My apologies in advance for the math here– I’ll keep it to a minimum to explain the points.

Below is a Google Analytics screenshot from a funeral planning site.  Facebook traffic represents sources 4, 5, and 8.

4) apps.facebook.com: This is self-serve PPC traffic from ads that show up on apps.  Most plentiful but has a bounce rate of 90.13%– about as bad as you can get.
5) facebook.com/referral: These are from people click on our links and posts– organic traffic.  66% bounce rate, which is high, especially compared to the site overall bounce rate of 40%.
8 ) facebook.com/cpc: Facebook PPC ads that show up other than on the apps.  Bounce rate of 81%.

We’re probably paying 15 cents a click for Facebook traffic, but the 80%+ bounce rate means we’re losing 4 out of every 5 folks on the landing page.  With Google AdWords, we’re paying 50 cents a click, but bouncing only 42%– in other words, losing 2 out of every 5 visitors.

So with Google, we’re paying 3 times as much per click, but keeping twice as many folks past the landing page.  On a per kept visitor basis, Facebook is still a better deal.

For the analytics smart-alecs out there, let me respond to your points:

• landing pages are the same: True, you’ll get different bounce rates if you send to different pages. Plus, you can’t compare a homepage bounce rate against a landing page bounce rate– they have different goals.  Well, we sent everyone to the homepage, as silly as that is– but it’s still an apples-to-apples comparison.
• not all clicks are the same: A higher bounce rate is likely to be a lower conversion rate, even if you adjust for the bounce.  Maybe– although it is true that EPC (Earnings Per Click) is what matters at the end of the day, not cost per click or bounce rate.
• data is not statistically significant: Okay, so the 80% and 90% bounce rates might actually vary 5-6% should we run a few thousand more clicks.  But given that we’re spending real money, there’s no need to blow a thousand dollars to find out that our estimated 80% bounce is more truly 77%– it’s still bad.

Were I Facebook, one thing I’d do is allow separate bids for application traffic versus non-application traffic.

Remember a few years ago when Google allowed advertisers to set separate content and search bids?  This is the same thing.

We’ve been selling discount caskets and headstones online. What’s interesting is that we do the PPC for a wrestling company, for which we bid on the term “undertaker”.  Now which undertaker is the one that you’re searching for– the one that will bury your loved one or the one that will give you a body slam? At a buck a click, if we attract the wrong searcher for the wrong client, we can bleed money very quickly.  And having highly relevant ad copy only partially helps.  If you’re above position 3– or especially first position– people will just click on your ads without reading them.  This is the “I’m feeling lucky” problem with Google and bidding to a high position.  Even with negative keywords– using “WWE” for the funerals and “funerals” for the wrestling, we have only modest protection, since searchers are more likely to type in “undertaker” by itself than with a modifier.

It costs us a buck a click and 1 in 25 clicks will result in a call. Thus, the cost per call is $25. To reduce the cost per call we can get clicks for less or increase our conversion rates.  Your best bet– increase conversion rates with better landing pages.  Let those other idiots play the bid optimization game.   When you increase your conversion rates, you can then afford to bid more and shut out the other guys on traffic.  When you try to reduce your bids to reduce your CPC, you end up going down a spiral of worse position, lower Quality Score, and less traffic.  So unless you are looking for funeral planning services, you should focus on conversion rates.

This week I had the pleasure of being a guest speaker at Penn State’s GOMCHA class, taught by Jim Jansen and Daehee Park (who won last year’s contest)– thank you for the honor, guys!  Daehee and Jim are coaching ten teams of five students, each team with a client that has $200 to spend on Google AdWords as effectively as possible.  There’s not much you can do with only $200.  Assuming industry prices of about $1 a click (and I understand that such averages cover up distortions in prices of certain terms), you get only 200 clicks and have but a couple weeks to optimize the spend.  The clients ranged from a bike shop to online news site to a rural real estate company.  Most were local clients, so geo-targeting limited the competition they had to beat for the keywords they were after.  I saw great use of negative keywords, tight ad groups, DKI (dynamic keyword insertion), and varying destination urls.

What I learned about GOMCHA

I suppose the point of being a guest speaker is that I’m supposed to share an old-timer’s PPC experience with newbies.  However, I found I learned a few things.  These students have a disadvantage from the outset because they’re  not given access to Google Analytics, not allowed to place AdWords conversion tracking, or use tools like Google Website Optimizer as tools to guide performance.  Thus, the GOMCHA class must optimize to impressions, clicks, and CTR.  If your goal is generating more clicks, then the side effect is that you’ll end up going after the lowest quality traffic.  Some clicks are subjectively going to be worth more than others– however, without a bounce rate or conversion rate, what’s the incentive to go after such terms?  Also, if you want a lower average CPC, then you’ll likely bid lower , get a lower average position, and thus, get a lower CTR.  Is a “low” CTR necessarily bad?  Depends on the average position that you’re in and whether it’s search versus content.  A 1% CTR is great for being in position 10, but not so great if you’re in position 1.

More is better

Many of the campaigns had each keyword on broad, exact, and phrase match.  Thus, if you have 200 keywords, then you have 600 terms in your account. I also noticed some ultra long keyword combinations– keyphrases of 7 words and more. And then there were low bids on these exact match terms of 7 words. For your first PPC campaign, you want to be so conservative and frugal in your spend that you might even optimize yourself out of getting any traffic to start. I learned that PPC takes courage to bid up and then quickly optimize, since hanging out only on low volume terms is not as safe as it may appear.  You get dinged by having massive keyword portfolios that have few clicks, as that’s more calculation that Google must do in deciding when to show your ads.

A great learning experience

Congratulations to Jim, Daehee, and all the GOMCHA class folks at Penn State!  I was impressed with what these students were able to do in such a short time.  I am hoping that Google will select several winners from this group– they have the best mentors available!  Good luck, guys!

Google placed in beta yesterday a new service to create free display ads.  This is a cleaner implementation, in my opinion to what Yahoo and AdReady have with MyAds.  Slowly, Google is moving from text-based to true self-serve advertising.  This is going to put a lot of low-value designers out of business.  This service allows you to fill out a form with text and then be able to customize colors and effects, plus upload an image.  It automatically creates banners in a variey of sizes, saving you time and effort.

So now you can count on Google for your local advertising needs– plus get free analytics tools, email services, and other items– all free.  Now all they need to do is provide a local listings tool that submits to all the directories to round out the whole package.