November 15, 2019
Incrementality Testing: What You Should Know About This New Marketing Tactic
AdWords recently announced the global rollout of a new conversion attribution modeling feature within Search Funnels (located in the Conversions section under Tools). For most knowledgeable PPC managers, this was not a question of if, but when. The new AdWords feature represents an important step forward, but still lags behind what third party PPC platforms have been doing for years.
What is Conversion Attribution Modeling?
Conversion attribution modeling can take on many different meanings. In short, it references the ability to attribute credit across various channels, devices, and clicks to determine what drove an actual sale. This information is extremely valuable for marketers so that they can determine which channels to invest in more actively.
For a more advanced understanding of the different types of multi-channel attribution models, this blog post at Occam’s Razor does a masterful job. For our purposes, we’ll focus on the issue as it pertains to paid search.
In AdWords, Google gives all the conversion credit to the last keyword with a click before a conversion occurred. Let’s say that Tito’s Ski Lessons is evaluating their PPC performance. Take an instance where someone searches: “ski lessons” and clicks on an ad, doesn’t convert, comes back four days later, searches “best ski lessons in Tahoe” and clicks on an ad, doesn’t convert, and then finally searches “Tito’s Ski Lessons” and converts. AdWords would credit the keyword Tito’s Ski Lessons with the conversion. However, in this instance, the non-brand and generally more top of funnel keywords ski lessons and best ski lessons in Tahoe clearly contributed to the eventual conversion.
What ends up happening is that many advertisers overvalue brand terms, which are already bid into top positions. Many non-brand terms, which represent the first click toward an eventual conversion, appear unproductive and may get bid down or paused.
Conversion attribution models provide a way to conduct analysis and to determine what credit should be given to the keywords that helped out along the way.
What Do You Do with This Information?
We have been using the Kenshoo platform for years in large part because they have one of the earliest and most sophisticated conversion attribution modeling platforms.
The reason that it’s so important to incorporate attribution modeling into PPC optimization is that it allows advertisers to more effectively set bids for non-brand keywords. Let’s take a U-shaped attribution model, which favors keyword clicks that initiated and closed the conversion funnel:
|Best ski lessons in Tahoe||Click||.2|
|Tito’s ski lessons||Click + Conversion||.4|
If you were to look in Tito’s AdWords account, over time there would likely be a ton of clicks and spend associated with ski lessons, but at a very high CPA. In the Kenshoo profile for that account, we will be able to see the real CPA after incorporating all of those fractions of conversions.
If that keyword spent $200 for 1 conversion, you are going to treat it (read: bid it) much differently than if it spent $200 and has 2.8 conversions.
How Much Will This Impact Your Account?
That really depends. The more clicks in an average conversion path, the more helpful this data becomes. We’ve typically seen that larger, considered retail purchases yield the highest number of clicks.
You can easily see how many clicks on average are associated with a PPC conversion by going to the Search Funnels section and looking at the ad clicks per conversion metric. Anything higher than 1.2 (every 5th conversion contains multiple clicks), means you would definitely benefit from stronger insight into the click conversion path.
“Without knowledge, action is useless and knowledge without action is futile.” – Abu Bakr
While the new data in AdWords should be celebrated as a step in the right direction, it is provided in a siloed report. Having this data is only as good as the ability to take action on it.
Without the ability to change how conversions are reported in the grid, this knowledge is much less actionable (and by Abu Bakr’s calculation, futile). Conversely, pausing and bidding down a bunch of non-brand keywords without proper data is not only useless, but actually somewhat reckless.
AdWords is already introducing more customization to their conversion tracking so it’s likely this is a change we’ll see in the next 1-2 years. Until then, it is worth considering a platform or agency that splits up conversion credit and uses that more accurate data to drive better CPAs and ROI.