Yesterday I, and every other author I know, got invited by BookBub to apply to be in the beta test of their new CPM advertising service. If you aren’t familiar with BookBub, it’s a relatively new player (about four years old) to the book advertising business. In a very short period of time, they built up a rather astounding mailing list and reputation. Their reputation is:
- Exceptional performance. Books advertised in their emails often see thousands of purchases.
- Ridiculous prices. Depending on your genre and what kind of a sale you are running, a single ad could be over $1000, and all ads are at least a couple hundred dollars.
- Highly selective. They take only a quarter of the ads for free books and a fifth of ads for books that are merely on sale. The exact chances depend on your Amazon reviews, the genre, and how deep a discount you are running.
I haven’t yet tried them. As you know, I will never give away my novels for free. However, I will run a $0.99 sale now and then. It’s a risk/reward issue. If an advertising experiment like Amazon Giveaways fails, I’m out a few bucks. If a BookBub experiment failed, I could easily be out several hundred (romance is one of the more competitive genres, so the prices are very high).
I have read dozens of blog posts by people who have run BookBub ads. Unfortunately, these are all authors, so almost none of them turned the numbers into a meaningful metric. A click-based ad campaign should be able to report:
- Impressions. How many people saw the ad
- CTR (Click-through rate). What % of people who saw the ad, clicked it
- Conversion. What % of people who clicked the ad bought the book
I don’t care what happened to your “bestseller ranking” because that doesn’t matter. It’s odd that nobody reports any of these numbers. Gross sales are nice to know, but that’s not enough information for anyone else to compare to. If your conversion rate sucks, but the impressions and CTR were great, then you won’t get good sales but it’s not BookBub’s fault. If your conversion rate is awesome, but you didn’t get many sales, then the service is overpriced. Without these numbers, I have no idea whether BookBub would be a good value or not for me.
Eventually I suppose I’ll have to cough up the money and find out for myself.
But anyway, that’s not the point of this blog post. BookBub is introducing CPM (Cost-per-mille) advertising. That means you bid a price to pay for 1000 ad impressions. A computer looks through all the bids that match a given email viewer, and they choose the highest bid. They charge you the price of the second-highest one, which makes you feel good, but with any large pool of bidders it doesn’t actually make any difference because every auction will end with a tie.
Most advertising platforms have moved to a CPC (Cost-per-click) model. In theory the two are basically the same, since for any ad you can pretty quickly measure the CTR and then do a little math to find an equivalent CPM, and then run the CPM auction I just described. For example, the CTR of my Twitter ads is about 1%. So when I bid $0.10 per click, Twitter actually treats that as a $1 CPM (1000 impressions x 0.01 CTR x $0.10 bid). This is better for the advertiser for a couple reasons:
- CTR is highly dependent on audience targeting, and it’s impossible to predict in advance. So a low-performing ad won’t lose you a lot of money, it’ll just win fewer auctions.
- The service can play fast and loose with the definition of “impression.” If the service counts impressions where nobody actually saw the ad, it will drive down the CTR for everyone equally. So it doesn’t really matter.
In contrast, if you are offering CPM service, an impression had better be an actual impression. Otherwise your advertisers are going to revolt. And this is the first concern I have about BookBub’s service.
They say that they are going to count someone opening the email as an impression. If the ad was a banner up top, then that’s fair. But it’s not. They are going to put the ads clear down at the bottom of the email. When was the last time you scrolled to the very bottom of a newsletter? I certainly don’t. If they were running CPC auctions, this slight-of-hand would not matter. But when you are selling impressions, you can’t do that. Given how few people are going to scroll all the way to the bottom of a newsletter, I suspect the CTRs of these BookBub ads are going to really, really suck. Instead of 1% like I get on Twitter, it’ll probably be more like 0.01-0.1%. (Update: turns out the answer is 0.4%, so not as bad as I predicted, but still terrible.)
An obvious question about running any kind of auction-based advertising in a newsletter is the time-travel question. How do they know what ad to put in the newsletter before they know if it will be opened? In this case, I think I get what BookBub is planning, and it’s genius. The email will contain an image link and a click link for the ad. When the email client asks for that image, the BookBub servers will run the auction at that moment and serve up the winning ad. That way, they don’t have know in advance which emails will be opened. It’s exactly how banner advertising or google ad words works on web sites.
However, there’s a really hard technical problem here: gmail. You might not be aware of this, but Google doesn’t actually show you the email you were sent. When gmail gets an inbound message, it goes through and replaces every image “src” attribute with a URL that goes to Google. When you view the email and your client (gmail app or web browser) asks for that image, Google’s servers request it from the source and then save it in a cache. This has been a nightmare for email marketers. It’s absolutely impossible for you to know if someone viewed an email more than once, because the second view will go to Google’s caches 100% of the time. That’s not a bad thing in this case, because it just means that the auction will only run once per person, instead of once per open.
The nastier issue is that Google ignores certain things about the URL when it does the caches. So if you aren’t very clever in how you serve up your ads, every gmail user is going to see the same ad. BookBub has been in the email marketing business for a while, and so I would hope that they are all over this particular issue. But it’s definitely something as an advertiser I’d want to know how they are addressing.
My other concern has to do with BookBub’s reputation and advertiser base. The vast majority of people advertising on BookBub have no idea what the fuck they are doing. And this concerns me because I’m afraid they are going to bid crazy high. There is such perceived value in being on BookBub that people will throw money at these auctions even if they don’t work at all. This isn’t BookBub’s fault, of course.
If you compare this to Twitter, Facebook, Amazon, or Google advertising, you can imagine that marketing-illiterate authors are going to make up a tiny portion of the advertising base. So the fact that they don’t know how to bid properly isn’t going go mess it up for everyone else. But imagine running an auction for a chocolate bar in a room full of kindergarteners. You’d have kids bidding 10-100 times what the chocolate bar actually cost. That’s what I’m afraid BookBub’s CPM auctions are going to be like.
Anyway, I applied for the beta just like everyone else. So on the very unlikely chance that they let me participate, I’ll report back.
Update: I got into the program and ran an ad.