive traffic to that URL. That explains one of the three.

  8. Rather than take the “declining auto sales – it’s a RECESSION!” bait, or attempt to explain specifically why some auto sites may have declined in viewership, I’ll attack Compete.com directly, then make a statement about the traffic on the sites mentioned.
    First, it’s a third-party measurement of site activity, and it’s inherently inaccurate. Heck, scripts on your own page can’t measure visits with complete accuracy! Add another layer of confusion to the mix and see what happens.
    Second, a change between two data points a year apart does not a trend make. One needs to see several data points in the interim to establish a trend.
    Third, and related to second, what is the average month to month variance of the ESTIMATED (since they are estimating and not measuring, see first above) traffic to each of these sites? We need an estimate of noise to compare to what we might interpret as signal.
    Finally, take a look at some of the more suspicious entries. Are we to assume that Nextag.com got 79,434517 hits in Dec 2006 (do the math on a 70% decline to 23.8 mil)? That means almost every other person with internet access in the U.S. visited that month! Alexa data shows the spike in Dec 06 versus Dec 07 to be about a 48% decline in percentage of total page views, and considering that total page views on the web are probably going up, then Alexa’s estimate is for a substantial decline, but one of considerably less magnitude.
    I would say that Compete.com, like any other traffic estimating service, has a large amount of error and noise in their measurement. At the extremes, they’ll have directional accuracy but be off by some substantial amount, and at the bulk of the distribution they’ll be useless.
    By “useless” I mean that, if my eyeball estimation of their error’s order of magnitude is correct, you could infer basically nothing from two websites with traffic levels within +/-10% or less of each other, or estimated moves of 10% or less.
    So there’s a good bet that the auto sites mentioned did decline significantly in traffic.
    SO WHAT.
    Without a measurement of traffic to ALL auto sites, there’s now way of knowing if it’s a global phenomenon. That’s like basing your retail sales estimate on your wife’s observations on her way to the mall. Does your wife go to EVERY mall?
    Even if you found a decline in site visits to ALL auto sites, you would still need to establish a correlation to auto sales, and investigate whether, statistically speaking, auto sales were correlated to moves in GDP, and whether those moves were trailing, coincidental, or leading indicators. Then you would have to extrapolate what traders or investors should do based on GDP.
    In summary, the information in the post is trivially interesting but computationally useless. Even the title is misleading, but probably sensational enough to get some reading.

  9. Largest commonality among gainers is social networking (wordpress, sixapart, squidoo, stumbleupon, ad nauseum…), not porn or vid. What jumps out at me are the sites with large % gains AND big traffic, e.g. Webkinz. With little kids of my own, I can confirm that the kid-centric (not merely “kid-friendly”, but actually targeting kids directly) web sites are the shiznit on my home pc, to the point where I don’t have enough time to look at what I want to look at myself!

  10. Media Metrix indicates a 41% year over year drop (worldwide uniques) for Autobytel Inc. in December (all properties, including autobytel.com and their new myride.com), a 67% drop for carsdirect.com, but a 57% increase for motortrend.com.
    I suspect that many of the third party dealer lead gen sites, that have historically been very active buyers of traffic, have seen the economics of acquiring online automotive shoppers become less favorable and subsequently the arbitrage opportunities are increasingly harder to find.