My passion for digital advertising comes from one particular item: data. The amount of information that we have the opportunity to look through and analyze is astonishing. From basic metrics like impressions and clicks to more intriguing insights such as website navigation paths or adds to cart, everything seems to be in place for complex questions and problems to be tackled with ease.
Yet, my experience has shown me that some areas of digital marketing still need improvement. In particular, some problems that I once thought were solved by the technology we have, are still relevant and recurring in the work that I do. Here are three areas that have remained problematic in digital advertising.
1. Attribution: The million-dollar question
Imagine running an ad on a radio station or a television channel.
What tools do you have at your disposal to measure impressions, streams, or views?
How do you know the average time spent by an individual listening to or watching your ad?
No matter how great you are with numbers, obtaining that information is probably not worth the time it would take.
Digital advertising gives us the luxury of so much data that sometimes we do not even know what to do with it. That said, when it comes to determining how an ad has influenced a purchase decision, the paid media world still struggles. In many cases, I have heard brands say:
- “Yeah, our retargeting video ad on Facebook had 10,000 views, but how many conversions would have happened anyway?” or…
- “I am not sure if display advertising is actually making a significant impact.”
Here we are, after years of technological improvements, still evaluating with relatively little preciseness, the impact of a video or image ad.
To an extent, rigorously designed tests can give insights into the impact of different channels. This requires isolating audiences and running robustly prepared campaigns for a period of time that is long enough for data to be relevant. Your display ads may not impact the consumer in the most direct sense of the term, but it helps keep your brands top of mind. The consequences of taking it off the table could be a slow but steady decline in brand interest and consideration, or maybe not. Who knows?
2. Tracking: the other million-dollar question
This one is actually the technical issue behind the first problem mentioned above. Sometimes it is not the interpretation of the metrics that make our life very difficult but rather the fact that the metrics themselves are not available.
A user sees a display ad on a phone, then does some research a couple of days later on his desktop computer and then converts on his laptop computer. In this instance, it is hard to track and credit devices and channels appropriately. Yes, improvements have been made with some big platforms such as Facebook, which can use a mix of browser-based and user-based tracking, but one can easily notice that some data is lost in the conversion path. Sometimes Google Analytics would call the data “unavailable”, “not set” or attribute to “direct” what was actually not a direct visit to the website with no ad-driven incentives.
As we worry more and more about our privacy on the internet and as related policies are implemented such as CCPA or GDPR, tracking systems will encounter more challenges. An issue with one of our clients which we have suspected was also due to privacy concerns was that Google Analytics was registering new users in instances where we were able to conclude that it was actually a duplicate of a previous user. To conclude briefly on this point, while we have significant insights into the overall customer journey, tracking is not at its full potential and there are reasons why it will not get to that point anytime soon. For now, let’s say that we want to be grateful of the fact that we have enough data to make impactful decisions as marketers, even though the precision is not there yet.
3. Automation: I will let you attribute the dollar amount
The tricky thing about numbers is that their mathematical nature hides their lack of accuracy. The same is true for algorithms, which are very powerful by their names, but in practice, you can perceive their weaknesses. Automated optimization systems are effective as long as human behavior can be tracked in a logical way.
Automated optimization systems do one thing very well: They detect countable trends, and they do it very fast.
But if a key characteristic of humans is their ability to think logically, another key characteristic is their ability to get outside of logical trends. In marketing, we deal with humans. In other words, we should be very careful about excessively using tools that rely so much on the logic that they become counterproductive. If you have ever tried an automated strategy that has worked on an account and unexpectedly failed on another, it is because they are not perfect and still need strategic guidance, meaning inputs from a human who understands other humans and the reality of the market.
Some automated bidding systems such as Google’s target ROAS for example tend to fail at spending and become very conservative as soon as the ROAS goal is set too high.
A test with a client I work with has been to first determine a set of campaigns for which the actual ROAS for the campaign was much higher than the set ROAS and then lower the target ROAS in order to allow the campaign to capture more traffic. What we were expecting was lower ROAS, but higher spend and of course higher revenue. It worked, for some campaigns, and failed for others. This revealed to us that simply lowering the ROAS target was not a scalable strategy. Automation is not perfect, but…
THERE IS NO REASON TO LOSE HOPE
Considering the technology today versus what existed 10 years ago, no one can deny the improvements. Nothing is impossible and it is our job to keep pushing everything to the limit and achieve what was once considered impossible. Let’s maintain our dedication and keep on testing and researching.
Maybe you will be the one answering one of the million-dollar questions mentioned above, just don’t forget to share it with me!