Are you in a bad relationship with your attribution model?
The correlation between relevance and response is strong in every channel marketers use to communicate to clients, and has been for years. As we increase the data we use to drive our advertising and marketing, we see an increase in customer engagement. Although eMarketer confirms that data-driven marketing is on the rise, some companies see a negative correlation between relevance and profitability.
There may be several issues at play here, from the cost of using more data to the time it takes to put together a good content strategy for a data-driven marketing campaign. One of the largest problems actually has to do with a bad habit we have as an industry: the use of last-click attribution methodologies to measure performance.
For marketers, it is terrifying to modify attribution methodologies. Decisions made, assumptions held, theories tested and budgets planned all break down once a new attribution methodology is introduced. Strategies that might have worked well in the past are questioned (similar to those relationships that seem so great at the beginning). Reputations are on the line and, quite frankly, jobs are at stake with these decisions. If we were honest, however, we would realize that decisions based on last-click attribution are simply poor decisions. It’s not you; it’s the methodology that is failing.
So, how do you become the change agent in the organization without damaging anyone’s reputation? As with anything worth doing, it starts with the fundamentals:
1. Good attribution starts with good customer identification. You need to have a consistent and accurate view of your customer, across ALL channels, to understand how your marketing works. We live in a world where our customers interact with our brands at their convenience and on multiple devices. This means that many times, a customer will receive a message through one channel, such as a mobile ad, and finish the transaction in another. Without a consistent understanding of what was seen by your customers and how and where they responded, it becomes nearly impossible to discern channel efficiencies.
2. Every marketing event has a shelf life. A display ad seen three weeks ago does not have the same pull on a transaction as the video seen today. Marketers need to take into account the order curve of the marketing event and allocate accordingly. While the email seen two days ago may have had a pull on the conversion, it is not as strong as the event that generated the click-through today.
3. Not all channels are created equal. People respond to marketing from different channels in very different ways. Consumers spend more time with catalogs from their favorite brands than they do with the marketing emails they just received. They may spend more time with the video ads they watch online than they do with the same ads shown on TV (that they likely fast-forwarded through on their DVR). A good attribution strategy will take this into account.
4. Marketing channels work together. Contrary to popular belief, the marketing funnel is not dead. Instead, consumers are moving through that funnel faster and through more channels than we have seen in the past. The trick is to understand how much sway each event has made on the final conversion.
5. Attribution is a data management problem, not a modeling problem. Fundamentally, models require data to operate. They use math to determine correlations between large data sets. To make sense of it all, models need to make massive assumptions and group together sets of data. But the conversion is a very unique event. Every transaction is different, and consumers are taking so many different paths to conversion that the model’s assumptions break down. Attribution needs to occur at a customer level first, then be aggregated to see the bigger picture.
While jettisoning your current strategy seems like a terrifying risk, what’s worse: knowing you’re wrong and doing something about it, or letting fear hold your entire marketing strategy hostage? If you can get the attribution done right, you will see the results that you expect from a data-driven marketing strategy. Attribution is a process that starts at the individual level, just as targeting is about being relevant to the individual customer. When you get the fundamentals aligned properly, your marketing performs better, your decisions become clearer, and you’ll make more money.