TVbytheNumbers by the numbers
Earlier in the week someone requested a view of TVbytheNumbers by the numbers. Given all our recent site outages, it sort of makes sense.
Our site statistics have been publicly available since our launch in September 07. As background, I found out about Quantcast, in a way, because of the iPhone. I wound up getting one the first day, though not until late in the evening. I avoided the lines, and was in and out of the Apple Store in about 5 minutes total.
I found out about Quantcast because of the iPhone
The next day I walked over to the local AT&T store to buy my first (and aside from replacement earbuds) and only accessory I ever bought for the phone. A plastic case with kickstand that allowed you to prop the phone up for watching video. I wound up using that for like two days. It worked, but it was a pain to take the case off, and I didn’t like the bulk it added to my pocket. I discovered pretty quickly that the phone could be propped up nicely by folding up something like a paper bar napkin. And though I am not easy on gadgets and have dropped the thing multiple times, it’s highly durable. I haven’t had any problems with it in the almost two years now that I’ve had it.
But after buying the stand I decided to walk across the street to Betelnut, an Asian-fusion restaurant in the San Francisco Marina and have a glass of wine (or three) and orders some chili encrusted calamari. In some ways, I ascribe some of TVbytheNumbers’ success to the calamari. Why? It was one of the first ways I found out about how the Google indexed things. Bill and I already knew we were going to launch this site, and I was playing around on a personal blog to try to get up to speed on the way the world works. I had somewhat of a background in search algorithms and indexing, but I’d been on the bench (or beach) for a while, and like most people, I just used Google because it worked.
Sure, I knew a little bit more about Boolean searching than most people, but mostly I just used Google and never really thought about the way it worked. Where the calamari played in was that in the first set of commercials that came out advertising the iPhone before it was actually available, one of the demos was on finding a restaurant with calamari. They picked Pacific Catch – only a few blocks from where I lived at the time. So I wrote a post about how the phone looks really cool, but the best calamari in that neighborhood was not at Pacific Catch.
Learning Google: iphone calamari
People were finding that post on Google, and quickly it became one of the highest ranked items if you did the search iphone calamari in Google. I wanted to understand why that was. So I played around a bit. I didn’t persist with it long, and if you do that search now you won’t likely find my blog anywhere.
However, if you do the search TV ratings you will certainly find TVbytheNumbers. Ahead of well, pretty much everyone. Ironically, including Nielsen. Don’t be overly impressed by this. In the end, all it really took was time. Being around a while is definitely helpful when it comes to Google’s search algorithms.
I don’t ascribe being ahead of Nielsen, Variety, Marc Berman, TV Week, etc. for the search ‘TV ratings’ to any particularly prowess on our part, instead, I ascribe it to those entities not understanding the environment they’re actually in and in many cases probably not even being aware that understanding the environment they are in is very important.
Anyway, I went over for the best calamari in the area, sat down at the bar, ordered, popped the earbuds in and started watching Caddyshack on the iPhone. It was glorious. But it was also pretty much “ground zero” for the iPhone, and everyone wanted to see…the iPhone. It was a good conversation starter, but by about 3 days later if you pulled out your iPhone in the you were pretty much considered a poser. But one of the people I talked with that day was a woman who was wondering whether she should buy one for her boyfriend. That was Krista Thomas, at the time the head of PR for a new startup, Quantcast. She’s no longer at Quantcast, but she was very helpful to us on the PR front when we first started out. And we dropped the Quantcast pixel into our code very early on and have been measuring since our early and humble beginnings.
Other than how Quantcast measures “people” all of the Quantcast data for us tracks extremely closely with our internal logs and measurement. People doesn’t jibe with our “unique visitor” counts, but that’s because Quantcast measures unique visitors differently (and perhaps more accurately). But if you look at total visits and page views, you are getting almost 100% accuracy.
The metrics I obsess over
In the early days, I was obsessed with the notion of having a site that generated a million page views a month. I’d share this obsession with Bill and early on, sometimes he wouldn’t say anything, he’d just look at me with a glazed “You.are.NUTS!” look. After about the first 13 months, I wasn’t sure I wasn’t nuts. But in our fourteenth month, October 2008, we went above a million page views. At the time, I’d e-mailed my old friend venture capitalist Fred Wilson, to let him know my obsession with a million page views was complete. Fred’s response?
If you can do a million, you should be able to figure out 5 million.
And I thought, “I love you Fred, but you.are.NUTS!” But perhaps not. Even with all of our May downtime, which has been too much, a lot of work staying on top of, and something that led us to switch hosts, if the site stabilizes after the move, we should hit 3 million page views or pretty close. It’s not unthinkable that we could hit five million next season, but I won’t be e-mailing Fred if that happens. He’ll just say “if you can get 5 million you ought to be able to figure out 10 million.
I’m not going to provide a tutorial on how to use Quantcast but you can view our traffic in a variety of ways, going back the whole time our site has been available. The problem is that some spikes make the scale sometimes fairly useless, but if you select “Page-Views” in the dropdown menu and click on month rather than day, you will get a good representation of the our real trend.
Beyond page views and visits the metrics that I mostly obsess over are how people are finding our site, and whether the amount of people coming directly to our site is increasing, and how much people are interacting with the site (are they looking at one page and leaving or are they sticking around?).
To give you something that’s perhaps easier to read with more data, here’s a table.
Engagement matters most (at least to us)
You’ll note that although we had more unique visitors in January, we had fewer total visits and page views. There’s a good reason for that. In January we had a couple of very heavy linking events from Drudge. This kind of traffic is awesome in a way (if your server can handle it, and believe me, anywhere you see a large spike in Quantcast is a day we were struggling to keep the site up and running) – it definitely increases visitors, and on Quantcast your people count (which increases your Quantcast rank). But it’s not the best traffic because in events like that, you have a preponderance of visitors who click the link to get to us and then click right back to where they were before.
That sort of thing bolsters many traffic metrics, but I’m concerned with engagement. Looking at things like average number of visits per month and average number of page views per visits. Huge traffic spikes are good for total counts, but they knock all the engagement measures lower. For whatever reason, we haven’t had any huge linking events recently, but our traffic is growing very nicely organically, despite whatever momentum we lose anytime the site goes down. Without site outages, April might have been double January in page views (and May might be anyway). That’s 100 percent growth in page views in three or four months, in months where we didn’t have any very heavy linking versus a month (January) where we had huge heavy linking.
I could focus on the sort of thing that would drive huge spiky traffic to increase our people count, and our ranking by unique people, but I’m more concerned with engagement. I’d rather have 500,000 people driving nearly 3 million page views a month, than a million people a month driving 1.5 million page views. Though I do understand it from the advertising perspective, I think the Quantcast approach to giving a much higher ranking to the million people a month who are driving 1.5 million page views scenario is flawed. Perhaps not in terms of the number of eyeballs a site is reaching, but certainly in terms of overall engagement.
Our regularly produced data, particularly the overnight reports and the Renew/Cancel index have had extremely nice organic growth. In the case of the Renew/Cancel index, it isn’t surprising. It’s not really surprising to me with the overnight reports either. When we started those, we felt the only value we could consistently offer versus the couple of places those numbers were already provided was making them easier to read.
Now there are more sites offering ratings data than when we started, and you might wonder why more of those sites don’t make easy to read data tables. It’s not always easy to be nimble and do that sort of thing when you’re in any kind of legacy publishing environment. Translation: it winds up being a lot of work, depending on your setup. In our setup, it’s not a lot of work. For the hard core ratings fan, being first is probably important. And we’re not usually first. But most people aren’t hard core ratings fanatics so “easy to read” is valued. And the truly hardcore ratings fans look at almost everything available.
The site’s community has also grown at an accelerated rate. Most days these days well over 1,000 comments a day are posted to the site.