The Ultimate Attribution Playbook in 2023

Unlock Revenue Growth With Data

Knowing where to invest marketing budget to increase contribution margin and overall revenue growth is the #1 pressing challenge for any marketing or growth leader.

As multichannel complexity and media budgets grow, attribution becomes one of those topics we really can’t ignore.

To truly understand the most valuable customer journey design, relying on default attribution reporting within ad platforms or Google Analytics just doesn’t cut it. In fact, it can even do more harm than good due to misattribution and double attribution - a big problem with these uncensored (and self-serving) tools. 

The trouble with free on-platform attribution reporting (like Facebook or Google Analytics) is that they are siloed walled gardens that work in isolation with their own limited data sets. Your most powerful and valuable attribution analysis needs to cover everything, directly tied back to revenue results. 

Without a proactive attribution strategy that connects all your customer journey and conversion data, optimized customer journey design will remain an elusive mystery. Highly influential channels like dark social or offline interactions are often underestimated or completely missing, while non-profitable campaigns are over-indexed. 

The difference in business results can easily stack up to millions of dollars in wasted budget and lost opportunities - especially where larger paid media budgets are involved. 

 Let’s explore how marketers can master attribution to start hitting revenue targets with much greater confidence and certainty.

The Impact of Not Using Accurate Attribution Reporting

The impact of not using attribution reporting - or using it poorly - is worth understanding. It can have multiple negative impacts on decision-making and overall business performance.

The common consequences are:

Incomplete Customer Insights Cause Poor CX

An incomplete understanding of customer preferences, motivations, and pain points hinders the ability to tailor marketing strategies to effectively engage and convert customers.

This can result in a lack of adequate content personalization and a poorer customer experience (CX), meaning your brand gets overlooked in favor of others by potential customers.

Unprofitable Resource Allocation

Struggling to accurately identify the marketing channels, campaigns, or touchpoints that are driving conversions or desired outcomes results in less effective use of resources.

For example, over-investing in underperforming channels, and underinvesting in high-impact touchpoints, wasting budget in the process.

Poor Revenue Growth and Limited Brand Equity

Incorrect assumptions about the impact of specific touchpoints or channels results in suboptimal marketing performance and missed opportunities.

If marketing efforts fail to engage and convert customers effectively over time, the business can suffer from stunted revenue growth, also putting a cap on brand equity.

Understanding the Challenges for Accurate Attribution

Marketers can’t fully rely on free attribution solutions for the insights they need to drive significant optimizations. Results can be significantly misleading when solely using free on-platform attribution reporting.

On-platform attribution issues:

• ‍No Cross-Channel Visibility - On-platform attribution doesn't have full visibility into the performance of other channels, or wider customer journey outside of their own ecosystem, acting as walled gardens. This limited view can make it difficult to understand the true impact of each channel on conversions and ROI (or ROAS). 
• Double Attribution - When using multiple platforms, there's a risk of double attribution - where more than one platform takes credit for the same conversion. This overlapping attribution may cause businesses to overestimate the performance of certain channels or campaigns, and consequent overinvestment stunts overall marketing ROI.
• Inconsistent Attribution Methods - Different platforms apply different attribution rules, leading to inconsistencies in how they assign credit to various touchpoints. This inconsistency can make it challenging to accurately compare the performance of different marketing channels or campaigns.
• Tracking Limitations - With increasing data privacy regulations, third-party platforms may face challenges in accurately tracking user behavior across channels. A custom attribution model can help overcome some of these limitations by incorporating first-party data and other tracking methods.
• Lack of Customization - On-platform attribution reporting may not be tailored to your specific needs, goals, and marketing strategy. A custom attribution model, on the other hand, can be designed to accurately reflect a business's unique customer journey, allowing for more precise insights into the performance of each marketing channel and campaign.

There is a compelling case for marketers to invest in their own customized attribution solutions. Especially when paid media investments start becoming more significant.

However, accurate attribution modeling isn’t one of the most straightforward tasks for a marketing department to tackle.

There are several hurdles to overcome to extract and benefit from the most valuable insights:

 1. Tracking Data Across Complete Journeys 

A typical user journey involves multiple devices, channels, platforms, and time breaks between visits, making it difficult to track a complete customer journey path from the first touchpoint to conversion. Cross-device tracking techniques are needed, such as device matching or probabilistic modeling.

 2. Data Privacy 

Tracking restrictions and cookie limitations can limit the ability to track customer interactions across marketing channels, sometimes requiring workarounds. Yet it's essential to adhere to data privacy regulations, maintain transparency and obtain appropriate consent from customers when collecting and utilizing their data.

 3. Offline Data Tracking 

Marketers may need to implement strategies such as unique identifiers, coupon codes, QR codes, or call tracking to link offline interactions to specific customers and attribute them properly. However, implementing and managing these tracking mechanisms may require additional resources and operational adjustments, including manual data entry from both customers and staff.

 4. Data Quality and Completeness 

Ensuring the accuracy and completeness of the data is crucial for building reliable attribution models. Marketers must establish data quality control measures, address data gaps, and perform regular data validation to maintain the integrity of the data.

 5. Data Integration 

Integrating data from various sources, both online and offline, can be complex. Offline data sources such as in-store purchases, call center interactions, or direct responses may not be easily captured and linked to other digital data. Marketers need to develop data integration processes to build a unified view of complete customer journeys.

 6. Attribution Modeling Complexity 

Choosing the best-fit modeling approach for marketing goals, and accounting for multiple touchpoints both online and offline, adds complexity to attribution modeling. Marketers need to understand statistical models that can properly attribute credit to different touchpoints based on their real impact on conversions. This requires analytical expertise, plus the budget for necessary data tools as marketing complexity grows.

Types of Attribution Data

Data attribution models are nothing without the data that you feed into them.

There are 2 main sources of attribution data.

1. Software-based Attribution Data

This utilizes digital tracking tools, such as analytics platforms or marketing automation software, to track and record user interactions and automatically attribute conversion actions to specific marketing touchpoints.

Pros - It provides objective and granular data on user interactions and conversions, and enables real-time tracking and analysis of customer journeys. The reliance on voluntary self-reporting and subjective recall is reduced.

Cons - Aside from missing touchpoints that are not digital or easily trackable by software, it can be complex to implement and require technical expertise. You’ll need the right tracking set up for accurate data and reliable insights, and the analytics tools.

 2. Self-reported Attribution Data

This is data collected directly from your customers and leads, who share information about the touchpoints that influenced their decision-making process. It’s usually collected via an online form or survey but can also be collected in direct conversation with customer-facing staff and then recorded in a CRM.

Pros - It allows for qualitative data collection, using direct insights from the individuals themselves to capture subjective factors and nuances that software-reported attribution may miss, such as offline interactions or word-of-mouth referrals.

Cons - It relies on individuals' willingness to provide information, and their memory and perception which may not always be accurate or complete. This type of data can be more time-consuming and resource-intensive to collect and analyze.

Hybrid Attribution Data

Combining both self-reported and software-based data sources into attribution modeling is what is known as hybrid modeling. It’s the ideal solution to mitigate the drawbacks of each data type, providing the most fully comprehensive understanding of your customers journeys.

Next, depending on your marketing activity and data tracking sophistication, you’re going to have some of the following types of data sets to work with.

The best way to categorize your data inputs is to split it into channel data and event data.

1. Event Data (What Happened?)

Event data typically includes:

• Conversion Data - Conversion data includes information about the desired actions taken by users, such as purchases, form submissions, or sign-ups. Conversion goals need to be set for each journey stage.‍‍
• ‍Behavioral Data – Any data related to customers' online behavior, such as organic website visits, page views, time spent on site, clicks, search queries, and interactions with specific content or features.‍
• Clickstream Data – This is a record of each click a consumer makes while browsing online. Tracking all these actions can help brands form an accurate understanding of the most effective consumer journey design.‍
• Ad Impressions and Clicks - Ad impressions combined with click data provides information on the number of times an ad was displayed to users, and the corresponding clicks made. This data helps gauge the effectiveness of specific ads.‍
• CRM and First-Party Data - This data provides long-term insights into customer behavior and can include survey responses, purchase history, and any interactions with the brand. CRM data is necessary to link the direct impact on revenue generation and CLV.

Note, there are two common ways to give credit to touchpoints within a conversion sequence: post-click, or post-view.

Post-click Conversion Data - If attribution is done on a post-click (not necessarily last-click) basis, clicked touchpoints will get a part of the conversion credit as long as the action happens within the defined lookback window.

Post-view (or view-through) Conversion Data – Here, the content a user viewed (impressions) within the specified lookback window also gets part credit for a conversion. Most of the advertisers who advertise on multiple channels will have video and social media as part of the conversion journey. These channels usually are not driving clicks, but still contribute to outcomes. This data is more challenging to accurately collect.

The lookback window is how far back a conversion action is included, usually measured in days. So a 7-day lookback window would only include advert impressions or clicks 7 days before the customer converted. In low-cost eCommerce transactions where the selling cycle is short, the most relevant lookback window only might be 7 – 14 days. Whereas for more complex sales like business software, a lookback window of 60 days could be used.

‍2. Channel Data (Where Did It Happen?)

Channel source data typically includes:

• Referral Data - This identifies the source that referred users to your website (or app). It can attribute from the high-level referral sources, such as search engines, social media or email, right down to the specific pieces of content.
• Device and Platform Data – This gives information about the devices and platforms used by users during their customer journey. It allows marketers to track cross-device interactions and attribute conversions across different devices. Device data is also helpful for providing location information.
• Offline Data - This gives information about customer interactions outside of digital channels, such as in-store purchases, phone calls, word of mouth, events, direct mail responses, etc. Offline data is typically captured through mechanisms like unique identifiers, coupon codes, or CRM systems.

An attribution model is essentially used to link these two types of data together to show which marketing touchpoints deliver the best results. 

Types of Attribution Models

There are several types of attribution models to feed your data into. Applying the right modeling for the goal or KPI is key.

A model essentially joins up your event data (what happened) to your channel data (where it happened) to show you the most profitable journey connections. 

The difference between attribution models is where they place most credit for achieving a desired conversion goal (like submitting a contact form, generating an MQL or closing a sale). Conversion goals should be set up for each stage of the customer journey to feed attribution analysis.

Multi-touch models attribute results to more than one touchpoint, allowing for the influence of consecutive touchpoints to be considered as part of a process that led to the final conversion.

 A model either uses: 

• Rule-based methodology - Analyzes data in a completely static approach. 
• Data-driven modeling - Typically uses AI and machine learning to help automatically customize multi-touch attribution based on the influence of touchpoints.

Here’s a breakdown of the most common attribution models:

Last Touch (or Last Click) Attribution

This model assigns all the credit for a conversion to the last touchpoint (or channel) that the customer interacted with before making a purchase or completing the desired action.

When to use it? - To understand which touchpoints are most influential for prompting people to take the final step in completing a conversion goal (e.g., submitting a contact form or making a purchase).

Limitations? – Although easy to use and collect data for, it’s not a great stand-alone model for longer and more complex sales cycles where conversion still heavily relies on the preceding touchpoints, particularly in B2B.

First Touch (or First Click) Attribution

The first touchpoint (or channel) the customer engaged with receives 100% of the credit for the conversion.

When to use it? - To understand which early journey touchpoints are best at first reaching new audience members who will eventually convert.

Limitations? – It ignores the influence that mid to late journey touchpoints have for final conversion. Data accuracy can also be more difficult to assure depending on your data tracking methods and lookback window

Linear Attribution

Equal credit is given to each touchpoint in the customer journey, recognizing the role of all channels in driving conversions.

When to use it? – To understand how touchpoints and journey architecture work together to nurture conversions over time, including cross-departmental touchpoints between marketing and sales for B2B.  

Limitations? – The data collection process is more intensive and may require cooperation with other departments to capture all touchpoints, taking time to implement fully. This may include qualitative touchpoints through manual data entry. Distributing credit evenly doesn’t account for which touchpoints have most influence.

Time Decay Attribution

More credit goes to touchpoints that occurred closer to the conversion event, with the assumption that recent interactions have a greater impact on the decision-making process.

When to use it? – For longer, more complex customer journeys where later touchpoints are most influential. Equally, it can be useful for very short sales cycles where decisions are made quickly and you want to see which touchpoints have immediate effect for impulse conversion.

Limitations? – The influence of earlier touchpoints for creating brand awareness or intent won’t be accounted for.

U-Shaped (Position-Based) Attribution

A higher percentage of credit goes to the first and last touchpoints in the customer journey, while the remaining credit is distributed evenly among the other touchpoints. It's based on the idea that the first and last interactions play a more significant role to create new leads and drive conversions.

When to use it? – When you want to understand which channels generate most new leads, and which drive most conversions.

Limitations? – Again, the influence of in-between touchpoints will not be fully understood, and it requires data collection to cover all touchpoints within the journey.

W-Shaped Attribution

Equal credit goes to three key touchpoints: the first interaction, the lead creation event (e.g., form submission), and the final conversion event. The remaining credit is divided among the other touchpoints.

When to use it? – To highlight the key journey milestones from early journey, mid journey and late journey.

Limitations? – The influence of intermediary touchpoints is not fully understood

Data-Driven Attribution

Data-driven models use advanced analytics, machine learning, or artificial intelligence to analyze customer journey data and assign credit to various touchpoints based on their estimated influence on conversions. This can be done by off-the-shelf software solutions specifically designed for marketing attribution. There are two widely accepted data-driven models for attribution: Shapley value model, and Markov chain model.

When to use it? – For more accurate full-journey attribution across multiple touchpoints, providing greater flexibility for integrating multichannel data silos and more balanced weighting criteria.

Limitations? - An attribution software subscription is required, some of which can be costly. How the algorithms are coded and applied is sometimes proprietary information that is not made fully clear or adaptable. Data sources still need to be set up and connected, including offline touchpoints. It can take months of work to fully set up and implement a data-driven model covering all marketing channels.

Fully Customized Attribution Modeling

Custom-built models are also data-driven, but can include as much complexity and adaptability as you’d like. They allow for full visibility and control of the combined data sets, rules and weighting in use. It allows the layering of many rules and granular data analysis so you can deeply understand and drive growth to a level that isn’t available any other way.

When to use it? - For larger media budgets where small adjustments see the $ results impacted by millions.

Limitations? - Fully customized attribution requires a specialist to implement because of the complicated algorithms and calculations, along specialized statistical software and coding. Like off-the-shelf data-driven solutions, it can take several months to fully implement.

Choosing Data and Models to Match Goals

For multichannel marketing across the customer lifecycle, marketers will have several different goals and KPIs, so there isn’t a one-size fits all when it comes to using attribution modeling. 

For example, marketing goals will vary by campaign, but also business lifecycle stage. As a business matures and can afford to allocate more budget in demand creation, longer payback periods become feasible in the name of sustainable growth.

For the most accurate results, several rules and weighting criteria may need to be layered together. This requires an understanding of how to choose the most appropriate combination for each goal or data set.

Here are examples of how different goals could affect the overall approach for assessing attribution against KPIs:

Brand Awareness - The main objective is to build familiarity rather than immediate conversions, so attribution models that consider upper-funnel touchpoints using a longer lookback window are most helpful.

‍Conversion Rate Optimization - Last touch attribution can provide insights into the most influential touchpoints in driving conversions for any journey stage.

Customer Acquisition – With a focus on identifying marketing efforts that drive most new customers, attribution models that emphasize first touch and last touch before sale conversion are a good fit.

Customer Retention and CLV - Attribution models that consider multiple touchpoints over the customer lifecycle are best. Time-based attribution models such as linear or time-decay attribution can help identify touchpoints that contribute to CLV over time.

Cost Efficiency - Attribution modeling using cost-per-click (CPC) or cost-per-acquisition (CPA) data provides insights into the cost of acquiring customers through different channels.

Channel Optimization - Models like time-decay attribution or position-based attribution can help evaluate the effectiveness of various channels throughout the customer journey.

Return on Ad Spend (ROAS) - Attribution that uses revenue conversion data along with position-based or data-driven models are most suitable for calculating ROAS. These models can help isolate the impact of an advertising campaign against other touchpoints.

Customer Engagement - Attribution models fed with click data are most valuable. Models like engagement-based attribution or position-based attribution can help attribute credit to touchpoints that generate higher engagement levels.

Campaign or Event Success - Campaign-based attribution or event-based attribution allow marketers to filter conversion data specifically for the corresponding campaign (or event) identifier.

Demographic Targeting - Companies that target audience segments based on demographic data, such as geography, need to be able to filter customer event data for segment-based attribution.

Social Media Influence - Models using multi-touch attribution with social media weighting can help more accurately attribute conversions or engagements specifically to social channels.

Experimentation with attribution models will help you find the most suitable approach for each reporting use case.

A Step-by-Step Guide to Building Custom Attribution Models

A customized approach to attribution modeling allows hybrid data usage to give the most complete and accurate view of your marketing effectiveness. (Reminder - a hybrid approach combines multiple online and offline data sources, reducing the risk of misleading insights).

With customized approaches, you can get journey clarity at the individual level. For example, you could isolate a new customer to see that their first website visit was 9 months ago, and they were exposed to 37 ads across 5 platforms. You can also use heat map tools to confirm how channels work together in order to predict where prospects will go next, targeting content messaging accordingly.

Here are the 6 steps to create custom attribution reporting that will truly allow you to start optimizing your marketing investments:

Step 1 - Clearly Define Your Goals

Identify the specific objectives that your marketing efforts aim to achieve, such as increasing conversions, driving brand awareness, or improving customer retention. They can be different for each channel or audience segment. As discussed, these goals will guide the rule options for your attribution model.

For each goal, decide what you consider to be a conversion for the journey stages, and whether you will need to include post-view data in addition to post-click data. The type of conversion is important, so you’ll want to identify the conversion events to look at for each specific goal, including the lookback window that will be most relevant.

Attributing marketing activity to revenue is the ultimate aim – this will give you the most powerful information to improve ROI and drive growth.

Step 2 - Identify All Your Data Sources

Start with accurately and consistently collecting all the data you possibly can for all customer interactions across all your active channels and platforms. You’ll need to UTM tag every link that matters, and have tracking pixels installed for all active marketing platforms.

Here’s a quick checklist of data sources:

• Website
• Social media (organic)
• Paid media campaigns
• Email marketing
• CRM system and revenue data
• Customer feedback
• Call tracking
• Offline touchpoints
• Third-party data providers

Notes:

• Self-reported attribution is most valuable when free text only.
• B2B buying decisions usually involve multiple people, so it’s better to track the customer journey at the account level instead by combining individual user data.

Step 3 - Bring in the Necessary Data Capabilities

Marketers need to have a deep understanding of marketing concepts and principles to be able to set up effective attribution models and make data-driven decisions.

You will need access to strong data analysis skills to be able to set up, manage and interpret the data for customized attribution models. Some technical knowledge is required to select, set up and configure attribution software tools, integrating them with existing data sources and systems. Knowledge of statistics is also necessary to understand, interpret and communicate the results of attribution models.

If in-house attribution data specialists are not in budget (or available), It can be more economical to use specialized data agencies to support you.

Step 4 - Chose + Activate Your Data Tools

Available resources are a big part of your consideration here. You’ll need to consider what is within means for your company in terms of ease of use, data integration capabilities and subscription cost.

There are 2 options here:

• Off-the-shelf attribution software

There are several software tools available that can help marketers combine marketing attribution data from different sources.

Tools with in-built machine learning and AI are better suited to help you analyze and weigh the contribution of different touchpoints and channels in your custom hybrid attribution model. This will give you more accurate insights.

Google Analytics (or Campaign Manager 360) are the best known off-the-shelf providers. However, data integration from other sources can be much more of a challenge with GA. Some other off-the-shelf options which offer better data integration capabilities include Northbeam, Wisely, Adobe Analytics and Improvado. 

However, the drawbacks are that you’re still handing over power to a platform that uses its own proprietary algorithms, not always allowing complete visibility or flexibility in how rules are applied or data is weighted.

• Build your own custom modeling

Depending on your resources, building custom modeling offers the greatest control and visibility of exactly how data is being weighted and analyzed for each scenario. 

If you’re doing this independently, you’ll need a data connector/warehouse solution to import and store your data from across your multichannel data sources. Custom coding and statistical tools can be utilized for advanced capabilities, allowing for layered algorithms and models tailored to any specific need or data set, including fully customized weighting criteria for data sets such as self-reported attribution.

The benefits over any other solution is the most accurate attribution possible, with completely granular insights depending on any criteria you’d like, allowing complete flexibility as variables such as channels, campaigns and customer or market dynamic shifts, and fully aligned for any goal you set.

With customized approaches, you can get journey clarity at the individual level. For example, you could isolate a new customer to see that their first website visit was 9 months ago, and they were exposed to 37 ads across 5 platforms. You can also use heat map tools to confirm how channels work together in order to predict where prospects will go next, targeting content messaging accordingly. 

Step 5 - Integrate Your Data Sources

Using your selected attribution tools, start collecting and integrating data from your multichannel sources.

This involves setting up data integrations between the attribution software and the data sources, whether through configuring API connections (recommended) or importing data files.

Automate the most relevant model-based analysis into dashboards, reporting on each of your specific marketing goals whether by revenue, channel, journey stage, customer segment, etc.

 Step 6 - Test and Iterate

Continuously test and refine your attribution model, adjusting the weights and methodologies as necessary. Monitor the performance of your model and make data-driven adjustments to improve its accuracy and effectiveness over time.

For example, data capture often relies on UTM tags, which requires links to be clicked before they are reported. This means some early-journey channels that rely on impressions rather than clicks (mainly social media and display advertising) will be underrepresented without qualitative self-reported data and weighting adjustments. Lift tests need to be run to help assess weighting criteria.

To test the influence of unclicked impressions, which is common for early-journey touchpoints and channels, you can use lift tests. Lift tests use test and control groups, only showing adverts to the test group. The difference in conversions between the two groups is known as lift, indicating the channel's real impact, and providing a helpful weighting metric. (Audience sample size and segment characteristics are important for statistically valid comparisons.)

Incrementality is a complementary metric to lift.

The Main Takeaways

Marketing attribution is critical to understand the impact of different touchpoints on customer behavior and conversions. 

While various simplistic attribution models exist, building customized data-driven models provides marketers with the greatest control and insight accuracy for their attribution analysis. This is essential to ramp up marketing spend with certainty of generating the required revenue results.

Custom data-driven attribution models offer several advantages over on-platform and Google Analytics reporting:

1. Report Against Goals - Marketers can tailor custom models to their specific business goals, customer behavior patterns, and available data sources. This level of customization enables a more accurate reflection of the complexities of the customer journey and the unique dynamics of the market.

‍2. Understand Touchpoint Influence Across Whole Journeys - Custom data-driven models empower marketers to attribute credit to touchpoints based on their true contribution to conversions, rather than relying on predefined rules or assumptions. And by integrating multiple (hybrid) data sources that include online and offline interactions, marketers can operate with a significant competitive advantage to drive growth forwards.

‍3. Allow Flexibility For Refinement - Custom models also provide the flexibility to adapt and refine the attribution process as the business evolves. You can more easily incorporate new data sources, update algorithms, and fine-tune attribution rules to ensure the model remains aligned with changing market dynamics and marketing activities.

Implementing a custom data-driven attribution model requires robust data integration and advanced analytical capabilities. However, the benefits of improved accuracy, granular insights, and informed decision-making make the investment worthwhile, potentially adding millions of dollars of additional annual growth. Particularly where larger advertising budgets are involved. 

By leveraging the power of custom attribution modeling, marketers can achieve industry-leading business outcomes.

If you need any support scoping, setting up or managing your attribution analytics, the team at Half Past Nine are here to help. We live and breathe marketing data! Just reach out.

‍What To Read Next:

• The New Era of Personalization Explained; A Guide to Building Profitable Customer Journeys With Digital Intent Signals
• Marketing Data Visualization To Fully Leverage Your Sources of Truth 
• Switching to First-Party Data: It’s Time to Stop ‘Renting’ and Start Owning Customer Data 

‍

‍