Thursday, August 27, 2009

Want To Lower Your Online CPA? Then Factor In Yahoo's Assist Conversion Data

For most search and display advertisers, reaching a low CPA (Cost Per Acquisition) goal is the holy grail of a successfull online campaign effort. But as most advertisers know from experience, it's easier said then done.

Getting more visitors to convert for less money is not an easy task by any means. There are a lot of factors that have to take place in order for CPA to drop. For example, search and display ads can't be too expensive, the ads need to drive quality traffic and be relevant, the landing pages have to be relevant, there can't be any roadblocks on the site's path to conversion, visitors must be engaged with your product, sign up forms need to be easy to navigate, shopping carts need to flow, etc., etc., etc.

As you can see, driving a lower CPA takes some effort. Too bad there isn't an easier way to lower that CPA dollar figure. Ahhhhh.....but wait. There actually is a way and it all boils down to how advertisers measure a conversion.

If we take a step back and look at the way that most advertisers measure a conversion, it usually goes something like this...

A conversion takes place when a visitor clicks on a banner ad or search ad, reaches the web site, and then signs up, makes a purchase, downloads something, etc. If the campaign receives a high number of clicks and a high number of conversions, then the advertiser considers the campaign a success. If the campaign receives a high number of clicks and a low number of conversions, the advertiser considers the campaign a dud and a waste of money.

So what is wrong with this description of a conversion? It's true isn't it? Yes, it is true, but the problem with this description is that the conversions mentioned above are single channel, direct response, conversions. Not all visitors who click on ads operate in a direct response manner. For example, many studies over the past few years have shown that on multiple occasions visitors will click on a display/banner ad, reach the web site and not convert, but later (down the road) run a search, click on the search ad, and then convert. This is what's called a multi-channel conversion (visitors who traverse across multiple campaigns before converting).

But when advertisers measure a conversion in a single channel, direct response frame of mind, the display ad in the example above would not have received any credit for driving the search conversion. If this type of scenario continued over time, advertisers would have considered the display ad a dud....it drove lots of clicks but didn't drive enough conversions on its own. This of course leads to a higher CPA. Due to a higher CPA, the advertiser would naturally kill the budget for that particular display ad.

In reality, what the advertiser has just done is kill off a very effective branding campaign for driving search conversions!

Because the advertiser was unable to give proper attribution conversion credit to the display ad that drove search conversions, they're stuck in the same boat of figuring out ways to lower the CPA for their display campaign. But it doesn't have to be this way. There are analytics solutions out there that provide multi-channel credit to ads that have been clicked on before the last clicked ad (direct response ads).

Yahoo's 'Full Analytics' (found in a Yahoo search advertiser's account) offers a multi-channel attribution metric called Assists. An Assist will attribute credit to an ad/campaign/keyword that contributed to the conversion of another ad/campaign/keyword. If a visitor clicks on a display ad and does not convert, but later converts off a search ad, the display ad will receive an Assist (conversion credit) and the search ad will receive the conversion.

What an advertiser can do with Assist conversion data is factor it into their regular CPA calculations to give them a better view of the performance of their ads. Here's an example...


Example 1: Single Channel - Direct Response CPA Calculation

An advertiser spends $100 on a display ad and it receives 5 conversions diectly (visitor(s) clicked on the ad and converted).

$100/5 Conversions = $20 CPA

Example 2: Multi Channel + Assist Response CPA Calculation

An advertiser spends $100 on a display ad and it receives 5 conversions on its own...but it also receives 7 Assist conversions that it drove to search.

$100/5 direct response conversions + 7 Assist conversions = $8.33 CPA

Results

Between Example 1 and Example 2, we see a CPA difference of $11.67. Depending on what an advertiser's CPA goals are, the Assist calculation should make a difference in how they manage their ad budget.

With Example 2, the advertiser realizes that the display ad brought in more than just 5 conversions on its own. So this contributing conversion data should also be factored into their CPA calculation.

With the data in hand, the advertiser would most likely want to spend more money on the display ad and expand its exposure (as it drives conversions across multiple channels) as opposed to pulling the budget or lowering the budget (because a $20 CPA was too high).

Using Example 2, advertisers realize that they are receiving more conversions for each dollar spent.


Conclusion

Without having to focus on issues such as web site optimization, ad quality, and shopping cart performance, we just showed how to drive a lower CPA by re-evaluating how conversions can and should be measured.

If CPA is going to be used as goal of success or failure, advertisers have to make sure that they are looking at a broader view of campaign conversion perfomance. Simply measuring CPA using a direct response caluculation will not cut it if you are running multiple types of ads (search display, email, etc). Direct response measurement does not provide enough insight into the true value of the ad. Advertisers need to make sure that they measure beyond last clicked (direct response) conversions and start giving credit to those ads that are responsible for driving conversions down the conversion funnel line.