January 16, 2007 Delicious Library (which I hardly ever do) I'd say, "All this without ever advertising!"
I stopped saying that because we do advertise now, and lying is bad. ("We" in this case is me and the frog in my pocket. I call him "Sir Jumpy.") I thought I'd share the little I've learned about advertising as a smallish software company. (Again, the Sir Jumpy and I incorporated a while ago, for tax reasons.)
- Advertising in magazines: pretty much sucks. Well, it might not suck, but you'll never know. First off, nobody reads magazines any more. Proof: what are you reading right now? Is it a magazine? No. There -- what more do you need? (If a single data point doesn't define a trend, then I guess our entire war on terror was a huge mistake?) Second, even if people did read magazines, you'd never know which ads of yours were effective, since there's no really good feedback mechanism of which I know. Third, magazine ads typically cost big bucks.
Magazine guys have told me, "Hey, advertising isn't to generate sales, it's to generate exposure!" OMGWTF?! If I wanted exposure I'd move the L.A. and go clubbing every night with Paris Hilton and her growing menagerie of hangers-on and/or venereal diseases. I WANT SALES, DAMMIT! I'm not paying for some squishy "exposure" thing. Give me money. That's what I want.
Also, magazine ads are really hard to compose, so you'll spend a bunch of time and effort trying to make one that looks all professional and shit, but you're a software developer, not a professional ad-making-type-person (hmm, there's probably a better word for that) so, if you'll forgive me for saying so, your ad is probably going to look like shit. And, you're probably going to forget to bring teh funny in your effort to look all professional, so you'll just end up alienating your core market of crazy independent thinkers with your attempts to look all sophistomicated. (Remember this is a community where the most remembered ads are a lady smashing a T.V. and "It sucks less.")
- Advertising on Mac-specific web sites: pretty good for me. My advice for this is the same advice I give for picking stocks: pick the sites that you yourself really like. (As opposed to "you notyourself" -- that guy has no taste.) I advertise on Crazy Apple Rumors because I think that guy is teh hilarious (that's two -- if I use "teh" one more time it's going to be funny, given the comedy rule of threes) and because it's a tiny site and he basically was willing to sell me the ad space for a hot dog and a cup of coffee.
Now, I don't get complete sell-thru data from C.A.R.S. (see next section), but I *do* get to track referrals from them to my site using the cool tools provided by my web host (in my case, it's called "Urchin"), so I can measure some of the efficacy of this, and I've found that, per dollar, C.A.R.S sends as many people my way as Google's ads. Since C.A.R.S. readers are also clearly Mac fans and have demonstrated their superior intellects by reading such a fine site, my estimation is they are more highly qualified customers than the generic Google referral, so this advertising was worth it to me, even though the total volume of hits I get is, of course, an order of magnitude smaller than with Google.
- Advertising on Google Adwords: good if you are VERY careful. For the two of you who don't know already, Google makes its zetazillion dollars a year by allowing advertisers to bid on placing ads on (a) search results pages, and (b) blogs and the like (called the "content network"). Because Google is made up of a bunch of well-massaged, well-fed, lovable computer wonks (seriously, Google employees are the Kobe Beef of computer programmers), they've designed an incredibly rich set of tools for setting up and monitoring your ads.
The basic idea is that the advertiser (me, in this case) says, "I'll pay up to 10 cents for Google to display my ad on a page that mentions 'Collecting' and 'DVDs'." If I'm among the top six or so bidders for any given page, my ad gets displayed. If my ad never makes it into the top six, or if nobody ever clicks on my ad, Google automatically disables it and tells me I better up my ante.
One of the coolest parts of their system, though, is that they automatically track customers who have used AdWords to get to my site, and I can tell Google if a particular "session" ended with a sale, so, for what I believe is the first time in the history of advertising, I can actually track exactly which ads, run where, cause customers to buy my product, not just "eyeball" it or be "exposed" to it or whatever. This is truly amazing.
There's a bunch of other factors in all this, but that's a rough outline of how it works. So, what's the problem? Well, the first thing scammers started doing was clicking on their competitors' ads; if you sell library software and you see an ad for your competitor, you might think, "Hey, I can just sit here clicking this link all day, and it'll cost him 10 cents every time! MUAH HA HA!" Now, I've set up a limit on AdWords on how much I'm willing to spend per day, but even so, if you false-click enough of my ads, they'll stop showing up for legitimate customers for the rest of the day, so you can essentially do a denial-of-service just by leaning on my ad, in addition to costing me money. CURSE YOU!
Sucks, huh? Well, Google claims they've figured out how to detect this, but of course isn't giving any details. In my experience, this kind of "click-fraud" isn't the primary problem with AdWords, but it's something to think about.
The latest form of fraud I've read about works like this: some unscrupulous site picks a bunch of words that are close to the words you'd use to advertise your product, but not quite (for example, common misspellings). They then bid, say, a penny for those words, since they are so uncommon. People who search for those words are led to a page on their site which contains Google content ads for YOUR site, for which you're paying ten cents a pop. Sure, it's not illegal, but it's kind of a sucky thing to do, in that it ends up getting around Google's automatic suggesting of similar, more successful searches, which would lead people directly to your ad and/or your site without your ad.
I don't know exactly of what kind of shenanigans I was the victim, but a couple months ago (November-December) I discovered that my cost-per-conversion (that's how many dollars I spent on advertising to generate a single sale) went from around $29 per customer to around $75 per customer. Yipes!
Now, even $29 per customer may seem high on a $40 product, but I like to fudge it in my head and think, "Sure, but think of the exposure!" I mean, there is a certain value in just having your name repeated, over and over, so I factor that in to how much I value AdWords. Also, there's a value to getting a customer for a 1.0 product at any price, because once gotten (gotten?) you don't have to advertise to her to get her to upgrade to 2.0.
But $75 per customer? So I'm losing $35 for each customer I get through AdWords? What am I, Sony?
So I started diving into Urchin, and the first thing I discovered is a VERY VERY high percentage of referrals to our site (ONLY during the months where I started losing money) came from some place called "www.losmejores-juegos.com/g-common2.googleadd.php". Take a good look at that URL... it just kind of sounds suspicious, doesn't it? I mean, a whole page just for "googleadd"? Why not just call your page "clickfraud.php"?
I went to that page (now defunct) and it was very much just a referral page full of ads. Hmm.
Then, on AdWords, I discovered that my ads on the "content network" were sucking up all of my ad budget every day, in just a couple hours, so I wouldn't run any ads for the rest of the day, in either the content network or for search pages.
Hmm and hmm, I say.
The smoking gun, however, was that these "content network" ads, while claiming to reach hundreds of thousands of people a week, were generating NO revenue. I mean, literally NONE. None of these "people" who were clicking through these ads (remember the user has to click on an ad for me to have to pay for it) to my site were ever buying my product!
Well, I know from earlier statistics (back before I was getting boned by los dickheads) what ratio of people normally buy Delicious Library after clicking through to our site, and it was WAY more than, say, 0%.
What was the solution? Well, in this case, I did two things: one is, I assumed that losmejores-juegos.com was up to no good, and added them to my AdWords list of sites that are simply not allowed to run my ads any more. Screw you guys, I'm going home.
Second, I discovered that historically, "content network" ads on Google really kind of suck -- they suck down the majority of my ad budget and almost never generate conversions (sales). So, to hell with them -- I lowered my bids on the content network down to almost nothing.
Result? My cost per new customer averaged $13.20 for the first ten days of January, instead of $75. Hooray for Zoidburg! Everything's coming up Millhouse! And other Groeing quotes!
I also wrote Google about all this (twice), but received a stock responses that basically said, "If you decide you don't want your ads on a certain site, just exclude them" and then "We watch for clickfraud in general, we're not going to look into this particular case."
So, to summarize: Google AdWords can be really good, but don't depend on Google to make sure you don't get boned. Concentrate on the search ads, and don't bid high for any ads on the content network right now. Honestly, the content network never really paid off for me even when I wasn't being actively defrauded, and recently it made me not only waste my entire ad budget, but also made me thus miss advertising to legitimate customers on the search side, because my budget was drained.
Sir Jumpy says: Google content network == teh suck.
I write the software for
I'm kept alive by Delicious Library
I'm kept alive by Delicious Library
- MacWorld EXCLUSIVE: Expect "all-new" Keynote '07!
- Pimp My Code, Part 13: The Pimp Before Christmas, ...
- Marketing Irony.
- Quel Horreur!
- You've Gotta Fight for Your Right to... OLPC!
- Thank God for George W. Bush
- Flame: My birthday present to me
- Pimp My Code, Part 12: Frozen in Carbonite.
- "It is what it is" + CPU giveaway
- Boot Camp of the future...
Pimp My Code
- Free Programming Tips are Worth Every Penny.
- I will insult your code!
- Part 1: Code Insults, Mark I
- Part 2: self = [stupid init];
- Part 3: Gradient TableViews
- Part 4: Returning late to return early
- JPEG2000: Cool but SLOW.
- Unit testing is teh suck, Urr.
- Part 5: Special Apple Sample Code Edition...
- Interlude: Free Code
- Pimp, Pimp Thyself.
- Frameworks are Teh Suck, Err.
- Part 6: The Pimp Before Christmas
- Thinking, boxes, & what kittens can do to them.
- Part 7: Pimplette?
- Part 8: Mary, Mary, why you buggin?
- Part 9: Beginner Code
- Part 10: Whining about Cocoa
- Part 11: This Sheet is Tight
- Part 12: Frozen in Carbonite
- Part 13: The Pimp Before Christmas, Redux
- Part 14: Be Inflexible!
- Part 15: The Greatest Bug of All
- Part 16: On Heuristics and Human Factors
- Part 17: Lost in Translations