Bring A Trailer Hits 15,000 Posts

Bring a Trailer, the most addictive car site for anyone who truly appreciates cars (if you’re at work, you’ve been warned), just hit their 15,000th post since Randy Nonnenberg and his team launched the blog in early 2007.

What does that mean? The site lists 15 posts at a time. If you wanted to browse every post on the site, hitting “Older” at the bottom of every page, you’d have to go through 1,000 pages to see every post.

Wow. Actually, it may sound like a lot of work to read through 15,000 posts, but the high bar the team has maintained in writing quality & community moderation makes the impossible possible: time-travel.

Because when you’re browsing BaT, time slows down. The clock is saying I’ve spent 60 minutes here. But I feel like I just got here 10 minutes ago. That must be goodyou know why?  If my body thinks only 10 minutes have passed in the past 60 minutes, it must have forgotten to make me 50 minutes older. Aging is all mental, right? Mostly?


Anyway, although the blog has been around since 2007, BaT is just getting started. Their relatively recently-launched auction platform is growing wildly, making the site financially sustainable and primed for growth. It’s a superb digital property that’s found a beautiful means to monetize its media beyond the obnoxious ads business to which so many other media properties fall prey.

To commemorate their 15,000-post-milestone, new commercial approach, and burgeoning community & success, I took the liberty to analyze their enormous, ever-growing oeuvre to see if I could uncover any interesting insights. I’m not a statistician (or a computer scientist…or any kind of -ist, really) but I quickly hacked together some scripts to analyze the 15,000 post titles and determine the most common 2-word, 3-word, and 4-word phrases in those titles. I used a very naive implementation of n-grams, an algorithm used in computations linguistics and probability.

Since BaT prides itself on hand-picking the cars it posts to its site, the hope was to see if any trends emerged in makes, models, and any other terms.

Below are snapshots of the top ~28 rows for each grouping (2-word, 3-word, and 4-word phrases). I won’t comment on the results (you can draw your own conclusions), but I’ve made the entire table of results available here (it’s a Google Doc, but it’s a pretty large one, so you probably don’t want to open that link on a phone).

Here are the results.







That’s all, folks! As I mentioned above, if you’re interested in taking a look at the whole data-set, here it is. That link includes the entire analysis, which is over 130,000 rows, so it may take 2-3 minutes to load.

UPDATE: since making this post, I’ve updated the analysis to be case-insensitive. There wasn’t a huge difference in the results, but for the sake of completeness, the updated analysis is at the Google Docs link.

Thanks for reading. Follow me on Twitter and check out my homepage to see what I'm working on.