It's been a struggle -- a big struggle -- to pull myself away from Twitter long enough to write this column. It's a cursed addiction. Don't go near it. But if you're already hooked, follow me.
I'm going to go out of sequence again and describe a new Google AdWords that will change PPC advertising in a fundamental way. Frankly I'm surprised that there's been so little discussion about the feature in print, forums, and blogs.
I'm talking about the Google AdWords Conversion Optimizer. Google quietly added this feature to every advertiser's AdWords account a few months ago. In a nutshell, it regulates keyword-level bids, promising to deliver as many conversions as possible, at or below a cost-per-conversion you specify.
That's right -- automated bid management, the Holy Grail that's been so highly-valued that companies offering it have been built and sold at a profit -- some more than once.
In a perfect world, automated bid management performs a function that is difficult or impossible for a human to perform: monitoring conversion behavior several times per day, and adjusting keyword bids so that the likelihood is high that the advertiser will pay no more and no less than the amount required to get conversions at or below their target cost.
Conversion Optimizer has two important advantages over commercially-available automated bid management software. The obvious first one: it's free.
The second: Google says (and we trust them, right?) that the bid optimization algorithms take into consideration data that only Google can possibly have. Google knows the conversion history for most of the search terms in your campaign -- across many advertisers. So they know with better precision than anyone how likely it is that your keyword will trigger a conversion (when associated with a proper ad and landing page, of course).
Google also factors in geographic location, because the likelihood that a conversion will take place can vary from location to location. So, for example, if people in New Orleans are more likely to convert than people in Tacoma, the Max CPC Google sets may be lower when the searcher from New Orleans sees and clicks on your ad -- which means your cost-per-conversion may be lower, and/or the volume of conversions at or below your target cost-per-conversion (Google calls this CPA) may be higher.
Sounds complicated, right? Actually, it's easy to set up. But first the caveats: Conversion optimizer only works on campaigns that have accumulated at least 50 conversions in the previous month. Can't meet that threshold? Move ad groups into one campaign such that the total number of conversions exceeds 50.
For any such campaign, you'll see the Conversion Optimizer option when you click on the Edit Setting link for the campaign, and then clicking on the "Change bidding strategy" link. On the next screen you just click the radio button next to Conversion Optimizer -- and you're in.
Google will suggest a Target CPA for the ad groups in the campaign. My advice: don't accept the suggestion. Manually set the target CPA to match your cost-per-conversion goals. The target CPA can be different for each ad group.
I hope you see the same thing we observed in our clients' accounts -- a nice increase in the number of profitable conversions. But you may need to be patient -- depending on the campaign velocity (the volume of clicks over time) you may need to wait a week or two before conversions rise and level off.
You may also need to manually adjust CPA targets -- especially during times of "natural" volume fluctuations, whether seasonal or news-event-driven. And for some reason, some of our campaigns would only level off at our target CPAs when we set them 25- to 75-percent higher than our "real" target. (I'm sure there's an explanation for this, and when we discover it, I'll tell you.)
Google's help files for Conversion Optimizer are very thorough, so I won't repeat their explanations and advice here. Do read through them, though, for nuances I don't have the space to cover.
Does it work? Yes -- perhaps imperfectly, but still very well. With some interesting wrinkles...
Some of you may be aware of problems associated with AdWords' Expanded Broad Match. If not, you can find a good discussion here, which also details steps devised by Matt Van Wagner and myself to minimize the detrimental effects. Basically we stopped bidding on one- and two-word broad match keywords, because AdWords was matching those to horribly irrelevant search terms, resulting in tons (yes, I mean tons) of expensive clicks that didn't convert.
So midway through our testing last year, I thought, "Wait -- if Google is responsible for capping CPA, will they somehow throttle the spend on those dangerous one- and two-word broad match keywords?"
So we tried it. Result: higher conversion volume at or below our target CPA.
But we tried it anyway. Result? Even higher conversion volume at or below our target CPA.
I hope you see similar results when you give Conversion Optimizer a try -- and please let me know how your campaigns fare (Tweet me!).
It's interesting to me where this will lead. Will it put vendors of automated PPC management software out of business? It might winnow out the weak, but smart companies have already started to add value to their offerings with tools and functionality that PPC advertisers love: campaign management and optimization, keyword research, etc.
Will CPA-based bidding catch on -- and will Yahoo and Microsoft follow suit? I think and hope so, and I believe it will lead to better PPC results as marketers' time is freed up to concentrate on optimizing aspects of their advertising that affect profitability even more than bid management -- like ad and landing page testing and optimization.
For all its near-perfection, Conversion Optimizer suffers from one major flaw that's common to almost every human and automated bid management I've seen: it attributes all the credit for the conversion to the action that directly preceded the conversion. Savvy marketers know that a conversion often occurs after multiple visits to the site. A typical chain of events might be:
- Visitor arrives on the site after using a broad search term -- but leaves before converting.
- Visitor comes back to the site after searching on a more-specific search term -- but still doesn't convert.
- Having remembered the name of the company, visitor comes back to the site after searching on the company name -- and converts.
Traditional analytics and human intuition concludes that the keyword associated with the last search term used is a great keyword -- exhibiting high conversion rates. That may be true, but other, broader keywords were responsible for earlier visits and deserve some credit for the eventual conversion.
While Conversion Optimizer is flawed in that it disregards the value of those earlier visits and associated keywords (and hence tends to under-bid on broad terms), its results are still very good, and much better than most mortals could achieve.
And there's good news ahead: several companies are rolling out tools that take the phenomenon (sometimes called "Latency", "Attribution Management," "Multiple Attribution Protocol," and a few other terms) into consideration.
Kenshoo Search factors in multiple attribution when it regulates PPC bid prices. ClearSaleing has recently added this functionality, and ClickEquations is rumored to be working on bid automation with attribution factoring. I've heard that other companies are working on similar tools, so expect a growing arsenal of great products that will help us all manage better PPC campaigns this year.