Marketing attribution: How to build an attribution model for your business

Far too many marketing teams are still seen internally as the arts and crafts department. 

They’re considered a support function at best and a cost center at worst.

And that’s frustrating considering all the work you’re putting into launching new products, producing awesome content, building campaigns, and well… ultimately driving revenue.

The good news is that there’s a way out. And that way is called marketing attribution.

By beginning to assign revenue to individual touchpoints like social media posts, ads, blog posts, website pages, and emails, your marketing team can escape the arts and crafts stigma for good.

And who knows, maybe it’ll even get you an invitation to the big kids’ table (aka the management team or eventually even the board of directors). ??‍♀️ 

But before we get ahead of ourselves, what exactly is an attribution model? How do you choose the right model for your business? And how can you get started with attribution modeling in practice?

You might want to fasten your seatbelt because you’re about to find out.

1. What is an attribution model?

An attribution model is a set of rules that determines how much credit of a sale is assigned to each touchpoint along the customer journey.

A well-designed attribution model will help you understand the revenue contribution of each individual marketing campaign, message, and tactic. 

Ultimately, you can use this data to optimize your marketing spend and direct more investments into the messaging, channels, and tactics that are working well. On the other hand, marketing attribution will also help you stop wasting money on tactics that are not worth the time and money you’re putting in.

But before we run through the different types of attribution models, let’s quickly discuss the limitations of marketing attribution.

The limitations of marketing attribution

Attribution modeling is cool and everything but is it perfect? Nope.

Will it help you attribute revenue to your brand marketing activities? Probably not. Will it allow you to attribute revenue to word of mouth? Nope. Or will it take into account what happens offline? Usually not.

The simple rule of thumb is that if most of the touchpoints your customers encounter before they buy from you happen offline, you probably shouldn’t spend a lot of time on attribution modeling.

On the other hand, if you’re in self-serve SaaS, ecommerce, DTC, or any other online-heavy business, it definitely makes sense to invest in figuring out your marketing attribution.

However, it’s good to remember that attribution models are simplified illustrations of complex customer journeys at best. That’s why instead of trying to build a perfect attribution model, your goal should be to build a useful attribution model. 

As with anything in life, though, done is better than perfect. Doing something always beats doing nothing.

And with that said, let’s take a quick look at five of the most typical attribution models.

2. Types of attribution models

The first thing you need to know about attribution models is that there are two main categories: single-touch attribution models and multi-touch attribution models.

In the next section, we’ll define each and break them up further to understand five of the most common models.

A) Single-touch attribution

Single-touch attribution models assign 100% of a sale to a single touchpoint. 

Even if a customer has visited your website 17 times, seen your ads on Facebook, Instagram, and Google, subscribed to your newsletter, and attended a webinar, all the dollars they’re bringing in will be attributed to only one of those touchpoints: typically the first one or the last one.

First touch attribution

In first touch attribution, 100% of the sale is attributed to the first (measurable) touchpoint on the customer journey.

Last touch attribution

In last touch attribution, 100% of the sale is attributed to the last (measurable) touchpoint on the customer journey.

B) Multi-touch attribution

As you’ve probably guessed by now, multi-touch attribution models assign revenue between multiple touchpoints. But since evenly distributing revenue isn’t the only option, let’s look at the most typical multi-touch attribution models.

Linear attribution

In linear attribution, total revenue is split evenly between all the measurable touchpoints a customer has engaged with before making a purchase.

Time decay attribution

In time decay attribution, total revenue is split so that the more recently the customer has engaged with a touchpoint, the more revenue is attributed to that touchpoint.

Position-based attribution

In position-based attribution, most of the credit is split between the first and last touchpoint, while the remaining credit is distributed evenly between the touchpoints in the middle.

3. How to choose an attribution model

Now that you know what kinds of attribution models are out there, it’s time to choose the best one for your business.

This is typically a simple, two-part process that consists of:

• Making some generalizations about typical customer journeys
• Testing a few different attribution models to see which one of them gives you the most useful data

Step 1: Make some generalization about your typical customer journeys

Since your ideal attribution model depends on your industry, audience, business model, average deal size etc. it’s a good idea to start by answering a couple of questions:

1. How many measurable touchpoints does an average customer encounter before they convert?
   • If the answer is 1, you’re fine to go with either of single-touch attribution models because you’ll get the exact same results from either
   • If the answer is 2, you might want to go with linear attribution.
   • If, on the other hand the answer is >2, you may want to experiment with different multi-touch attribution models.
2. What’s the buying process like in general?
   • If you’re working in ecommerce and your average order value is fairly low, your customers are typically able to make purchase decisions pretty quickly. However, this doesn’t mean that most people would go directly from the first ad to purchase, which may suggest that you should experiment with different multi-touch attribution models.
   • If you’re in B2B and have long sales cycles and multiple people involved in the buying process, attribution modeling gets even more complex. But especially here, multi-touch attribution on the account level might be your only option if you want to get valuable data.
3. Are all touchpoints equal in value?
   • If you have a new brand or otherwise low brand awareness in the target market, the first touchpoint is actually pretty valuable. After all, without it, the sale wouldn’t be possible. If this is the case, you might want to consider position-based attribution.
   • Sometimes staying top of mind to customers during longer sales cycles is just as valuable as making them aware of your solution in the first place. If this is the case, a linear attribution model may be your best bet.
   • If your brand is well-known, you may well want to attribute more revenue to the touchpoints that are closer to the sale. In this case, a time decay model might work for you.

Now that you have some kind of an idea which models to try, it’s time to move on to the next step: experimentation.

Step 2: Experiment with a few different attribution models

Remember what I said about doing something being better than doing nothing? Well, now is your chance. 

Pick two or three attribution models from the list and start collecting data. 

Depending on your sales volume and the typical length of your sales cycle, you may need to run the experiment anywhere from a few weeks to several months.

But the good news is that once you’ve set up tracking and committed to doing the work, experimenting with a few different attribution models is no problem.

Speaking of setting up your attribution model(s)…

4. How to set up your attribution model in practice

When it comes to setting up your attribution model, you basically have three options:

• Good old Google Analytics
• A cloud-based data warehouse like Google BigQuery or Snowflake
• A dedicated marketing attribution tool like Dreamdata.io or LeadsRx

Attribution modeling with Google Analytics

If you want to quickly test out a few different attribution models or you don’t have a lot of budget flying around, Google Analytics is by far the fastest and cheapest way to get started.

The Model Comparison Tool in Google Analytics’ MCF (Multi-Channel Funnels) is perfect for comparing up to three attribution models at once.

To learn how to do this in practice, read Google’s support materials on attribution modeling.

Attribution modeling with Google BigQuery

If you’re serious about attribution modeling and you have a ton of data across your CRM, customer data platform, social channels, ad platforms, and your website analytics tool, Google BigQuery (or a different cloud-based data warehouse like Snowflake or Redshift) might be your best bet.

This is where Supermetrics comes in handy. We’ll help move all your marketing data from siloed platforms (like your CRM, marketing automation platform, paid channels, social channels, and web analytics) to your data warehouse with a few clicks. 

Instead of wasting months of your engineers’ valuable time in building custom integrations, you can literally set up automated data flows in minutes with Supermetrics.

With all your marketing and sales data in BigQuery, you’ll be able to stitch all the touchpoints together by session and create a custom attribution model that makes sense in the context of your business.

To learn more about getting started with attribution modeling with BigQuery, check out this post by our friends at CXL.

For inspiration, read how and why we’re using Supermetrics for BigQuery at Supermetrics.

Attribution modeling with a custom attribution tool

If you don’t have a lot of technical knowledge or analytics resources, your third option is to go with a built-for-marketers attribution tool like Dreamdata.io, Attribution (clever name, eh?), or LeadsRx.

For more information on how to choose the best attribution tool for your business, check out this awesome article by Segment.

No more arts and crafts ??‍♀️

As you probably noticed, marketing attribution can be a hairy beast. 

But the cool thing is that your first attribution model doesn’t have to be your last. 

By constantly testing and experimenting with different models and relative weights, you’ll start gaining the important information you need to drive more revenue with marketing.

And with that data at your disposal and those impressive results in your back pocket, you can finally say buh-bye to the sticky stereotype of the marketing team as the arts and crafts department. ?

Happy attribution modeling! 

Psst! If you’re not quite ready to start attribution modeling but would like to make better use of your marketing and sales data anyway, you can now start a free 14-day trial of any Supermetrics product. No credit card needed. ?