Attribution data: What it is and how to use it

What is attribution data?

There’s no single type of attribution data. It comes in many forms, from many sources. Cookies are one example, although third-party cookies are fading fast; the tracking pixel added to web pages for analytics is another. Tagged URLs, with the long series of extra characters you see after clicking a link, also count, as do email opens, social media clicks, and other interactions recorded in your CRM. 

But there are commonalities. With attribution data, what matters most is that it’s first-party information: the result of an individual action, created by an identifiable person. Whether that’s a prospect just entering the sales funnel, or a customer with a years-long relationship. 

And when those different events are seen together in context, as a customer journey mapping out that person’s behavior on their way towards the sale, it’s possible to understand what drove them to act at each stage, what worked and what didn’t. All of which is marketing GOLD.

Why is first-party data so important?

There are important differences between first and third-party data. Where first-party data points are created as the direct result of a relationship between you and your customer, that visit to your website or purchase from your store, third-party data is more distant, often shared from marketing partners or resold by brokers. (That’s why you have to click through  all those “Your data may be shared” notices whenever you visit a new website!) 

So privacy legislation and corporate policies are moving away from third-party data, with Big Tech mostly agreeing it’s fading into irrelevance. But there’s a more concrete reason to go first-party …

… in the EU, third-party data like tracking cookies must be deleted after 7 days. So if the initial touchpoint happens at the start of the month and the next happens over a week later, that first touchpoint isn’t visible in the customer journey and won’t be part of your analytics. Which creates errors and biases that roll forward into your business planning.

That’s why we’re all-in on first-party data at Billy Grace. It’s compliant with regulations everywhere, it’s more accurate in your analytics, and it lets you build a stronger relationship with your customer. 

Types of attribution data

Let’s list some more types of attribution data. Note: not every method uses all types of data, or even many of them. The trick’s in deciding what is most useful to you.

• Most analytics apps use clickstream data: it’s the sequence of pages and links a visitor experiences as they navigate your site. When Seen as a connected customer journey rather than a set of clicks, it helps marketers understand what they were looking for – and whether their needs were answered. 
• Referral data adds to the customer journey, telling you not just where customers are *going*, but where they’ve *come from*. It’s all the sources that sent them to you, from links in emails to search results. 
• Conversion data captures the results of actions. If a user completes a signup form or hits the Buy button, that’s a major event, but never forget the actions that led them there are just as important.
• Engagement data is created when people interact with content: watching videos, reading emails, dwelling on a web page for a few minutes. Linked to this is social data: all those Likes and Shares that demonstrate a deeper interest. 
• Platform / device / connection data. People these days carry several devices on their person, phone, pad, laptop, with several more at home and work. To understand your customer as an individual it’s necessary to know who they are across them all.
• Geographic data. Knowing where a user is can provide useful context. If someone’s downtown in daytime but the suburbs after six, it’s likely the data is denoting work and home.
• Demographic data. It’s old-school marketing, but data about age group, gender, and social class still matters. Of course, your goal is to know each customer as an individual, but demographic data helps group them into audiences and segments, so you can focus your marketing activities.
• Campaign and channel data. It’s often an email campaign or PPC ad that gives someone the final nudge they need to make a purchase; knowing the campaign ID and ad variation that delivered each clickthrough gives insight into what worked.

Of course, data isn’t much use with doing something with it, and that’s where an attribution model comes in.

Types of attribution models

An attribution model is how you critically assess attribution data and decide how much “weight” to give each interaction’s contribution to the sales process. (We have a whole article on this if you’d like to learn about attribution models.) Here’s a summary.

First touch and last touch attribution models (FTAs and LTAs) “weight” responsibility for the sale towards the beginning or end of the customer journey. A large brand campaign might influence buyers heavily at top-of-funnel; it’s a reason to go first-touch. But an action-oriented PPC special offer might persuade more buyers to click now; you might use last-touch here. It’s all about the right model, for the right situation. And both have the bonus of being simple. 

Linear and multi-touch models take it further, splitting the contribution to a sale between all touchpoints on the customer journey, equally with linear, and customized with multi-touch. They can be very accurate as long as you set them up to model reality precisely.
 

Another step up is Marketing Mix Modeling, where the customer journey surfaces from correlations between events; it’s possible to see the sequence even without directly connected data points. Even better is Unified Marketing Measurement (UMM): it takes all channels into account, including online & offline activities like drive-to-store, using unique identifiers like offer codes to stitch the journey together. (No prizes for guessing which model Billy Grace is passionate about.)

But there are other factors in using attribution data effectively, windows and modes. So let’s look at both.

Types of attribution window

Attribution data tells you the “where” and “what” of an interaction; an attribution window is all about the “when”. A click on content a year ago doesn’t have much connection to a sale today. But a click in the morning and a sale that afternoon? There’s a far stronger (and more likely) relationship between them. So an attribution window represents the timeframe in which a user’s action (such as a purchase, signup, or download) has reasonably been influenced by a specific marketing effort. 

For greater flexibility it’s easy to change the attribution window based on what matters most to you. A single day may be correct for a tactical sales offer. Other approaches may work better with windows of 7 or 30 days, or even one without limit if your business has very long sales cycles. 

So how do you decide what your window should be? There are several methods.

• You can decide based on the type of interaction. If most of your sales come from clicks on PPC offers, the window will be short. (Known as a clickthrough window.) But if a lot of people buy after browsing your content (not always clicking through) the window may be longer: a view-through attribution window.
• Or you can decide based only on duration. You may want to limit attribution to conversions made within 24 hours of an interaction. Or 7 days. Or up to a month for longer sales cycles. You get the idea.
• Or on attribution model. A First-touch attribution window will usually need to be longer than a last-touch one, to give time for the whole customer journey to happen … but not always. As always, the uniqueness of your business matters.
• Even on device or platform. If sales are strong on the mobile channel only, or one social media site delivers more than others, you may want to set the attribution window based on data from these modalities. 

Good marketing management software lets you change the attribution window, so it’s right for your business. (Needless to say, that’s a feature of Billy Grace.) Next up: modes.

Types of attribution modes

It should be clear by now that attribution isn’t a singular “look” at the data. Attribution data can be viewed in different ways, through different models and windows depending on what makes sense for you. Here’s another way to “cut” the data: attribution modes. 

There are various types of attribution mode out there, but let’s look at the duo Billy Grace customers find most useful:  session-based and event-based. 

Session-date attribution modes

Taking the session-date-based choice means you’re allocating credit for a sales event along the whole customer journey (more precisely, the length of your attribution window). 

If the brand experience happened on Tuesday, the Facebook interaction on Thursday, and the PPC-driven sale on Saturday, and you decide each touchpoint contributed a third of the sale, the total contribution will be spread across three dates. This makes session-based mode great for seeing a “snapshot” of your return on marketing spend at any given time – each useful interaction is credited for the day it occurred, even if the sale wasn’t in the same period.

The session-based mode is like a Work-in-Progress report, showing you the results of your activities even for customer journeys that aren’t complete yet.  

As with any way of looking at data, there are drawbacks: that’s why we offer so many different ones! And one negative point of going session-date-based is that with longer attribution windows, especially the “unlimited” one, it’s less clear when all those early interactions will deliver a concrete sale. That’s why some customers use event-date-based modes as another option.

Event-based attribution modes

In an event-date attribution mode, all those touches over time, the customer journey, are credited to the day the sale occurred, without including interactions that haven’t yet led to a sale. This carries plenty of advantages. An event that pushed the customer deeper into the sales funnel, like that Meta visit several weeks ago, is recorded on the date of the sale itself, with the various touchpoints given their credit weightings on the same day. 

This makes it clear which interactions delivered the sale, even if they stretched back into the past. Remember your attribution window can stretch back to the dawn of civilisation! While ensuring each touchpoint gets the right credit, letting you see which channels are performing well for you “at the point of sale”. 

The downside of the event-date mode? Since there’s a difference between the actual touchpoint date and the date it was given that credit, it’s harder to relate those touchpoints to your daily spending, since not all events happened on the same day. 

So, in brief: use session-date-based attribution mode when you’re reviewing ROAS at a specific point in time (like your daily sales figures) and event-date-based when you’re looking at longer timescales like quarterly campaign performance. Whatever your priority, there’s a mode for it.

There are many more attribution modes, but most fall into one of these categories. And once you’ve chosen the one that works, it’s crunch time: how to use it.

How to interpret the data

Once you’ve set your attribution models, windows, and modes, and data is being gathered from your sources and applications with integrations and connections, it’s time to do the critical thinking of interpreting it. Short version: Billy Grace can give a lot of insights, if you know how to use the data. Here are some general interpret-tips:

Begin with your business goals. (Always a good place to start.) Think of what outcome you’re looking for or what you want to discover: length of customer journey? Contribution of different touchpoints? Total sales on any given day?, and choose your “cut” on that data to suit. Examples: 

• If you’re focused on BOFU (which we wouldn’t advice, but nevertheless) for example, for a short-term tactical campaign running at high intensity for a single week, consider a last-touch attribution model with a short attribution window and event-date mode. It’ll give you a good picture of what happens each day in a given period.
• If you’re interested in the effects on sales of an early interaction, like a big branding effort for the January sales, try a TOFU approach, which will show you the sales hike that correlated with your big campaign.
• If you want the long-range view of how earlier touchpoints affect later outcomes, say, for complex, high-value purchases with longer sales cycles, switch to a multi-touch attribution model with a longer attribution window. You’ll see all the contributing factors spread out over time, and a different perspective on the data. 
• If you want to bring quantitative and qualitative data together for a Big Picture of your overall branding and marketing spend, use UMM and MMM to see how other factors like offline branding efforts influence customer behavior. 

IN SUMMARY: the right view on the right data

1. With easy-to-understand charts and graphs providing a single view of the truth to as many people as you want, the raw data of clicks, views, and conversion events comes to life. By seeing each customer journey as a journey, and not a bagful of separate interactions, you’ll notice some complete faster and more efficiently than others. Which means you can focus your marketing activities on them. 

2. Second, by seeing which individual interactions contributed most to each result, you can boost your ROI even further. It’s not always obvious that a brand ad seen early in the sales funnel had a huge effect on sales a month later, but seeing all the data in context makes it much clearer. There may be a difference between what *seemed* responsible for an event and what was *really* responsible … and the right approach to attribution data can find it. 

3. Third, automated optimization of your ad campaigns can take away a great deal of drudgework each month, as you adjust budgets and bids in the hope of sales uplift. By basing such decisions on insights from actual data and letting software perform adjustments to campaign resources within limits you set, your people can be freed for more creative work. Like writing the next quarter-busting ad campaign.

CONCLUSION: Attribution data is about more than data

The world is divided into those who have data and those who know how to use it. At Billy Grace, we work with people who want to be in the latter camp, and we’re helping over 200 of you grow your sales and profits every day.