Attribution reports and results can be complicated and can take a lot of energy just to set up correctly. Once your marketing team has the results, how do you use them to affect and influence your future campaigns to bring in more conversions? Just shifting a marketing plan can cause a large headache for your team.
So how can you effectively change your company’s plan to put the results to work? Below are a couple of strategies I’ve seen work for many companies when they’re acting upon their attribution data.
Campaign Level Analysis
Whenever I’m performing analysis, I start the exercise by asking myself the question, “What was my goal when I began setting up attribution?” Most often, I’m not looking at making wholesale changes to my overall marketing program, but I’m tweaking specific campaigns to make them as effective as possible.
One of the biggest stumbling blocks in attribution modeling is understanding how one campaign can affect another. In other words, if you take budget from one campaign to add it to the other, will the overall conversions between the two still go up?
I think the best approach is to look at two campaigns week over week for some time. Each week, change the budget of one of the two to see what the results are.
Over time, by comparing the effectiveness of the campaigns week over week, you can start to see trends on how one may affect another.
One thing I get push-back on when recommending this strategy is that the sample size is too small to make meaningful adjustments. Though I understand the trepidation, an attribution software in today’s business climate needs to be agile enough to provide daily information.
With so many other factors like unmeasured marketing, seasonality and other external elements at play on a daily basis, you need to keep your analysis windows short. Otherwise, too many variables will come into play that may undermine your analysis.
This is also why I recommend making your analysis on the campaign level, rather than on the individual placements. Generally speaking, placements or individual ads will not see enough traffic to allow a marketer to make an informed change.
Looking at the campaign as a whole, however, can be a more statistically relevant marker.
Testing On Various Levels
If you do have placements with enough data (in my opinion, that’s 3,000 conversions per view period), then they are very ripe for acting upon, as well. In this case, take care to measure placements together in the same campaigns or campaigns in the same strategy or strategies in the same channel.
It’s easy to go down the rabbit hole and try to measure the performance of search keywords against display placements, or a mobile campaign against a social campaign. But because each reaches a different audience at potentially different times, taking action can be a bit difficult when trying to run an isolated test.
I tend to choose two or three different items at each level, and, as I’ve mentioned above, look at their relative performance week over week as I raise or lower spends to find the optimal balance between all of the players in my game.
Adding Or Removing Placements
Attribution analysis will quickly reveal players which seem to be adding next to no value to your marketing efforts or who have CPA numbers off the charts compared with some of your better performers. Of course, your team worked hard to develop and implement those placements, so should you just remove them outright?
I think again that the answer is to test to see the impact on your marketing as a whole. There are always a few questions you must answer before removing a player from your campaigns.
Are other placements eating up a much bigger portion of your budget? Is much of the activity of this particular player going to your baseline numbers?
If you do determine a player is worth removing, I definitely recommend making incremental changes. Then, watch how the other players trend as players are removed or reduced.
The same is true when adding new players to a campaign or new campaigns to a strategy. Making wholesale changes will not allow you to analyze how your marketing is changing as a whole.
And when one new player is inserted, take care to understand how the other players suffer or thrive as you introduce a new player. Again, if wholesale changes are made all at once, you can no longer compare them to your past games.
As mentioned by fellow columnist Lewis Gersh in a recent article, it is very important that you use your marketing attribution reports to solve the problems they were meant to solve.
Assuming your attribution software is robust enough to provide updates daily, analyze it daily, make incremental changes to your programs and watch the trends over time for the full effect of what you can gain for marketing attribution.
Comments