Attribution Part 2: Source Codes and Reconciliation

This is the second part in a series on attribution. For an explanation of what attribution is, start out with Part 1 here.

Ok, now that you know what Attribution is, this post will show you how to get started. This is where most companies should begin if they are currently not doing anything. Usually at this point you have some kind of customer relationship management system (CRM) so that you can track what happens to leads, and the deals that arise from them.

In your CRM system you want to set up a custom field called Marketing Source Code. The iron clad rule from this point going forward is that no list of leads will be imported into the system without a value in this field. That’s the easy part, more challenging may be updating any forms or other integrated tools that are creating leads in the system for you.

I’ve found that it’s best to keep this field hidden from all but marketing users. It’s not that it’s a big secret, but the important thing is that you leave it locked (or read only) depending on how your CRM system works so that it never gets changed.

If you already have thousands of leads in your system you may also want to consider a data cleanup project where you fill in the marketing source code for your existing leads, although this is not critical because you will also want to create a field called Deal Source Code that will be required for any new deals created by the sales team. In most CRM systems Deals are usually an object of their own (a record in the database) they are not just fields added to a contact record. As a result you’ll want to know the source of each and every deal that is associated with your contacts.

Initially you’ll want to create this as an open text field. The idea here is that when a sales rep is working a lead and gets to the point where they are going to add a deal to the pipeline they will give it a name, and set the forecasted close date and amount. Deal Source Code will be a required field on this Deal creation screen and the rep can put down any information they have (the more granular the better) on where the deal come from. After your first 30 deals or so you’ll want to review what’s been entered and start to work on creating a pick list for the most common options.

A best practice here is to keep a granular feed that can always have open text for description, but also create strict categories for classification on top of that. For example, the code entered by the sales rep may be “June 2020 Vegas Sales Boondoggle Event”. This record could also have the Marketing Source Code of “Vegas Event 2020 Booth Scans”. Over time you may also use fields such as Source Code Minor Category: Trade Show Booth Scans and Source Code Major Category: Trade Shows.

The big idea is that the Major Category will map to your marketing budget categories so you can easily see which programs are doing better. With the minor category you can get more detailed reporting on which specific marketing programs in the category are most effective. Many organizations will make statements about major categories “Webinars work very well for us”, but more agile organizations are able to say “These types of webinars do well for us” and can easily compare things like how webinars they run compare to webinars they pay a third party to run and provide registrants for.

This is always a work in progress. You’ll want to be reviewing closed deals quarterly or better to check your data to make sure it’s valid (confirm if the Deal Source Codes match back to the Marketing Source Codes – the odds get lower the more complex the sales cycle). Until the Deal Source Code list gets solid you’ll want to be doing reconciliation with the sales staff “Hey, the Thompson deal? Where did you meet the VP?”

While would like to tell you “Hey! Congratulations! You’re doing attribution!” we’re not there yet. If you’re using source codes and everyone is happy then you could say that you’re doing attribution at a cocktail party at a marketing event and probably get away with it.

The bad news is that the first time the system fails you, that’s when you’re going to be doing the real work of attribution. You’ll be looking back on a closed deal with one of your sales reps and your Marketing Source Code will be the above mentioned Boondoggle event, but the sales rep will say “Actually, she was also at the Fall VIP event, that was really where she said she had budget and wanted to get a deal going.” This is the point where you’ll have to decide on a more complicated structure. Do you want to attribute half the deal value to each? Many marketers would love to attribute the full value to each, I’ve found that Sales Execs and the C-Level suite have a big problem with double counting. Making all results suspect at best, elimination of your position at worst.

As you attribute a deal to multiple sources you reach a point where individual source code fields no longer scale. One solution is to use a CRM system that tags contacts with all the campaigns they have taken part in, this can get you usable data for multi-stage attribution. You also start getting into frameworks to calculate the value of each touch. Do you consider any of the previous sources if there’s more than 6 months gap to the next touch? Do you disregard the older touches? Discount them? Prioritize the last touch? You can make arguments for all of these and if you want to go down that rabbit hole check out the previous post and dig into the Avinash material.

There’s also another option using Machine Learning to validate which touches are most relevant, and that will be the topic of the third and final part in this series!