Chin Music Press - Adventures in Publishing - Google AdWords, our broken savior

Google AdWords, our broken savior

Online publishing

With the Internet positioned to be one of the most empowering mediums for small businesses to spread their gospel, Google's AdWords program — a service of Google allowing you to buy certain keywords and have ads for your products and services appear in context-sensitive searches and relevant websites — seemed superficially like the lever that could move the earth for us small guys. That was, until we actually used it. The lever, as it turns out, was rotten on the inside, and we don't have enough time or ducktape to fix it.

The idea is simple — you bid on a few keywords relevant to your product — in our case, a book (Kuhaku) — set the maximum price you want to pay for a click and then, when the ads appear in contextually relevant spaces, you hope to get a few sales. This is nice because it isn't as brutal as random advertising and one assumes Google's geniuses have done a good job with the algorithm that chooses when and where to place your ads. We've always thought that Kuhaku sales were stifled only by a lack of marketing. We hoped AdWords would be an economically sensible way to bridge our gap between product and consumers. It turns out it's all more complicated than we had hoped.

So we signed up for an AdWords account, which costs a measly $5.00. We picked a bunch of keywords and phrases that we felt were relevant to Kuhaku and set a maximum daily budget. We created a special Kuhaku sales page to which we directed all of our ads. It was simple and direct, contained photos, reviews, quotes and obvious buttons to purchase the book if you were so inclined. In a sense, we made sure the front door to the house we wanted buyers to enter was as obvious as possible.

Within hours of putting the ads up, the hits started pouring in. Hundreds and then thousands. All we needed was one person out of 500 or so to buy Kuhaku to break even on the marketing costs. Assuming the ads appear only in places people with an interest in Japan or travel literature frequent, we figured this wasn't a very tall order.

But there was a small problem — nobody was buying. Looking at our logs, we quickly realized that 98% of the hits were coming from Search Engine Optimization or obscure search sites covered in google ads. In other words, Adwords spam sites — sites that cover themselves in Google ads and surreptitiously click themselves to the big bucks. This is a no-no in the Google rulebook, but they don't seem to have found a way to stop them just yet.

One assumes these spam sites winnow down which keywords will get them the largest per-click return. So let's assume "Asian travel" is a high-paying keyword. These sites then induce false searches on their servers for "Asian travel." Google, thinking these "search results" pages have something to do with "Asian travel," then causes the relevant ads to appear. These sites then click on the ads through some automated mechanism. It's easy money for them and a complete waste of money for us.

Sure we got a dozen or so clicks from real websites like japan.about.com or Google searches for Japanese literature, but a dozen over 10,000 is a horrible ratio if you're trying to get sales without going broke.

So we went and pored over Adwords tip sites, newsgroups and forums. Initially some of our keywords were too broad. We dutifully narrowed them down and increased the overall number of more specific phrases. Again, we waited, and now, with the number of ad impressions considerably lowered (and hence, one assumed, more specifically targeted) we still saw a similar and monotonously disturbing pattern — even these specific phrases were being hit by spam sites. So now with the number of click-throughs down from the thousands to the hundreds, the ratio of real/fake visitors hadn't changed one bit.

This shifting of words, phrases, ad text and click-bids continued as we hoped to find something that would weed out the spam and bring in the real humans. However, after a week and a non-trivial hunk of change, we despairingly shut off our AdWords account.

So what went wrong? How can Google's cash cow be a farce? Or perhaps more apt — are we just too dumb and impatient to figure out the system?

Clearly it can't all be a farce or else it wouldn't be making them trizzillions of dollars every fiscal quarter. What seems to have been implied by our little experiment is that the best products to sell on AdWords are those with high price-tags. Ones where you need only get one sale out of 10,000 to justify the costs. Another thing is that it takes time, a lot of time, determination, focus and willingness to become an AdWords tech-head and make the system work for you. You need to spend weeks refining your keywords and rewriting your ad copy. This is not a system that works "out of the box." (Not that we expected to sell thousands of copies of Kuhaku overnight, but we did expect the ratio between click-throughs and sales to have been more ... rational, and that Google would have a better handle on false-clicks).

In this sense, I wonder if it wouldn't make more sense for a small publisher to spend that time and energy on building up a word-of-mouth reputation. At least then you can gain the added benefit of connecting with your customers and supporters directly instead of through a peripheral and, clearly, somewhat dirty system of advertising.

Regardless, I'd love to hear from anyone who has made AdWords work for them in selling something like a book.

Maybe sometime in the future we'll give this another try but for now, we've got a big fat book on New Orleans to hunker down and finish up.

Additional Reading: "Google Goes Las Vegas" by Robert X. Cringely. He discusses how the algorithm deciding how much you get for your advertising buck is more complicated than Google would have us initially believe.

Craig Mod >> October 25, 2005

Comments

we use adwords pretty successfully. the real trick was to do "conversion tracking". it ties the session cookie from google to our order/checkout page, so you can see how many dollars are generated by a click.

once you get it setup, i found you needed to check it every week, so see which keywords had a high "dollar-to-click" ratio.. the ones that didnt, we dropped the amount we big for, and the ones that did well, we increased it, or increased the budget for it.

if you want, email me your login, i'll poke around and cant see if i can help out the setup. works for us @ loberman. :)

later craiggers...
-m

miguelito at October 29, 2005 02:38 PM

Mike!

He's alive!

Conversion tracking -- well, maybe I shouldn't say this, but we get so few online sales directly from our site that there isn't much to track. That is to say, if there were any sales we would have assumed them to have been from Google. For the record, we ran the ads for about a week with lots of tweaking and sold 0 books.

Thanks for the encouragement though. Maybe we'll give it another try when we get this next book out to the printers at the end of the month.

Craig Mod at October 30, 2005 02:53 AM

Tried Google ads in a few places for fun. It was a good first step for getting people in the door. Most people wouldn't buy something or join into something straight up. But if there was a reason for them to come back later - e.g. content that they might like for example - then they would come back. Once they actually trusted the place or decided they needed something, then they hooked in. There's been dubious clicks when running AdWords, but the search ads were pretty solid for getting "quality traffic." The new media option (pick a target site and pay CPM) is good for flashing yourself in appropriate places. The time between clicking an ad and buying/joining varies. If you're selling something that lots of people need, it tends to be short. If you sell something that a small number of people want then it can be a very long time from click to decision (which is a real bugger if you haven't got a hook to keep them coming back, but even then they might keep a tab on your site for getting that interesting Xmas present). Thousands.... then hundreds of clicks per day sounds very high (to my impoverished mindset). I would look at limiting clicks (either daily budget, only selecting the Google search for ads, or selecting the best-matched keywords) and extending the test period as I think the sale cycle for a Kuhaku would tend to be longer (or seasonal - Xmas, returning gaijin season in March, etc) due to it's ultra-hip niche status. Another useful Google tool in conversion tracking, is using the page accesses option - I forget the name, but you drop a niblet of code on every page to see if people are actually trawling the site or just getting out of Dodge ASAP. If they're rifling through the pages, then it means you got their attention, if they leave ASAP it suggests a (non-sale) hook is missing. On the other hand, SEOing the site can do wonders for pulling people in from regular searching (e.g. getting the database to spit out individual page titles, keywords, descriptions (metatags are still used by some search engines I think). As for dealing with people, I'm about as sociable as a tarantula, but I can see meeting/talking to people being more valuable for investing most of your time than trying to create the miracle from search engine ads (have them there, but get them set up to a level that's good for you and monitor the results while nursing the Black Nikka at night). That's my thoughts anyway.

Brundle Fly at October 30, 2005 09:24 AM

Brundle,

Thanks for the excellent followup. Great points on the seasonality of sales for something like Kuhaku -- perhaps we should put the time in to do a limited campaign for the holiday seasons.

It looks like -- based on what you and mIke have said -- that with enough time to devote to getting adsense up and working for you, it's probably a solid compliment to face to face handshaking.

I think we may give another go to it at the end of the month.

Now to find some Suntory.

Craig Mod at October 31, 2005 05:25 AM

You've got an excellent post here on the hazards of using Google Adwords for low-margin products like books. I've had the same experience (as a self-publishing author of one book). I dumped about $4,000 into Adwords to market my book and was shocked to discover I couldn't trace more than a few sales to the ads.

I just wrote about my experience at my blog and linked to your post to show that publishing companies have noticed similar results.

A few months ago when Google allowed you to target specific sites, I was able to get much better click-through by showing my ads only on the few niche content sites related to my book's topic. But it still came back to the fact that the ad fees were way, way more than resulting sales revenue.

I thought if anyone had a chance of making Adwords work with book sales, I could do it because 1) I had a niche nonfiction book without much competition, 2) I priced the book high and therefore had $10 per unit I could have dumped into marketing -- if I could have found something that worked!

Since then I've been concentrating on a blog and sales have been much better with free search traffic -- with no ad bills.

Steve Weber at March 3, 2006 01:18 PM

Steve, thanks for the comments. Good to hear others are finding similar results/issues with adwords. And it's also nice to hear that good old fashioned community building via blogs, etc, is the more productive and economical way to move things like books.

Craig at March 4, 2006 04:04 AM