Advertising ROI, Day Eight
by Joey Manley
Well, it’s been a week and a day. I know I promised daily updates, but yeah.
Here’s the good news: the Google AdSense campaign finally kicked it. It started delivering ads on 6/25 and has been running at full tilt since. I can only assume that, since this was an image ad, the Google Robotic Brain had been holding it until a human being could review it and approve it. I didn’t see any kind of note to that effect, though, which is annoying, if that’s what happened. I didn’t do anything else, make any changes to the campaign or any of that, so I can’t imagine what else the problem could have been.
But no matter. The ads are running now. Here’s the data:
Google AdSense
Clicks:
Google count: 154
Our internal count: 232
Impressions: 425,980
Click-Through Ratio: 0.04%
Average cost-per-click: $0.39
Average CPM: $0.14
Total spent so far: $60.08
Clicks:
Facebook’s count: 162
Our internal count: 254
Impressions: 197,729
Click-Through Ratio: 0.08%
Average Cost-Per-Click: $0.53
Average CPM: $0.43
Total spent so far: $85.76
Project Wonderful
Click-Throughs:
PW count: 225
Our internal count: 327
Impressions: 1,458,332
Click-Through Ratio: 0.02%
Effective CPM: $0.07
Total spent so far: $103.22
Note: click-through ratio and cost-per-click are based on the click numbers provided by the ad platform, not on our internally tracked numbers. Our own internal click-through count has been consistently higher than any of the ad platforms’ numbers. I have a new theory. Some of the extras are caused by the ad server itself (Google’s, or PW’s, or Facebook’s) double-checking the URL to make sure that the page it’s supposed to be linking to is still in existence. I arrive at this theory because, even before Google started showing the ads, there were exactly two “hits” a day to the tracking URL I used when I set this campaign up. That had to have been Google. Our internal tracking simply counts the number of times a particular URL is hit, and I used a different unique URL for each ad. People could also, possibly, be passing the URL’s around to their friends. This is unlikely, but would be a pleasant reason to have your stats screwed up, wouldn’t it? Either way, I am going to stop reporting our internal click-tracking numbers from now on, and just go with the ones provided by the ad networks. If our numbers had been consistently lower, or if they’d been higher for some networks, and lower for others, I would still care about them. But I don’t.
ComicSpace blog reader (and creator of Calamities of Nature) Tony Piro made some helpful suggestions in the comments of the Day Three post. In particular, he reminded me that the people commenting on the blog can be helpful resources, who may be able to point the way to improving the performance of the campaigns, but only if I share more information about the campaigns themselves. So here’s the banner we’re running through Google and PW:
And here’s the ad that’s running on Facebook:
One of Tony’s suggestions was that we might get more clicks if the ads weren’t obviously for merchandise — and that once somebody clicked, he or she might actually poke around and look at stuff and maybe buy something. We considered that before starting this campaign, but decided to be very, very upfront about the t-shirtiness of the thing you’re clicking on, not out of any benevolent goodwill or anything, but just because we wanted to only get clicks from people who were actually honestly interested in buying a t-shirt, maybe. This is a very different prospect from advertising a free webcomic, just to get people to pop over and have a free bit of entertainment. I think those kinds of clicks are a lot easier to get — but it’s also impossible to measure ROI on them, or, at least, very very difficult to.
Speaking of ROI: there is none yet. We still haven’t made any shirt sales as a direct result of these ads. We’re selling shirts here and there, mind you. But none as a direct result of these ads. Yipes.
Tags: advertising, roi, Store
June 30th, 2009 at 4:15 pm
If the intent is to get people who aren’t already American Elf readers, maybe you shouldn’t devote so much of your ad real estate to talking about American Elf and James Kochalka. Could the shirt design be bigger/more visible? The design is what people buy the shirt for, after all.
July 3rd, 2009 at 12:22 am
I find this approach interesting. It’s basically the convention approach to merchandising, where a lot of the buyers you have at a con haven’t heard of you before, so will be purchasing your wares based on the designs you have.
As for the ad design, I would group several shirts together in the same ad, according to themes: cute shirts, witty slogans, etc. But basically, put 2-3 designs instead of just the one.
Thanks for providing your numbers. They’ve been very enlightening.
July 3rd, 2009 at 8:06 am
[...] out Optimum Wound’s 23 Ways for Comic Artist to Survive and check on Joey Manley’s advertising blitz ROI results so far. The Webcomic List also has some advice for comic creators and debates the merits of posting comics [...]
July 3rd, 2009 at 7:30 pm
That’s a thought. Maybe you need a shirt that uber-geeks AND hipster douchebags will like.
Something about Harry Potter in a pixel font. You got one of them?
July 4th, 2009 at 4:01 am
I have to agree that the design of the ad is where a lot of the “click problems” may be. Unfortunately the ads have go the “seduce the readers” route.
You’re paying an about 50 cents per click and that’s a lot more than you need to. With good looking ads and if you maually can place the ads (PW at least) you can bring in clicks for much less.
July 4th, 2009 at 9:38 am
[...] Manley continues to document his experiment in return on investment for various ad [...]
July 5th, 2009 at 5:00 pm
I totally agree with making the shirt design bigger. In the Google/PW ad, the words on the shirt are hard to read, and I even opened the image in a new tab to make it un-squished. The page design squishes it(some browsers might crop it).
In any ad, the words should be big, and limited. The only crowd that would care who made it would be the PW clickers, since these ads are mostly on comic sites and there might be some value to the branding. Adsense and Facebook, however, won’t care who’s selling it.
I recommend a shot of just the design(not on the shirt), with the photo of the model, and a VERY short blurb about the shirt, maybe a quip like “Click if you agree”. The url can be in small text as a footnote.
The picture for the Facebook ad looks good, but less text below would make it friendlier to viewers. It’s good to stick with just the strong words, and “design” adds nothing to the description. “T-shirt by James Kochalka, because God’s a cutie!” is good. Sell the emotion. I left the name in there because “t-shirt” needed to be in a sentence, and the words are better-separated from the imagery.
The “Only at…” statement also isn’t necessary. It’s safe to assume that a shirt in a web ad is exclusive to that site, considering the number of independent t-shirt designers out there.
July 11th, 2009 at 12:37 pm
You were tracking the sales based on these ads and these shirts.
Were you able to track whether or not those clicks bought another item instead?
July 17th, 2009 at 12:38 pm
I’ve just caught up with your past three posts about this ROI experiment and I find it extremely helpful and interesting. Thank you for taking the time to do this.
July 22nd, 2009 at 7:31 pm
Frank: unfortunately no, not technologically, but as I send out shirts I send emails to people asking them how they found out about the shirt. Of the approximately 50% who respond, none of them have mentioned the ads.
November 2nd, 2009 at 12:25 pm
Hi, I think google adwords/adsense doesn’t like it if you display information like that. Don’t ask me why, but if you want to be on the safe side better not display it. Just a heads up.