Proximity matchback attribution

Key to robust attribution modeling is the breadth of data points available for analysis. Blending historical data points together is the basis of matchback analysis, and when used properly, it can guide your monetary and time marketing efforts towards those activities that generate the highest comparative value. 

Digital vs. non-digital attribution

Each of your campaigns will have a subset of attribution parameters.

For digital media you'll have likely set up UTM tracking, and you can further dig into location and technology-specific parameters. Using a variety of click-based attribution methods you can assign value accordingly.

For above the line media attribution becomes considerably more creative. You'll be relying heavily on voucher and vanity url use specific to your campaign, as well as more inventive means of matchback attribution. If your marketing campaign was the equivalent of a water droplet falling into a larger body, matchback attribution would be your methodology on how to measure the number and density of ripples the droplet generated.

Vouchers and vanity URL's have different conversion types

Vouchers and vanity URL's both attempt to force direct response attribution onto a non-direct response medium. Importantly, these tools measure very different things.

Vanity URL's measure attribution from browsing intent, where you've called out that a particular marketing campaign's call to action results in attributable visits. Additionally, you can view the visitor flow funnel for uses of campaign optimization. The micro conversion path for a vanity URL will result in a series of individual page views, and the associated actions between each step.

Vouchers measure attribution from actioned intent. In other words, you attribute 100% of a macro conversion to the sale, and that's it. Vouchers do not give you particularly extensive insight into visitor actions in the upper part of the funnel.

Given the differences in conversion types between the two tools, their respective data points must be normalized in a structured way such that you can have actionable cohort results. This is where your business would develop an attribution decay model, where a series of attribution rules define the final attribution priority of a marketing campaign. At a basic level, it is typical to define attribution priority from all unattributable revenue to vouchers first, and then divide the remainder unattributable revenue on a per-visit basis such that each vanity URL visit is assigned a micro conversion value. I'll discuss more about this in the future. 

And yet, vouchers and vanity URL's only scratch the surface of matchback attribution.

Matchback attribution

After you've accounted for the indirectly attributable voucher and vanity URL conversions, you must decide on a methodology for proximity matchback attribution.

The reason I define it as "proximity based" is because you constrain your attribution into a series of rules that measure the likelihood of a conversion originating with a particular marketing campaign. The likelihood of conversion (or degrees of statistical significance) are meant to be a function of how close the conversion was to specific marketing touch points.

The reason it is termed matchback attribution is because you are reattributing value (based on proximity) to a source that would have otherwise not accounted for a certain conversion value. You can match proximity by location, device, incoming URL... the possibilities are endless, all it takes is logic on what best defines your marketing campaign's visitor.

Matchback attribution in practice

My favorite example is that of a mobile bus advertisement. Here's the scenario:

Advertising Bus
"Your company decided to purchase an old clunker of a bus at a bargain $9,000. You completely cover the bus in advertisements at a cost of $10,000, and invest an additional $1,000 in a crazy sound system and a disco ball to attract further attention to the rolling advertisement."

At a fee of $20,000, the mobile advertisement is the same cost as some of your single-day newspaper campaigns, or equivalent to a 2-3 week spot on an actual local transit bus. How do you justify your spend?

Your ammunition for matchback attribution is:

  1. A voucher code on your bus
  2. A vanity URL on your bus
  3. GPS coordinate logger

First, you immediately attribute any voucher sales (macro conversions) to the bus.

You then figure out the value of a single unattributable visit, and attribute a micro conversion value to each vanity url visit.

Next, because you have a GPS coordinate logger you can apply rules around matching site visitors to those visitors seeing your bus marketing campaign. Since you have the route and time of your bus advertisement, you can plot direct visits, branded organic search, and branded SEM search IP addresses against a map of where the bus was. Next you'd define a time threshold with which you're comfortable reattributing visits to the campaign, such as 5-10 minutes as is common for reattributing visits on TV advertisements. This is proximity matchback attribution.

Finally, because you've seen direct conversions, you can reattribute remaining unattributable revenue by matching proportional unattributable revenue against marketing campaign spend. So that you're calculated about this, you'd only want to reattribute unattributable revenue generated on or after the date of your bus campaign.

 

Brilliant work - such a wide array of touch points for proximity matchback attribution! Next, define how to attribute revenue for a Lamborghini advertisement such that you can write it off from your frozen yoghurt business...

Lamborghini Advertisement