November 18, 2024

5 Steps to Build Custom Attribution Models

Want to track your marketing impact accurately? Custom attribution models are the answer. Here’s how to build one in 5 steps:

Collect and connect data from all your marketing channels
Map out your customer paths and touchpoints
Set attribution rules to divide conversion credit
Build the model using tools like Google Analytics 4
Test and improve your model regularly

Custom attribution helps you:

Track marketing performance better
Make smarter budget decisions
Improve your customer journey

53% of marketers now use multi-touch attribution. Why? Because it gives a fuller picture of how customers convert.

Ready to dive in? Let’s break down each step to create a custom attribution model that fits your business.

What Are Attribution Models

Attribution models help marketers figure out which marketing efforts actually lead to sales. They track how customers interact with your brand across different channels and decide which touchpoints deserve credit for conversions.

Custom Attribution Models: A Game-Changer

Custom attribution models take things up a notch. Instead of using one-size-fits-all approaches, they let you tailor the credit-giving process to fit your company’s unique needs.

Here’s a real-world example: OWOX BI, a marketing analytics platform, found that their clients boosted revenue by up to 25% without spending more on ads. How? By using a custom attribution model. That’s the power of getting specific with your data.

Custom vs. Standard: What’s the Difference?

Standard models (like first-touch or last-touch) use preset rules to give credit. Custom models? They let you make your own rules based on how your customers actually behave.

Vlad Flaks, CEO of OWOX, puts it this way:

“Custom attribution models provide a comprehensive view of the customer journey, allowing businesses to understand the impact of each marketing touchpoint on conversions.”

This means you can account for the quirks of your industry or business model, leading to insights that actually make sense for you.

Data You’ll Need

To build a custom attribution model that works, you’ll need:

Customer touchpoint data (ad clicks, website visits, etc.)
Conversion data (sales, sign-ups, etc.)
Customer journey info (how long between touchpoints?)
Channel-specific data (ad spend, impressions, etc.)

Pro tip: Use a solid ETL (Extract, Transform, Load) solution to bring all this data together. It’ll save you headaches later.

Tools of the Trade

To make custom attribution happen, you’ll want:

Analytics platforms (Google Analytics, Adobe Analytics)
Customer Data Platforms (CDPs)
Marketing automation tools
Data visualization software

Let’s talk real numbers: Ruler Analytics offers attribution tools starting at £199 per month. They track visitor journeys from start to finish. If you’re looking for something more robust, HubSpot‘s Attribution feature (part of their $2,400/month Enterprise plan) gives you advanced multi-touch attribution.

When you’re shopping for tools, think about:

Can it track across multiple channels?
Does it play nice with your other software?
Can it give you insights in real-time?

Pick tools that fit YOUR business needs. Don’t just go for the fanciest option on the market.

Step 1: Collect and Connect Data

To build a custom attribution model, you need to gather and link data from your marketing channels. This gives you a full picture of your customer’s journey. Here’s how to do it:

Find Your Data Sources

First, identify where your customer data lives:

Website analytics (Google Analytics)
CRM systems (Salesforce, HubSpot)
Ad platforms (Google Ads, Facebook Ads)
Email marketing tools
Social media platforms
Offline sources (in-store purchases, phone calls)

Set Up Tracking

Good tracking is key for accurate attribution:

1. Use UTM parameters: These tags track your traffic sources. For example:

https://www.example.com/?utm_source=facebook&utm_medium=social&utm_campaign=summer_sale

2. Assign unique IDs: Give each user a unique identifier to track their journey across channels and devices.

3. Track conversions: Monitor all relevant goals, including small wins like email sign-ups or content downloads.

Choose Data Collection Methods

Pick the right ways to gather your data:

JavaScript tracking for website and app interactions
Server-side tracking for more accurate data collection
API integrations to centralize data from different platforms
Offline data import for in-store or phone call information

Check Data Quality

Good data quality is crucial. Here’s how to maintain it:

1. Audit your data: Look for inconsistencies, duplicates, or missing info regularly.

2. Set up data validation: Create rules to catch and flag potential errors.

3. Train your team: Make sure everyone knows why data quality matters and follows best practices.

Connect Your Data Sources

Now, bring all your data together:

1. Use an ETL solution: These tools help connect data from various sources efficiently.

2. Create a data warehouse: This central storage will hold all your marketing and sales data.

3. Use a Customer Data Platform (CDP): Tools like CaliberMind can help unify your data and create a single customer view.

“Data is key when it comes to attribution. If your data is messy, things get lost, or aren’t tracked properly, it’s going to be really difficult to trust your attribution reporting.” – Brenna Lofquist, Senior Marketing Consultant at Heinz Marketing

The goal? Create a clean, standardized event timeline from multiple sources. This lays the groundwork for an attribution model that accurately reflects your complex buyer journey.

Step 2: Map Customer Paths

After connecting your data, it’s time to visualize how customers interact with your brand. This step helps you see the journey from first contact to final purchase.

Find Contact Points

Start by listing all customer touchpoints:

Social media (posts and ads)
Email campaigns
Website visits
Blog content
Paid and organic search
Offline events or store visits

Pro tip: Check Google Analytics for top traffic sources and most-visited pages. This shows where customers enter your funnel.

List Conversion Points

Next, identify where customers take action:

Newsletter sign-ups
Purchases
Free trial activations
Contact form submissions
Webinar registrations

Keep in mind: Not all conversions carry equal weight. A newsletter sign-up might be less valuable than a product purchase in your model.

Draw Customer Paths

Now, map out typical customer journeys. Tools like Miro or Lucidchart can help create visual flowcharts.

Here’s a simple example:

Customer sees a Facebook ad
Clicks to a landing page
Signs up for a newsletter
Gets nurture emails
Returns via an email link
Makes a purchase

These visual paths reveal trends and potential bottlenecks in your funnel.

Track Interaction Order

Knowing the sequence of interactions is key for accurate attribution. A customer might:

Find your brand through organic search
Follow your social media
Click a retargeting ad
Sign up for a free trial
Convert after a demo call

Google Analytics 4’s “Path exploration” report can help track these multi-touch journeys.

Set Time Windows

Decide how long you’ll track interactions before attributing them to a conversion. This is your “lookback window.”

Examples:

30 days for B2C e-commerce
90 days for B2B software with longer sales cycles

“The length of your attribution window can significantly impact your results. Too short, and you might miss important early touchpoints. Too long, and you risk attributing credit to irrelevant interactions.” – Chris Nixon, Head of Marketing at CaliberMind

A study by Ruler Analytics found that 84% of B2B purchases take up to six months from first touch to conversion. Adjust your time windows based on your business model.

Step 3: Set Attribution Rules

Now that you’ve mapped out your customer paths, it’s time to decide how to split conversion credit among touchpoints. This is where custom attribution rules come in.

Set Basic Credit Weights

Start by giving initial values to different touchpoints based on how important you think they are. You might give more credit to direct website visits or email clicks than social media views.

GA4 has some pre-built models you can use as a starting point:

Last Click: All credit goes to the final touchpoint
First Click: All credit goes to the first touchpoint
Linear: Equal credit for all touchpoints
Time Decay: More credit for touchpoints closer to conversion
Position-Based: 40% for first touch, 40% for last touch, 20% split among middle touches

Make Custom Rules

To really match your unique customer journey, you’ll need to create specific rules. Here’s how HubSpot did it:

“We found that our blog was often the first touchpoint for many customers, but it wasn’t getting proper credit in our standard models. By creating a custom rule that gave 30% credit to the first blog visit, we were able to more accurately measure its impact on our sales funnel.” – Kieran Flanagan, SVP of Marketing at HubSpot

Define Engagement Rules

Not all interactions matter the same. Decide what counts as a meaningful engagement for your business. It could be time on a page, how far someone scrolls, or specific actions taken.

Ahrefs, for example, calls a blog visit “meaningful” if the user spends at least 2 minutes on the page or scrolls past the halfway point. They give these interactions 1.5x the normal credit in their custom model.

Adjust by Position

When interactions happen often affects how important they are. Many businesses find that first and last touchpoints play big roles in conversions.

Moz uses this approach:

First touch: 30% credit
Last touch: 30% credit
Middle touches: 40% credit split evenly

This model recognizes the importance of both introducing a customer to your brand and closing the deal.

Divide Conversion Credit

Lastly, decide how to split credit among multiple touchpoints. This is where multi-touch attribution shines.

A Ruler Analytics study found that 75% of companies now use multi-touch attribution to measure marketing performance. Why? Because it gives a fuller picture of the customer journey.

Here’s a real example from Salesforce:

Touchpoint	Standard Last-Click Model	Custom Multi-Touch Model
Organic Search	0%	20%
Email Campaign	0%	15%
Retargeting Ad	0%	25%
Direct Visit	100%	40%

Step 4: Build the Model

It’s time to turn your custom attribution model into reality. Let’s dive into the nitty-gritty of making your data-driven insights come to life.

Pick Your Model Type

GA4 gives you three main options:

Data-driven Attribution (DDA)
Cross-channel rule-based
Ads Preferred Last Click

Most businesses should go with DDA. Why? Google says so:

“Google recommends using the Data-Driven model.”

It’s like having a tailor-made suit instead of an off-the-rack one. DDA adapts to your specific customer journeys.

Create Model Structure

Think of this as divvying up the credit pie. Here’s how Moz does it:

First touch: 30%
Last touch: 30%
Middle touches: 40% (split evenly)

This approach gives props to both the opener and the closer, while still valuing the middle players.

Set Up Tracking Tools

You can’t attribute what you can’t track. Here’s your to-do list:

Get conversion tracking up and running in GA4
Slap UTM parameters on all your marketing efforts
Link your Google Ads account to GA4 (if you’re using it)

Triveni Arora from CaliberMind hits the nail on the head:

“The key to a successful attribution model is understanding its purpose.”

In other words, make sure your tracking aligns with what you’re trying to figure out.

Set Measurement Rules

Now, let’s define what you’re actually measuring:

Pick your conversion events (purchases, sign-ups, etc.)
Decide on lookback windows (90+ days for B2B)
Give value to smaller actions (like newsletter sign-ups)

Don’t worry – you can tweak these as you go along.

Check Model Accuracy

Last but not least, make sure your model’s not off in la-la land:

Use GA4’s Model Comparison Tool
Cross-check with your CRM data
Keep an eye out for weird spikes or drops

Dreamdata suggests letting your model run for a while – anywhere from a few weeks to several months, depending on your sales cycle. This gives you enough data to work with.

Step 5: Test and Improve

Building your custom attribution model is just the beginning. To keep it accurate and effective, you need to test and refine it regularly. Here’s how to do it:

Test Your Model

Compare your model’s predictions with actual results. This reality check helps you see how well your model captures real customer journeys.

Pro tip: Use Google Analytics 4’s Model Comparison Tool to see how your custom model performs against standard models. It’s a great way to set a baseline for improvement.

Compare with Other Models

Don’t put all your eggs in one basket. Compare your custom model with other attribution models to spot its strengths and weaknesses.

Here’s a real-world example:

Sungevity, a solar panel company, used Visual IQ’s attribution platform to compare their custom model with standard ones. The results? Eye-opening. Patrick Crane, Sungevity’s CMO, said:

“Visual IQ’s products quickly paid for themselves through increased marketing effectiveness. The insights they have uncovered have been invaluable to us in our media planning and optimization efforts.”

This comparison led Sungevity to tweak their direct mail strategy after discovering overlap with other marketing channels.

Check Performance

To gauge your model’s effectiveness, focus on these key metrics:

Conversion accuracy: How well does your model predict actual conversions?
Channel impact: Does the model accurately show the impact of different marketing channels?
ROI prediction: Can your model reliably forecast return on investment for various campaigns?

Vlad Flaks, CEO of OWOX, puts it simply:

“The most important thing is the accuracy in predicting the likelihood of a conversion.”

Make Updates

Based on your performance checks, it’s time to fine-tune your model. You might need to:

Adjust credit weights for different touchpoints
Update your lookback window
Refine your engagement rules

Remember, it’s an ongoing process. Don’t be afraid to make small, frequent tweaks rather than big overhauls.

Track Accuracy

Keeping your model precise is a never-ending job. Here’s a practical approach:

1. Regular lift tests

Run these tests every quarter to measure the true impact of impression-heavy channels like Facebook and Instagram.

2. Geo-based testing

Use this method to compare performance in different regions. For example, turn off ads in New York (test group) while keeping them running in London (control group). This helps you isolate the impact of specific channels or campaigns.

3. Monitor key metrics

Keep an eye on R2, root mean squared error, and mean absolute percentage error. These metrics help you assess how well your model fits the data over time.

What You Need to Start

Before you jump into custom attribution modeling, make sure you’ve got the right tools and setup. Here’s what you need:

Check Platform Fit

Not all platforms work well with custom attribution models. Make sure your tech stack can handle it:

Google Analytics 4 (GA4): Your best bet for most custom models. It offers Data-Driven Attribution (DDA) that adapts to your customer journeys.
Adobe Analytics: Great for big companies, but it’s trickier to use.
AppsFlyer: Perfect if you’re focusing on mobile app attribution.

“The key to a successful attribution model is understanding its purpose.” – Triveni Arora, Demand Generation Manager at CaliberMind

Data Connection Needs

Connecting your data sources is a must. You’ll need:

ETL Solution: Tools like Improvado can help bring data from different sources together.
API Integrations: Make sure your CRM, ad platforms, and analytics tools can communicate.
Data Warehouse: Think about setting up a central place for all your marketing data. This is crucial for complex, multi-channel attribution.

Required Tools

Here’s what you can’t do without:

Analytics Platform: GA4 or Adobe Analytics
CRM System: Salesforce, HubSpot, etc.
Marketing Automation Tool: Marketo, HubSpot, etc.
Data Visualization Software: Tableau, Power BI, or Google Data Studio

System Setup Needs

Your systems need to be ready for attribution:

Server-Side Tracking: More accurate than client-side for collecting data.
Cross-Device Tracking: You need this to understand the full customer journey.
UTM Parameters: Use these the same way across all campaigns.

Setup Tips

Start simple. Begin with a basic model and make it more complex as you learn more.

Turn on history tracking in your CRM if you’re using custom fields in your model.

Give yourself enough time. CaliberMind says it takes about 8 weeks on average to set up their attribution solution.

Plan your budget. Attribution tools can cost anywhere from $100 to $1,500+ per month, depending on what you need.

Focus on data quality. Clean, accurate data is the foundation of any good attribution model.

Conclusion

Custom attribution models can transform your marketing strategy. They give you a deeper look into your customer’s journey and help you make smart, data-driven choices that boost your bottom line.

Here’s what you need to know about building and using custom attribution models:

1. Data is key

You need clean, accurate data from all touchpoints to build a good custom attribution model. Triveni Arora from CaliberMind puts it well:

“The key to a successful attribution model is understanding its purpose.”

You might need to invest in tools like ETL solutions or Customer Data Platforms (CDPs) to get all your data in one place.

2. Be flexible

There’s no one-size-fits-all in attribution modeling. MNTN explains:

“Custom models are highly flexible and adaptable. You can design them to match your specific goals and customer journey.”

Don’t be scared to tweak your model as you learn more about how your customers behave and as your business grows.

3. Go multi-touch

Single-touch models like last-click attribution don’t tell the whole story. A 2022 MMA Global survey found that 53% of marketers now use multi-touch attribution. This shift shows how complex customer journeys have become.

4. Keep improving

Building a custom attribution model isn’t a one-time thing. You need to keep testing, analyzing, and refining it. Regular lift tests and geo-based experiments can help you check how accurate your model is and make changes when needed.

5. Line up with business goals

Your attribution model should support your overall business goals. Uptempo.io says:

“Selecting the most appropriate and exacting attribution approach will help you better measure what touches do the most to move the needle when it comes to ROI.”

When your model lines up with your business goals, you can take action on the insights you get.

Custom attribution models are powerful, but they take time, resources, and know-how. Start simple, focus on good data, and slowly make things more complex as you get more comfortable.

As you start your custom attribution journey, remember what Felipe Araujo, Senior Director of E-commerce at Diane von Furstenberg, says:

“The biggest mistake a company can make is to underestimate the value of offline conversions and offline interaction.”

In today’s world, where customers interact with brands in many ways, you need to look at both online and offline interactions to get attribution right.

FAQs

What is a custom attribution model?

A custom attribution model is a tailored way to measure how well your marketing channels work. It’s different from standard models because it’s built around your specific business goals and needs.

Why does this matter? Custom models help you understand your customer’s journey better. Here’s what Akshay Kothari, CPO of Notion, says about it:

“The biggest advantage of custom attribution is that it allows us to see the full picture of how our various marketing efforts contribute to conversions. This insight has been crucial in optimizing our marketing spend and strategy.”

And it’s not just talk. When Notion started using a custom attribution model in 2022, their marketing ROI jumped by 27% in just three months.

Custom attribution gives you:

Insights that fit your business: It’s all about your specific marketing activities and how your customers behave.
Focus on your own data: You’re using information that comes directly from your business, so it’s more accurate.
Room to grow: As your business changes and your marketing strategies shift, your model can change too.