You're up one day and the next you're down.
It's half an inch of water and you think you're gonna drown.
That's the way that the world goes 'round.
~ "That's the way the world goes 'round" by John Prine
Well, it’s been three months since the road to god knows… launched on GirlAMatic and I thought it would be good time to chat a bit about numbers. I’ve been meaning to do this for some time but I wanted to get enough data together to actually make it worth my (and your) while. I now have three months of data and 56 pages of story up on the GaM site. So I can definitely start chewing things over now. And share 'em with you.
First, a quick backgrounder for those who don’t know. The road to god knows… launched on GirlAMatic on December 3rd, 2007 (though a preview page was up before that). The story updates every Monday with four new pages of comic and I’ve never missed an update yet. This gives me the luxury of being able to collect and correlate data, at least on the update front, very consistently. Road is probably not the most ideal webcomic compared to many others. Unlike strips and other “gag a day” comics, road is a 142 page serial. It has a definite beginning, middle and end. On top of it, the subject matter (focusing on a teen girl’s struggle with her mom’s schizophrenia) isn’t easy or light. There is humour in the narrative, but overall the story is meant to be a lot closer to The Tale of One Bad Rat then, say, Dinosaur Comics.
Now, I’m going to show you everything I possibly can. I know sometimes people seem quite proprietary about readership numbers but I prefer transparency. Besides, there’s not really much point to keep these kinds of things secret. To put it plainly, my numbers aren't great and I don't mind people knowing that. Obviously, someone might take issue with some of these numbers and there’s not going to be much I can do to change that. If you believe I'm fudging the source numbers there's not much I can do to change your mind. I’ll put forth the data as best I can and then you can chew it over on your own. You’ll have to be come to your own conclusions.
So what am I going to be talking about here? A number of things. We’ll be chatting about what the actual stats are showing (so page views, unique visitors and all of that) plus how I’ve been finding my Project Wonderful advertising experiences. I’ll be doing a comparison with Warren Ellis’ FreakAngels (is that how you spell the bloody thing?) just so we have some context. And a few other things besides.
I also want to be able to show you exactly what I’m working with, so I’ve uploaded a number of spreadsheets to Google Docs. I’m pretty sure you have to have a Google or Gmail account to view these, so don’t shoot me for that. The data heads amongst you will be able to take a close look at the raw numbers and compare and contrast them to what I’m saying. Who knows, I might be missing something and it’s always cool to have as many different eyes as possible looking at my work. For the non-data heads who wonder why the hell Von tracks all of this stuff, please remember that I used to run a bookstore. I believe in tracking things. As many things as possible. I also hate relying on instinct or gut feelings. Numbers, at least their proper interpretation (however you define it), can clarify assumptions. Or, as Paul Zimmerman once said, “…because statistics are like love; you’re either hooked or you’re not.”
Before we go any further, it’s time for a few caveats. I use Statcounter to track all of the page views, unique visitors and whatnot to my GirlAMatic site. On top of it, I have two ad boxes on the site from Project Wonderful. The nice thing about the latter is that you can actually use PW to track statistics, too. And since it lines up quite nicely with Statcounter, each verifies the other. Or, depending on how you feel about these things, make the same errors. This could be important because GirlAMatic provides it’s own stats, too. But there are a few problems. The first one is that GaM's stats only tracks page loads. No further data is provided. No unique visitors. No “came froms.” Nothing. This wouldn’t be so bad on it’s own, but it’s made worse by the fact that the GaM stats have a very different page view count then Statcounter (as verified by PW). For the three month window, GaM estimates my all-time pageviews at approximately 44,000. However, Statcounter shows pageviews of 22,965. About half. It’s annoying because I don’t know why the discrepancy is so vast (yes, I’ve asked – there’s no answer). On top of it, I’ve noticed the same anomalies on my Webcomics Nation page. Again, everything is inflated. It’s possible, of course, that the GaM pageviews are actually correct. But since there’s nothing to “hang my hat” on, I’m not going to use them. I’ll stick with Statcounter (as corroborated by Project Wonderful) instead. Just keep this discrepancy in the back of your head as we go forward.
I'm going to try and keep things as straight forward as possible (so as not to delve into a full math exercise). The spreadsheets will show more and those who are keen can go dive into them. With that said, on we go!
Let’s start with a relatively simple comparison. Back on February 22nd, Warren Ellis released the first week numbers for FreakAngels. These are his unique visitors for the first week of the webcomic and, because he made them public, I can compare them to mine. The Google Docs spreadsheet is here.
As you can see, I’ve got about 1% of Ellis’ unique visitors (less if we compare same day to day and a little more if we compare first week day to day). What does this prove? Well, nothing save for the fact that Ellis is vastly more popular than I am. No surprise there. I’m new and he’s way (waaaaaay!) more established. It is slightly more interesting, however, when we think about some of the stereotypes with webcomics. Tom Spurgeon actually hit on one when he was discussing these same first week numbers. Spurgeon wrote that the fact that, "...'webcomics gets a lot of visitors' isn't really news..." As much as I enjoy his columns, this is the type of throwaway line that aggravates me. It is news. Why? Because so few webcomics actually are able to manage this. We can all refer to webcomics like Penny Arcade, Megatokyo, and a few others, but gaining an audience is extremely difficult. Ellis has the advantage (though admittedly a hard fought one) in that he’s an established in print author who has long used the internet to connect with readers. From that point of view, it’s probably no surprise that FreakAngels hit the ground running. I think, though, that he’s the exception that proves the rule more than anything else.
Nuts and Bolts
Ok, time to go back to my own readership. Since the road to god knows… launched on GirlAMatic, I’ve generated the following numbers:
Page Views: 22,965
Unique Visitors: 8161
First Time Visitors: 7238
Returning Visitors: 923
For those wanting to “drill down” into the real data, the Google Docs spreadsheet for every single day is over here. Plus the totals (though ignore the Project Wonderful data for the time being. I’ll be getting back to that). For those not keen on going through the numbers, the following image should help. It only shows pageviews and unique visitors, but you can get a good sense of what’s happening on every single day.
What do I see? Well, it’s pretty clear that I’m slowly but surely bleeding page views and unique visitors. That’s annoying but probably not a huge surprise. The move to GirlAMatic did get some media attention and I think we see that in the December numbers. On top of it, December was my most heavily advertised month (we’ll get there, bear with me!). In addition to that, the site was “stumbled” in early December and that gained some extra page views at that time, too. As the weekly updates fell into a routine and the advertising budget scrolled back, we see the pageviews and unique visitors shrink. This makes sense.
My frustration, however, is that I had kind of hoped for the opposite. This is probably a little naïve on my part, but I did have some hopes that the story might go viral. Yeah, yeah, I know. Wishing on high. Gotcha. That clearly has not happened. If anything, I suspect what’s happened (as is alluded to here), is that some readers have stopped coming every week. Instead, they come by every few weeks when enough material has been collected. As the creator and, for lack of a better phrase, “updater” I have no control over how and when a reader visits the comic. The best I can do is update the story when I say I will (every Monday!), do a bit of marketing (so simplistic to say but so hard to do!) to keep it in the public eye, and see what happens.
One piece of good news is that as each week goes by and four new pages of story goes up, the ratio of page loads to unique visitors is going up. That's exactly what should be happening. With that much more content every single week, there's more for a visitor to look at it and they are. That's a good thing, especially when you think that some visitors will have already read much of the story and are just looking for their latest update. So that it's going up, on aggregate, is positive and I'll cheer for that.
It'll be interesting, as I go forward, if the ratio really spikes. It probably won't go to something like 50 (50 pageviews per unique visitor) but who knows. It might!
It is also very difficult to define what a “reader” is. You’ll notice that so far I’ve spoken in terms of page views and unique visitors. Readers are far more difficult to define. In their lecture “I Can Haz Money Now,” Phil and Kaja Foglio discussed their calculation of a “reader” as being the number of unique visitors per week divided by the number of times the comic updates. In that case, mine are pretty easy to calculate – the sum of my weekly unique visitors divided by one (the number of times that road updates). If you do this on a monthly basis you can calculate a “readership” base. But man, is it ever inexact. I mean, no one truly knows. It isn’t like measuring sales (or rather, units sold). We can’t count it at all. Which is, y’know, frustrating because every creator (hell, every blogger) would love to know exactly what their readership is. In reality, all anyone can do is make an educated guess. Based on the above, I’d measure my readership (normalized for a 28 day month) as somewhere between 400-600 (see the Google Doc spreadsheet here). But I really don’t have any idea.
Here’s what I do know, though. I’m pretty damn skippy that the readership is small, that it’s not growing in any appreciable way, and I probably won’t really know where I stand ‘til the story has been posted in it’s entirety. I do know that some people really love it while others hate it. Such is life. It will be quite fascinating to see what happens with the pageviews and unique visitors once the story is finished.
Project Wonderful, for those who don’t know, is nominally a fairer and more balanced way to handle banner advertising. In some ways, I think this is true. It is certainly fantastic for sellers - those wanting to advertise something. For those who are using Project Wonderful as a way of generating revenue, however, it’s not so good. I'll be discussing the former in depth. The latter is trickier due to my site's low traffic. But I'll touch on it all the same.
First, the background. As I’ve yabbered about before, I feel I’ve spent quite a bit of money already on traditional advertising that failed miserably. So going into road’s new life as a webcomic, I knew that I had to find a very low cost way to spread the word. Some of this was done in a non-web manner. I was a participant in a big art show at Ottawa’s Parkdale Gallery that netted me some fantastic advertising (including a TV spot and an interview on the CBC). It certainly raise my profile and enabled me to do a few teaching gigs on top of it. This worked extremely well and really only cost me time. I knew I wanted to do a bit of outreach with banner advertising but I knew I had to do this in a far more cost effective way then is the norm. With a little experimentation and research during the course of 2007, I knew that Project Wonderful would give me the degree of control I wanted.
Banner advertising is a tricky business. The industry average is that only 0.5% of banner views turn out to be a click (so 5 out of 1000 views). If it costs you, say, $5.00 for those 1000 banner views, then it’s costing you $1.00 per click ($5.00/5). Which, from where I sit as a seller, is insane. I had no budget for that. My goal was to get a cost per click down to about 10 cents. Preferably less than that. And, I’m fairly pleased to say, I have managed to do that quite well. My average cost per click is down to 3 cents and I can live with that. To put it in context, since the entire campaign started on December 3rd, I’ve spent $108.60 (in US funds) and received 4112 clicks (the Google Doc links will follow in a moment). The click through rate is slightly higher than the average (at 0.78%) and certain banners have a much higher one (more on that in just a sec). $108.60 for three months is only about $33.00 per month. That’s a pizza and a beer or two. Not bad at all.
Let’s put that in context. I can’t really say that the 4112 clicks are unique visitors. They could be but there’s really no way of knowing. Odds are at least some of those are the same person clicking through more than once on the same or even multiple days or weeks. You’re guess is as good as mine. Project Wonderful actually lacks in one key area – none of the advertising data I’ve collected is available as easily downloaded CSV files. It can’t be just downloaded and imported into something like Excel. If I could do this, I could probably eliminate common IPs and "came froms" and make a stronger calculation. Instead, I’ve had to collect all of this stuff by hand. I’ve got it down to a science and I’m pretty quick at it, but I'd love to see them correct this soon. It also means that it's way more work to track daily hits. And you'll notice that I don't bother. I look at things on a weekly basis and really track most of my advertising on a monthly basis (I do it weekly to drop the hammer on any ad that's really doing badly).
I’m very particular when it comes to how I evaluate how effective my ads are. The non-math people will glaze over at this point so feel free to skip to the next paragraph. I use a four point ranking system based on the following criteria: Cost Per Day, Percentage of Clicks, Cost Per Click, and Clicks Per Day. Each gets ranked by a number based on whether it’s a single site or part of a campaign. I divide each by my actual results and get a raw number. Then I just convert it to a 10 point scale (the raw number divided by 4 and then the result is multiplied by 10) and evaluate them accordingly. I’m pretty happy with any score higher than 8. Any ad (on a single site or as part of a campaign) that can’t at least hit 7 out of 10 (generally after a week or two run) gets killed. I call this 10 point scale a Q Rating. Those who are keen can see all of the spread sheets on Google Docs. The campaign results are here while the single sites are here and here. Note that December and January are together. I really meant to separate them out but I was really experimenting with different qualifiers to see what gave me better Q Ratings. As a result, the two months are together. A total tabulation of all banner results are over here. You'll notice, for instance, that my animated banner ad had a much higher click through rate then any other ad (2.45% vs. the industry average of 0.5%) and that's probably something I'll be experimenting with more.
Project Wonderful uses an auction process for it’s buyers and sellers. If you want to run an ad to attract visitors to your site, you need to compete with other sites doing the same thing. You can do this on a site by site basis (someone can advertise, for example, on one or both boxes on my GirlAMatic site) or you can do it as a campaign. In the latter case, you enter in criteria and have PW bid on your behalf on sites that meet your specs. Personally, I use a combination of both. For instance, I find that I get a really good Q Rating on Wahoo Morris, probably because the two stories are somewhat similar and both Craig Taillefer and I are Ottawa-based creators. ‘Course, maybe it’s something entirely different. Regardless, I keep on bidding on his site because I like the results that I get.
If we assume, just for a sec, that those 4112 are unique individuals, then my $108.60 ad spend is pretty amazing. Even if it’s quarter of that (say 1000 unique individuals), then I’m only spending 11 cents on each person. When I compare that to how much I spent going to various conventions slogging my work (keeping in mind that I went to both the Alternative Press Expo and the Small Press Expo from an advertising point of view and NOT to sell things), it’s remarkable. $100.00 doesn’t even cover the table costs, let alone manufacturing, travel, lodgings and food. I would never get that type of awareness going to a convention and I’ve certainly learned my lesson. Your mileage, of course, may vary.
Really, my overall goal is to drop my ad spend to about $20.00 per month (US) and focus on getting my cost per click down to two cents. Some of you may be thinking that spending more money and aiming for a higher traffic site might be better. I’ve tried this a few times and what I’ve noticed (and again, your mileage may vary) is that while clicks do go up, the click through rate is poorer (like around 0.18%) and the cost is just too excessive. I’m big on the slow, steady and affordable way of doing things rather than spending money excessively. How you define excessive, though, may be far different than my definition.
So, Project Wonderful is great for me as an advertiser looking for affordable clicks. From the point of view of trying to monetize my GirlAMatic site, however, it doesn’t work at all. This is the problem with a low traffic site. Road does not have a big readership at all and Project Wonderful tracks this. So as much as it is very cool that I can do affordable advertising for road, I get hammered from the other side in that I can’t generate any revenue from my story, either. Now, that doesn’t surprise me and I wasn’t expecting to (unless, of course, road had gone viral). But it pretty clearly illustrates that Project Wonderful is a double-edged sword. I’m a minnow swimming in a far bigger fishbowl.
So what do I really know? Not a huge amount save for the fact that pageviews and unique visitors aren’t growing. As I reduce my ad spend (and thus the total amount of clicks, though bulk quantity is not the same as cost-effective quantity) I’ve noticed my pageviews and unique visitors dropping. This indicates to me that referrals and other linking in isn’t happening the way I had hoped it would. It will be fascinating to watch how the next number of months go until road finishes being serialized in the late summer. Traffic may or may not spike at that point. I’ll lose my weekly viewers (currently running at about 35% of my weekly unique visitor traffic) but I might gain more from those who want to read the whole thing knowing that the story is finished. I do know that, as of this juncture, I’m a far cry from the Foglios or Ellis.
That’s ok. I’m still moving forward. Stargazer is coming and it is very cool, despite any readership woes, to have road finally out there. If that, at the end of the day, is all I get out of it, I can live with it.