Direct response advertisers take heed. You are doomed if you're still relying on last-click methodology to evaluate your media buys.
That's a strong statement, and it doesn't necessarily apply to all direct response advertisers. Somewhere out there, there's a low-consideration, high-margin product that can probably continue to use this model and derive some measure of success from it. But for the majority of DR advertisers, the last-click methodology can't continue to work. Here's why.
Just about every DR advertiser takes a "test and refine" approach to digital advertising. Typical commitments for testing a given media property tend to be in the $10-20k range. Assuming a $5 CPM (just to make the math easy), a $20k test buy will let an advertiser sample the property with a run of 4 million impressions.
Typical click-through rates (CTRs) might approach 0.1 percent, and perhaps be lower for certain creative executions that are over-exposed all over the web. Assuming the 0.1 percent figure, that $20k will drive 4,000 people to the advertiser's website to consider the offer. Measuring on a last-click basis, this means only a handful of people will convert, and the test buy will be judged on a basis where a few more clicks or a small number of conversions will determine whether or not the buy was successful.
In the early days of digital advertising, the differences in last-click effectiveness between test buys were significant, and the judgments concerning which media properties to scale up and which to cut from the buy were clear as night and day. However, on the modern landscape, certain market forces are leading to a homogenization of CTR performance. Among them:
- The proliferation of ad networks -- This one has been discussed over and over, and it's hard to bring it up without beating a dead horse. Suffice it to say, how can last-click advertisers expect to see meaningful differences in ad network performance when many of the networks they choose are selling the same inventory? In order to have a shot at seeing deltas that can be exploited, an advertiser needs to buy networks that have truly differentiated inventory.
- Scale (or lack thereof) -- There are a limited number of ad properties that can devote the volume of inventory necessary to be successful at scale, given the numbers I cited above. Most DR advertisers know this and limit their tests to properties that can generate scale for them if their test is successful. Even in the case in which a media property believes it can scale a DR buy, it often falls short when it runs out of inventory that reaches the right prospect. They loosen the reins on targeting, campaign performance suffers, and the advertiser moves on.
- Dropping CTRs across the board -- Simply put, people just aren't into clicking on banners much. This effect has been constant, with CTRs gradually declining over a long period of time. Now, we see reports claiming that most ad clicks are generated by a very small group of people.
Some might look at this evidence as an argument to support exchange-based media buying, or as an exercise to understand the real difference in inventory quality between ad networks. I see those things as short-term solutions to a bigger problem.
The truth is that if you're evaluating effectiveness with last-click methodology, you're focusing on a very small slice of your success. A DR advertiser will see more conversions from people who did not click on ads -- the result of the mere ad exposure. If 20,000 ad views result in one click-based conversion, it's probably playing a role in generating four or five more view-based conversions, in concert with the rest of your ad buys.
Understanding how to optimize these view-based conversions and the combinations of ad exposures that drive them is the premise behind engagement mapping, which is an exercise just about every DR advertiser should undertake.
Without this understanding, many DR advertisers quickly hit a brick wall with their tests. They scatter $10k and $20k media buys all over the web and find that the differences in performance between these tests either aren't significant, or they fall apart at scale. Buys are canceled, ad reps shrug, and the advertiser is left with nowhere to go until they look at performance through something other than a last-click lens.
If you find yourself hitting a wall like I've described, I'd suggest that you attend ad:tech New York, where the Microsoft Advertising Institute will be conducting a session that can get you up to speed with engagement mapping and its concepts.
Tom Hespos is the president of Underscore Marketing and blogs at Hespos.com.
On Twitter? Follow Tom at @THespos1 or @_MarketingLLC. Follow iMedia Connection at @iMediaTweet.
