Back to the Learning Center

By: Casey Grimes on February 1st, 2023

Preparing Your Data for Attribution: The First Step in Your Journey

As we kick off 2023, it’s not uncommon for marketing leaders to start enacting their plans for major initiatives for the year. And more and more companies have been asking, “Should we implement marketing attribution?”

That said, you might feel a little lost if you’ve tried to read up on whether it makes sense for your company. One of my biggest frustrations around marketing attribution has been how we, as marketers, talk about the process. The conversation tends to go like this:

1. Purchase an attribution solution.
2. ???
3. Talk about attribution models and how to structure reports.

This trope is so common that multiple blog articles follow this pattern: an introduction to attribution, some hand-waving, and then an extensive discussion of attribution models. The reader is left with some knowledge of models, but the articles omit critical components that provide foundational understanding.

Those question marks for Step 2 are extensive and should be broken into about 20 different steps. Even beyond that, you’ll need a Step 0 for readying and structuring your marketing department for attribution success. When starting an attribution journey, talking about attribution models is like worrying about what custom color you’ll paint your car before you even consider purchasing a vehicle.

So, if you’re at the point where you want to invest in marketing attribution and revenue accountability, where do you start?

You need to go through the work to set up a customer journey.

If you’re on the road to attribution measurement, this might be your biggest requirement: Make sure you implement the work you need to have an accurate customer journey so you can have the most accurate attribution data possible.

A customer journey, or lead lifecycle, is simply a measuring and reporting tool to understand the velocity of a record as it moves through a company’s interactions, from prospect to customer and beyond. The system only listens and reacts to the marketing, sales, and customer success practices already in place in an organization. This is often why we consider deploying a lead lifecycle as only the first half of the work that needs to occur in consulting. Once a lifecycle is let loose in the “real world,” it will expose non-standard practices and unexpected behaviors that weren’t considered during its development.

It’s not uncommon to find new challenges post-lifecycle deployment, even if extensive interviewing and prep work are done. Two common examples include finding inside sales prospecting with unexpected lists or a partner marketing program with different data needs. The measurement of records brings up new challenges for standardization and alignment that need to be handled before your reporting is “accurate”—or at least aligned with the company’s common expectations of behavior.

Something similar happens when an organization decides they want to make a serious effort at handling its attribution data. Once a solution is deployed, the same problems will occur: They’ll be unable to understand what behaviors are touchpoints on the customer journey to measure and have no way to discern their impact.

We strongly recommend any company serious about implementing attribution go through the process of setting up a customer journey measurement program first. The prerequisite work of establishing marketing and sales alignment and having a clear and commonly-shared understanding of the customer journey will speed up your attribution preparedness. You’ll also benefit from having time and velocity as an additional layer for attribution analysis, allowing you to make better-timed decisions as a marketing leader.

You need a single view of your prospect or customer.

Whether it’s a lead lifecycle, attribution, or any other form of measurement, your data will need some way of understanding who an individual is. This may sound straightforward, but it can be a major consideration for some businesses.

In the world of marketing automation, there’s a drive to have a single person as the “record of truth”: After all, if you have three copies of “test@example.com” in your database, it can be difficult to see which version of the record receives an email, fills out a form on your website, and so forth. The activity of what someone participated in gets strewn across different copies of the duplicate. Where possible, companies that want to embrace attribution data should find some way to distill who an individual is to measure their behaviors. In most cases, this comes down to an effort to de-duplicate your database of prospects and customers.

However, while some companies may be able to perform deduplication, other structural problems, especially for enterprises, may prevent marketing from having One Record for One Person.

For example, we work with a major enterprise that can sell to multiple layers of an organization: They may have contracts with an individual grocery store, a grocery store’s regional distribution center, and the grocery store’s corporate headquarters all at the same time. These are considered different accounts by this company, as their billing and accounting system treats them separately. Thus, if someone is responsible for an individual grocery store and the regional grocery distribution, their record is duplicated to show up for each account. This can’t be structurally changed, so other work needs to be done to ensure every version of a person’s record is aligned, regardless of the account it is attached to.

Defining what makes a person a discrete record (such as “unique email and phone combination” or “unique name and email address”) will help you determine how you want to define an individual. That definition can then be applied to handle alignment and measurement, which can be motivations in and of themselves. For example, even if you’re not ready for attribution, you’ll still want to be able to keep track of consent and messaging preferences across multiple versions of a duplicate record. Once you know who is 1:1 a unique person, measuring their behaviors and providing attribution across the customer journey becomes far more straightforward.

(If you need your intentional duplicates aligned and acting as one unified data layer, I’ve heard a certain consultancy can do that)…

You need a unified way to talk about your marketing campaigns.

One of the things we talk about in marketing operations is how critical data hygiene is to your marketing efforts—standardizing the way you write customer data for things like geodata, lead sources, company names, and so on. We do this for two major reasons:

1. To simplify our work, such as being able to select “Country: US” instead of “Country: US, USA, United States.”
2. To help machines that will not know that “US,” “USA,” and “United States” are the same thing—even if humans know that.

Standardization, in other words, doesn’t just help our work—it helps technical processes know what data points are the same. While that can be managed within one database, what happens when you try to track commonalities across your systems?

Let’s take a look at a real-world example: hosting a webinar.

  • You send out your first invite email to the webinar with the UTM structure ?utm_source=marketo&utm_medium=email&utm_campaign=webinar-name-here
  • When someone fills out the form to register and are brand new, they get a Lead Source of Webinar and a Lead Source Detail of Webinar Name Here
  • The program in Marketo hosting the webinar is named 2023-03-WBN-Webinar Name Here
  • The campaign in Salesforce linked to the Marketo program needs to be named a little differently due to sales’ using campaigns to call behind, so it’s titled Webinar – Webinar Name Here. Because it had to be created separately, the Channel in Marketo is listed as “First-Party Webinar,” but the Campaign Type in Salesforce is “Webinar.”

While humans can understand that webinar-name-here, Webinar Name Here, and 2023-03-WBN-Webinar Name Here are the same thing, machines—including those that power marketing attribution—will not be able to do that sort of fuzzy matching well. Therefore, it’s important to work within your marketing department and with your organization to develop a standardized way to discuss campaign efforts. So, how do you tackle this on a practical level?

  • Get organizational buy-in on how you talk about campaigns: Having one centralized place to describe and name campaigns helps prevent many naming issues before they start. Using a tool like Workfront or your project management system, you can synchronize each component of your campaign to use names and labels that keep your data aligned—so, for example, both your digital analytics team and your Marketo power user know to use “2023-03-WBN-Webinar-Name-Here.” Pre-emptively providing this information as part of a marketing brief reduces the need for your teams to guess what makes the most sense for their part of the campaign.
  • Create naming conventions that scale for all aspects of a campaign: If you’ve used Google Analytics, you may have noticed that the tool treats capitalization as a unique item. “Facebook,” “facebook,” and “FACEBOOK” are all considered separate sources—and that’s before you get to things like “FB” or other variants. As a result, you’ll want to set a naming convention and standard that keeps to one specific standard regardless of where it’s used.

If you call a campaign “2021-06-WBN-Webinar-Name-Here,” you’ll need that label to match everywhere possible to draw larger insights. You should aim to have your campaign standardized and written in the same way on:

    • Lead Source Detail (a secondary field that describes the exact Lead Source)
    • Acquisition Program Name
    • Marketo Program Name
    • Salesforce Campaign Name
    • Analytics platform campaign name
    • Other third-party platforms that need alignment

Think of your campaign’s name as a common denominator or key that can be used across platforms to understand campaign performance and attribution. You’ll want to come up with a standardized naming convention that works with the limitations of each of your platforms and can be easily understood and adopted by anyone impacted by campaign data.

Are there other ways to align your data between systems that don’t require this method? Of course. However, the most straightforward, consistent, and easy-to-implement method is simple: Use the same name, written the same way, everywhere.

  • The best naming system is the naming system everyone will use: One of the issues I’ve seen in companies trying to adopt naming standards is that there’s usually one department pushing for standardization—and other departments, who aren’t affected by data discrepancies, don’t have buy-in. When standardization begins, there’s often a drive to cover all potential scenarios and account for edge cases. However, this can result in a reluctance to adopt a new naming system. If you can’t remember whether you include a content category or certain ID in the name sometimes, it’s probably easier to go with “good enough.” Your naming system should feel intuitive to someone who’s had the campaign explained to them in 30 seconds. Otherwise, it’s probably too complex.

You need to ensure your best practices are consistently set up and accomplished.

This last part may sound a little silly, but it’s the biggest problem I see when talking to marketing departments: If you want to measure something, you must give yourself something to measure consistently.

While this isn’t true universally, a common problem can occur when multiple people work in a platform like Marketo: consistency. As a basic example, if you were going to send a one-off email/landing page as a marketing campaign, there are quite a few ways inconsistency can cause problems:

  • Marketers may differ on what counts as a click in an email or whether to use a status like “Bounced” for campaign members
  • A campaign in marketing automation may not be properly synced to the CRM
  • Analytics tracking parameters on email links may be missing or incorrectly set up
  • The landing page may be missing a web analytics tag or the form doesn’t properly capture analytics values
  • Names and IDs for the campaign may be different between different platforms

These issues can seem minor at first: They’re not the sorts of problems that would prevent a campaign from launching. However, they all cause complications for measuring the outcome and impact of the campaign.

The good news with this problem is that it’s solvable with some internal effort. There are several practices you can implement to ensure your marketing efforts are consistent, such as:

  • Implementing a marketing campaign QA checklist and enforcing its use
  • Providing additional training on marketing best practices for your company
  • Interviewing marketers who have had issues with consistency: Should a process be changed, simplified or removed to make things more uniform?
  • Finding ways to automate or foolproof your campaign setup and measurement

Whether you find yourself re-assessing your marketing readiness and need to get up to speed or know you’re ready to go in 2023 with marketing attribution, DemandLab and Insentric can help: Get in touch with us.