Attribution Management: Good Theory or Good Practice?
Most businesses that advertise online set some rule as to what they can and cannot afford to pay for a conversion. A common practice is multiplying their average sales price times their average margin and setting that as the maximum cost per acquisition that can be allowed. In our example below, we are looking at Joe’s TVs, which has an average sales price of $1,200 and a margin of 35%, which leaves $420 to acquire a sale at breakeven. Any advertising source producing sales that cost more than $420 is in need of change. Below you will find an example of two keywords from JoesTV.com, one for the general keyword ‘HDTV’ and another for their branded term, ‘Joes TV’. The two keywords have combined to produce 86 orders, and within those 86 orders, 12 orders were produced as a result of a Purchase Path (two or more ads) beginning with ‘HDTV’ and closing with ‘Joes TV’. With these rules in place, let’s take a look at the performance of their keywords to determine which are acceptable and which are not, and then take action on which of these keywords works.
The first view of the keyword set below is using your traditional last click model. The keyword ‘HDTV’ does not perform below the target $420 CPA, so that keyword is up for elimination:
| Last | ||||||
| Keyword | Conv | Convr | CPA | Revenue | Profit | ROI |
| Joes TV |
58 |
2.22% |
$33.74 | $69,600.00 | $22,403.25 | 1145% |
| HDTV |
28 |
0.40% |
$563.23 | $33,600.00 | $(4,010.52) | -25% |
| Total |
86 |
0.89% |
$206.13 | $103,200.00 | $18,392.73 | 104% |
In this next example, the data displayed is using an even attribution model, meaning that both keywords are being given even credit for the conversion. The keyword ‘HDTV’ is still costing above the targeted $420 CPA, so that keyword is up for elimination:
| Even | ||||||
| Keyword | Conv | Convr | CPA | Revenue | Profit | ROI |
| Joes TV |
52 |
1.99% |
$37.63 | $62,400.00 | $19,883.25 |
1016% |
| HDTV |
34 |
0.49% |
$463.84 | $40,800.00 | $(1,490.52) |
-9% |
| Total |
86 |
0.89% |
$206.13 | $103,200.00 | $18,392.73 |
104% |
In this next example, the data displayed is using an even attribution model with exclusions, meaning that we have selected to exclude the branded term from credit since the user initially found the site by searching for ‘HDTV’. We’ve found that users search a branded term at the end of a Purchase Path for navigational purposes, so we exclude it from receiving credit. The keyword ‘HDTV’ is now performing at an acceptable CPA and is producing a 7% ROI:
| Even w/ Exclusions | ||||||
| Keyword | Conv | Convr | CPA | Revenue | Profit | ROI |
| Joes TV |
46 |
1.76% |
$42.54 | $55,200.00 | $17,363.25 |
887% |
| HDTV |
40 |
0.57% |
$394.26 | $48,000.00 | $1,029.48 |
7% |
| Total |
86 |
0.89% |
$206.13 | $103,200.00 | $18,392.73 |
104% |
In this final example, the data displayed is showing that if you use the metrics in the first and second example you may move forward with eliminating the ‘HDTV’ keyword, which will then result in a large loss in revenue and profit that could have been prevented by using an even attribution model with excluded terms.
| If HDTV Eliminated | ||||||
| Keyword | Conv | Convr | CPA | Revenue | Profit | ROI |
| Joes TV |
46 |
1.76% |
$42.54 | $55,200.00 | $17,363.25 |
887% |
| HDTV |
0 |
0.00% |
$ - | $ - | $ - |
0% |
| Total |
46 |
1.76% |
$42.54 | $55,200.00 | $17,363.25 |
887% |
A lot has been written about the theory of attribution management, but very little has been put into the practice and types of results it yields. The previous examples took you through a scenario with a view of a campaign, and another view of that campaign where attribution management is applied. As you can see, attribution is not only a good theory, but a best practice that all online marketers should implement sooner than later if their goal is to increase profits.