Log In


DIY Attribution: 6 Data Hygiene Basics

Posted June 7, 2023

Welcome to another step forward in our series meant to help you design a DIY attribution model! One of the most critical steps toward getting attribution “right” is putting rigor in place around data collection. As the old analyst saying goes, “Garbage in, garbage out!” Of course, we know it’s not as simple as changing how we structure our data going forward. A lot of the work goes into fixing what was done in the past.

Andrew Sawusch, Head of Customer Operations & Onboarding, at CaliberMind, shares his insights on what needs to be done prior to implementing attribution, the concept of the virtual campaign, and some best practices when it comes to UTM parameters.

Tip 1: Decide Which Data Is Critical to Include

When we are implementing attribution, one of our first focus points is the systems we have to collect data from. For people who are thinking about building attribution themselves, the first major step is deciding which systems need to be included and how to prioritize data cleansing.

“First and most important system is where you have all your data for your accounts, your contacts, your leads, and your opportunities. When we’re onboarding customers, that is the first thing we look at. The hygiene of these records is vital – and so is the organization because everything should eventually roll up to an account. We recommend beginning with the CRM platform that you use, whether it’s Salesforce Dynamics, HubSpot, or Zoho.

“We then start uncovering where other critical information lives. Typically, there’s a marketing automation platform. If you use HubSpot or some of the combined CRM and marketing automation systems, you’ve already mapped out your marketing data because it lives in the same place as your sales data. It’s a one-stop shop. However, whether it is Pardot, Eloqua, or Marketo, we’ve seen a lot of different flavors and varieties of marketing systems that customers try to pull information from. 

“And then we have ad platforms. You go where some other niche platforms might be. There are so many different platforms out there. I think there are 11,000 different Martech products in the market right now. It’s unreal how many different platforms there are. And trying to connect and organize these activities so that you can make sense of them is really the most difficult part. How do all these events join and get organized along a timeline associated with a person and an account?”

Tip 2: Use Your CRM Campaign Object

As consultants, we’ve seen marketing teams live and breathe in their marketing automation platform and don’t really think about pushing campaign data to their CRM. Andrew provided some insights on what is different about the campaign structure between marketing automation and Salesforce, and why it is important to organize campaign data in the CRM.

“A platform like HubSpot documents send lists for emails, form fills, and other key digital interactions. Offline events like tradeshows, direct mail, or other activities are where platforms like Salesforce shine. Salesforce has been integrated with so many systems, including these marketing automation platforms. 

“So say you had a send list in HubSpot and have a connector that goes into Salesforce and all these individuals are added to a campaign as campaign members. You’re keeping a running tally of how those people interact with your brand. Form fills are also added to Salesforce. But within Salesforce, you also have the ability to upload offline data. Think of in-person events. You might have individuals who came to your booth, who signed up for this webinar, and registered through these different portals. All this information can be collected into the source of truth, which is the CRM.”

Tip 3: Decide Who (Contacts or Leads or Both?) to Include in Account Aggregation

Another thing we see customers struggle to overcome on their own is a lack of the concept of an account or company object in some major marketing automation platforms. Even within some CRMs, people are split across lead and contact objects, which also makes aggregating information at the account level difficult.

“When we are talking about the general individuals who show up on your sites, they’ll identify themselves and give you a little bit of information, but oftentimes, you don’t know their company. You don’t know where they are from. In most circumstances, you just have a list of individuals and you have their email addresses and their contact information. It’s important to organize the data in your data warehouse or a tool like CaliberMind so you can identify when these individuals are part of the same buying group.”

Tip 4: What are “Important” Interactions?

When it comes to attribution, by its very nature, it’s a combination of opportunity data and brand interactions. When marketers want to look at this data, they’re probably thinking about demonstrating campaign effectiveness. Andrew elaborated on the common missteps people should correct prior to implementing attribution.

“I’ve seen some of the dirtiest and most ill-set-up campaigns. I’ve seen some of the best setups, too. It’s vital to have a consistent campaign response definition. These should be your hand raises, or the key engagements that could signal a readiness to engage with sales. Typically, within a Salesforce or HubSpot marketing automation platform, you’ll also track member statuses that wouldn’t be considered a hand raise. For example, email marketing campaigns where you’re tracking clicks and opens. Maybe you want to track whether they clicked on a Google ad or LinkedIn ad prior to filling out the form. These are all things we have to consider.

“We also have customers who want to include sales interactions in their attribution data. They want to see meetings and responses to sales outbound activities. 

“Businesses have to decide whether they just want to optimize their campaigns or if they want to represent department efforts through attribution. 

“We also need to think about how we want to think about campaign activity. Do we want to compare individual campaigns of the same type? Or do we want to group them by an initiative? Or do we want to be able to split out campaigns by business unit and the product we’re focusing on?

“That’s when we start seeing people get creative with campaign hierarchies in Salesforce. Unfortunately, layering campaigns through the parent object makes reporting extremely difficult. We’re trying to navigate layers and organize things using a single field (the parent campaign). Instead, companies should really think about leveraging custom fields. They can be used to classify campaigns in a way that is easy to report on.

“Too often we see companies limit themselves to out-of-the-box campaign features: type, cost, and name. If you want to organize by business unit, product, region, or some other factor, create fields.

It really behooves people to keep the campaign tables as flat as possible. This could include adding UTM parameter fields on the campaign member record (more on this lower in the article). This allows companies to track the call-to-action or campaign type (like demo request, webinar registration, or form fill) and see what paid medium the campaign member interacted with (LinkedIn paid ad, organic social, or Google Display, for example).

Tip 5: Don’t Limit Yourself to Campaigns

Historically, attribution was thought of and tooled to be dependent on campaign members. Some tools recommended workarounds, like creating campaigns for sales to manually indicate the deal was sales or channel sourced, and some creative ops people duplicated key sales interaction as a campaign member. Instead of duplicating activity or requiring manual data entry, consider a “virtual campaign” or database/data lake attribution activity pool.

“Typically, a virtual campaign represents an activity that wouldn’t traditionally warrant a campaign member and campaign. A virtual campaign allows you to take a group of similar activities and aggregate them. For example, you could decide to track page visits on your website. Pushing that data into a campaign could be cost-prohibitive in your CRM, but storing that information in a data warehouse is typically more affordable. 

Other examples of virtual campaigns could include sales meetings, successful phone calls, and partner referrals. It could even include product or in-app signals. 

“When we talk about marketing campaigns, we’re talking about a very small subset of the interactions necessary to successfully close an opportunity. Using virtual campaigns allows us to expand what we include in attribution. You can also get more granular about all those various marketing activities (like website visits) that might not necessarily have been thought about as a traditional campaign.”

Tip 6: UTMs & Campaign Structure

Traditionally, we’ve seen organizations create a parent campaign for the brand interaction (like form fills, webinar attendees, etc.) and then create several child campaigns to separate UTM channels. We’ve also heard debates about whether campaign type should really tie to channel or UTM  parameters or whether campaign type should represent the CTA. To date, there isn’t a “right” answer and it depends on what people are trying to report on. Andrew gave his perspective on what he’s seen work for B2B organizations.

“It’s helpful to have UTM information down to the ad or even keyword level. You can create custom fields to capture granular UTM data at the campaign member level in Salesforce. You can stamp the UTM source, UTM medium, and UTM campaign; and then you can create reports in Salesforce. You can use custom fields to dissect the data however you want. Doing it this way works much better than stamping the UTM data on a lead or contact.

“If you stamp the UTM data on the lead or contact, you’re either snapshotting old information or constantly overwriting what the person clicked on with the next interaction. You could create a first UTM interaction field to stamp how they first came in, but if you’re going to continuously stamp the same field over and over, it really doesn’t help you. Yes, you can track field changes and have historical records. But reporting on the history object is far from simple.

“Pushing UTM data down to the campaign member allows you to see both the campaign data or brand interaction and then how they arrived at that CTA through UTM data. You also have a running history of what people are interacting with over time. Where have they been coming from? Has it been consistently Google ads? Are they known customers? A pattern often emerges.

“There’s a standard practice of what UTM parameters you should have. UTM source, UTM medium, and UTM campaign are usually a given. Some companies include keywords. The problem is that companies often reuse campaign names. You could have five derivatives of an ad that are named the same thing. Because of this, we recommend adding the campaign ID.

“On some platforms, they’ve got these dynamic track variables where anytime an ad is served up, they click on this campaign, and it automatically populates the creative ID, the keyword, the source, the medium – you name it. 

“The best practice is to provide as much information for yourself in the future as you can today. And once you land on what all you would want to report on, apply the same standard to all of your digital campaigns.”

For more great tips on what to watch for, backfill, and put in place for better future data hygiene, watch or listen to the full interview above.

View Our Other Thought Leadership