Marketing attribution models will lie to you. A lot.
Sometimes the lies are flattering. Who doesn’t want to hear that their brilliant email campaign is responsible for millions of dollars of new pipeline?
But the lies always hurt the business. Companies that don’t understand what makes customers buy are doomed to spend money on ineffective tactics. That burns cash and hurts growth.
We’ll take a look at the 3 biggest lies in marketing attribution, and share the best way to determine the truth in what marketing and sales tactics matter. And I’ll be focusing on opportunity attribution, not lead creation.
Let’s jump right in.
First Touch Marketing Attribution
First touch is one the simplest attribution models out there. It gives all the credit of a lead or opportunity to the first tracked engagement.
This approach gives too much credit to channels that drive a lot of traffic. Tactics like social media, paid search, and retargeting ads do well by this measure. But if you’re focused on generating revenue, there’s a big elephant in the room: “What about the rest of the customer journey?”
First touch attribution is super easy to implement. Many website tracking tools like Google Analytics provide this right out of the box.
First touch models ignore all the other stuff that happens AFTER someone has hit your website. Companies with long sales cycles have a lot of other stuff to consider. Think about lead nurturing emails, webinars, and tradeshow attendance.
It’s helpful to put yourself in the customer’s shoes here. If you are buying a $100,000 per year software tool, would a banner advertisement be the most important thing to your purchase decision?
There are so many more meaningful engagements necessary to convince a person that your product is the right fit for their needs.
There are so many issues with first touch attribution – why use it at all?
If you are measuring brand campaign success, first touch can help quantify the impact. A spike in direct traffic (someone typing your website URL directly into browser) and organic search (someone searching for you company via search engine) after putting up billboards is really valuable information. It shows that you are doing a good job of getting the word out about your company.
But if you’re in the business of tracking what marketing matters for revenue, first touch is the wrong way to go.
Last Touch Marketing Attribution
Last touch attribution is another simple model marketers commonly use.
To be fair, it has a few good traits.
It makes sense that the final marketing activity a prospect has before deciding to buy is much more likely to be the cause of the decision. So it seems like a better choice than a first touch attribution model.
Better yet, last touch attribution is another out-of-the-box option provided by many marketing and advertising platforms.
Clear winner, right? Wrong.
Last touch suffers from the same problem as first touch. It does not consider the entire customer journey. And by failing to do so, it gives too much credit to a single set of tactics. Neil Patel explains this beautifully in this post.
Imagine what would happen if you designed your marketing strategy around last touch attribution. Bottom of the funnel content like case studies and product comparisons would get a lot of credit. But if spent all of your time talking about why people should buy your stuff, you will neglect other important steps. Creating brand awareness and helping them define their problems are vital parts of marketing as well.
When should you use a last touch model?
Last touch attribution can provide powerful insights about sales enablement content. Which whitepapers or case studies are most effective in making prospects decide they want to buy? Very important questions to answer.
Weighted Marketing Attribution Model
Weighted models are a significant improvement in marketing attribution approaches. Weighted models consider the entire customer journey, and assign credit to marketing touchpoints all along the way.
There are many weighting schemes out there. Here’s a few popular ones:
1. Linear attribution: This is the simplest approach to weighting. It gives each marketing activity along the customer journey equal credit.
2. Time decay attribution: This adds a wrinkle to the linear attribution model. It assumes that activities which occurred closer to the opportunity conversion date have more value.
3. W-shaped attribution: This approach puts 30% of weight on the three big conversion events: first touch, last touch, and lead conversion. Then it spreads the remaining 10% of credit across the remaining marketing activities.
Each of these approaches have their strengths and weaknesses. Linear attribution is easy to implement, but doesn’t try to understand the relative importance of different marketing activities. Time decay attribution is more nuanced, but it underweights marketing activities that occur early in the buying process.
Of the three options, W-shaped attribution offers the best mix of identifying critical events in the journey while also considering all the other activities that have occurred.
But even the W-shaped attribution model has a flaw. A flaw shared by the other weighted models, first touch, and last touch attribution schemes.
It assumes you know the right weights.
At the end of the day, all of the attribution approaches shared make big assumptions about what is important and what is not.
For some, the logic is much better than for others. But the reality is there is no one right answer. Every business is different, and each customer’s response to marketing will differ too. Choosing a scheme that pretends to know how everyone behaves stops you from really investigating and understanding the truth.
Enter Predictive Marketing Attribution
Predictive attribution flips the problem on its head. Instead of making an assumption about which marketing activities are most important, it allows the data to decide. Machine learning models can be used to identify and weight the most impactful marketing activities using your own customer data.
In a world of off-the-rack marketing attribution, machine learning gives you a custom-tailored model.
And while the math behind machine learning models is quite complicated, the underlying logic is simple.
Looking across your lead data, a predictive model looks for marketing activities that are present in customer journeys that result in opportunity creation and NOT present among leads that fail to convert.
Activities that are never associated with opportunity conversions are ignored. Activities that are only associated with successful conversions get a lot of weight. Most activities will fall somewhere in the middle.
Take a look at the graphic below to see the logic model:
The answer provided by predictive attribution is specific to your individual business. You don’t need to rely on “best practice” or industry benchmarks – you are finding the right answer for your own company.
While the predictive approach is the best one, it isn’t perfect. It requires having plenty of data available to generate the models. And as you get new customer data, the model needs to be refreshed to reflect the changes.
The biggest challenge for predictive attribution isn’t a technical issue. It’s a people problem. Making it understandable for people without a degree in statistics can seem hard.
The other models described are easy to explain and implement. Predictive modeling may feel like black box magic if you try to explain the math behind it. No one wants to look dumb in front of their boss, trying to explain how machine learning works (or get fired if it doesn’t work). That makes it “feel” safer to choose a simple model that you can describe easily. Even if the simple model is very, very wrong.
How I Learned To Stop Worrying and Love Predictive Attribution Models
There is no better way to identify the most impactful marketing activities than using a predictive attribution model.
Single touch and simple weighted models make incorrect assumptions about your prospects. This means you’ll spend more money on tactics that are ineffective and miss out on growth opportunities staring you in the face.
Want to make the right conclusions from your marketing data? Predictive attribution holds the keys to sustained growth.