Unattributable revenue is a common frustration for marketers and managers alike. After all, how do you know which campaigns to scale if some of your best performing cohorts lack attribution? Direct traffic, referral traffic, social traffic etc. - different companies deal with these mediums with varying degrees of confidence.
The methodology I frequently revert to is a weighted reattribution of unattributable revenue to the most likely source of acquisition.
Why reattribute revenue?
Your unattributable traffic comes in a variety of shapes and sizes. Here are a few of the most common:
- Direct traffic
- Referral traffic
- Brand organic search traffic
- Brand SEM traffic
- Traffic with "broken" tagging
On their own, these traffic sources often outperform other cohorts, as the nature of the traffic is primed for interaction with your brand. However, the truth is that most of these visitors would have not existed without some form of supporting marketing-based stimuli.
Unfortunately, some of the most expensive advertising mediums are frequent victims of poor attribution. A few examples include:
- Newspaper & magazine advertising
- Billboard advertising
- Public transit advertising
- Direct mailing campaigns
- Radio advertising
- TV advertising
A few of these have their own basic but generally accepted methods of attribution, such as post-code validation for direct mailing campaigns. But how would you measure success when your sale is of an entirely digital product with no credit card required for transacting? The answer lies in blending the data you've collected.
Blending data for reattribution
You have a variety of data points at your disposal related to campaign transactions, such as voucher code use patterns and visits through your vanity url's. These provide the basis for reattribution. Constrained against a pre-defined time period, the data points you'll need are as follows:
- Cost of 1 advertising campaign
- Total purely attributable revenue for 1 advertising campaign
- Total cost of all marketing campaigns
- Total revenue from all unattributable sources mentioned above (and any others you have)
In its simplest form, the methodology matches weighted unattributable revenue to the proportional spend of a campaign. For every $1 of campaign spend through "unattributable" sources, you generate $X in proportional untracked revenue. Here's an example:
- Single campaign marketing spend: $1,000
- Toal purely attributable revenue for the campaign: $600
- Total marketing spend: $15,000
- Total revenue from all unattributable sources: $30,000
The campaign accounts for 1/15 of the marketing spend. You multiply the weight of the campaign within your acquisition efforts by the total unattributable revenue. In this case 1/15 of $30,000 equals $2,000. Consequently, the reattributed revenue for this campaign is $2,600 ($600 directly attributable + $2,000 reattributed). You now plot this against your campaign profitability thresholds to see whether the campaign pays for its share of overhead expenses, as it has already covered its $1,000 marketing spend.
Revenue reattribution limitations
This methodology isn't perfect by any means. It punishes campaigns generating zero purely-attributable revenue, and potentially overcompensates for those campaigns that show even the most marginal suggestion of monetization.
However, the methodology at least begins to tackle the issue of untracked cohort revenue with arguably minimal effort and zero technical integrations. The methodology can be particularly useful to businesses not looking to invest six-figures into attribution software, which itself, is based on assumptions.
Good luck reattributing!