How to Use Attribution in Marketing

Attribution can help you look in the right place to avoid costly mistakes. However, you need to understand and apply it correctly to get the right insights. In this piece, we’ll go over attribution and how to use it to help you make the right decisions in marketing.

Why is marketing attribution important?

You’ve heard it a million times; the buyer’s journey is ever so complex and trying to track everything your audience does can be a headache! Despite this, your CEO might have said something like this to you: “if it’s digital, we should be able to track everything so why can’t you give me the ROI for every activity we do?”

Unfortunately, this isn’t always completely true and it can be difficult to explain to your stakeholders. Although digital is easier to track than traditional marketing activities like TV advertising, there are still gaps and data can be interpreted in many ways, leading to costly mistakes. Attribution doesn’t fill all the gaps but it can still be a great ally!

What is attribution?

Marketing attribution uses data to measure the impact of each marketing channel or activity based on a specific attribution model. This can be used to understand ROI (Return On Investment) from your marketing and sales activities, or to decide where to invest your budget, time, and resources.

Think about it like a football match. The person who scored the goal didn’t do it on their own, their teammates had to pass the ball at the right time and with the right power for them to score. Attribution uses a mathematical model to calculate the impact of each teammate (marketing/sales activity) across the journey. Each activity is then given a weight depending on how it contributed to the goal (sale/conversion).

When analysing your overall marketing and sales activities, you can get a more accurate picture of which ones were the best at reaching your goal, allowing you to make informed decisions.

Benefits of attribution

Attribution paints a more accurate picture of how people converted online (and offline if you can get this data). Omitting some interactions from the overall picture or attributing the same value to every interaction doesn’t always represent the reality. Some activities will be more convincing than others and the action that sealed the deal might not have been the defining one.

Here are a few benefits of using an attribution model to look at your marketing/sales data:

• Allocate your budget to higher performing activities/creatives to maximise ROI – attribution will show you which channels or content performed best so you can keep improving and getting the best return from your time/money investment (think about how you justify asking for more budget to your CEO).
• Better understand your buyer’s journey and how they decide to buy from you – attribution models can show you at which point of their journey your prospects preferred using each channel. From this, you can adapt your messaging to the stage they’re at to ensure it resonates best with them.
• Identify activities and channels that don’t perform well to improve them – it’s great to invest in what’s working but don’t forget to look at what’s not working! Why is this channel or piece of content not contributing as much as the others? Is it because your prospects don’t visit this channel or don’t care about the content? Or could you improve and convert them even better if you tried?

Limits of attribution

As mentioned earlier, you can’t track everything, even with digital. Cookies expire or people don’t accept them; some channels like WhatsApp automatically fall under dark traffic and can’t be tracked; people browse your site in a fractured way across different devices and browsers or leaving long periods of time in between their visits; people have real life interactions that can’t be tracked; and finally, they don’t always follow logical paths!

Attribution helps you to be more precise about your buyer’s journey but it simply cannot fill in all the gaps and will never be 100% accurate, whatever model you use.

• Attribution is only ever as good as your data – ensure your tracking is set-up correctly, your data is combined between different platforms, and your teams collect data correctly too.
• The data you collect is based on behaviours, not the reason behind people’s choices – data will show you what a visitor has done but won’t tell you why. Talk to your customers and prospects to understand why they choose you and why they convert.
• Accept you cannot track everything – privacy is a growing concern among consumers and they’re actively avoiding data collection by declining cookies or using blocking software/tools. The disappearance of third-party cookies is also affecting how we track data across devices and browsers.
• Platforms are biased – ever wondered why Facebook Ads might report higher conversions than Google Analytics? It’s in their interest to show you attribution models that are biased towards their platform so you spend more with them. Keep that in mind when looking at attribution within different platforms!

Rand Fishkin from SparkToro wrote an insightful piece on why you sometimes need to trust your gut instead of trying to solve attribution.

The different marketing attribution models

There are many attribution models used across varied platforms. For simplicity, we’ll go over the attribution models used in Google Analytics as it’s usually the widest platform used to analyse marketing data. You might come across other models but these are the most common ones used across various platforms.

Last and first click attribution models

Last and first click are the most commonly used attribution models, and they shouldn’t be! First click gives 100% of the credit to – you guessed it – the first touchpoint when last click gives 100% of the credit to the last touchpoint.

When to use?

They can be useful to understand how you acquired visitors or which touchpoints tend to ‘close the deal’. However, avoid using these models to make decisions on which channels to invest in as they will be heavily skewed.

Position-based and time decay attribution models

The position-based model gives 40% credit to the first and last touchpoints, then distributes the 20% remaining to the other touchpoints in the middle. This is because the first and last clicks are usually the hardest ones to get (attracting a visitor for the first time and closing the deal).

The time decay model gives most of the credit to the touchpoints as they get closer to the conversion. This focuses on closing the deal as the hardest part and that’s why it rewards proximity to the conversion.

Source: Google Academy on Air

When to use?

Time decay is best suited for market leaders with a conservative growth strategy. They’re already well-known in their market and so the first click is easier to get. What matters to them is closing the deal, which has more value in this model.

Position-based is better for businesses new to the market with a growth-oriented strategy and high competition. In these markets, the first and last clicks are the hardest to get and therefore, the most important.

Linear attribution model

The linear model simply distributes the credit equally between all touchpoints.

When to use?

It’s great to look at every touchpoint without bias and truly understand the full picture. We use it a lot to understand how channels are performing when a data-driven model isn’t available.

Data-driven attribution model

The data-driven model distributes credit statistically between all touchpoints. To do this, it compares different paths, calculates conversion probability for each, and then attributes a weight to specific touchpoints based on their incremental impact. See an example below:

 

Source: Google

When to use?

If you have enough data, we recommend using this as it will be the most accurate representation of all. However, it’s not always available for smaller or niche businesses as it needs enough data to be able to calculate statistically significant probabilities.

Choosing the right attribution model for your business

There are many attribution models out there, from the most popular ones we’ve just highlighted to more specific ones within different platforms or even models you can build yourself. Choosing which one to use will depend on what problem you’re trying to solve.

Identifying the right channels to invest in

Use a data-driven or linear model to ensure you capture all touchpoints throughout the full journey. If using GA4, ensure your channel data is being attributed correctly and create custom channel groups if needed.

Understanding where traffic comes from

Use first click attribution to identify which channel brings you the most traffic. Be mindful of cookies though as they’ll expire after a certain period of time. If your buying cycle is longer than 90 days, ensure you have the right platform to capture this information as GA4 might not be as accurate. A platform like HubSpot could be useful as it keeps track of the original channel a contact came through.

Understanding which channels ‘close the deal’

Use last click attribution to see which channels tend to convert for your business. This information can be used to optimise these channels for conversions to maximise revenue. A platform like HubSpot might even go further than GA4 as it will also have information about deals for B2B rather than conversions like signing up for a demo.

Identifying best performing content or assets

Use data-driven or linear attribution (depending on your business, you could also use position-based or time decay) so you record all touchpoints.

Beware of platform biases! If you’re analysing the data across channels, ensure you use a ‘neutral’ platform like GA4 or HubSpot. Bias shouldn’t be a problem if you’re analysing the data within one specific channel like Facebook. Since you’re just interested in which assets performed best within this platform, the bias in attribution won’t be an issue. Remember the platform might still over-report on conversions or revenue though.

Same advice if you’re running ads, we tend to choose data-driven or position-based for our clients as this covers all touchpoints. This means you might see decimal numbers for your conversions when analysing the data i.e., 50.7 conversions from Google Ads. This is because part of the conversion will be attributed to another channel.

Calculating ROI or Cost per Lead

Use data-driven or linear attribution so you look at the full journey and touchpoints. A platform without bias is critical here as you could end up with very different results. We tend to use HubSpot (sometimes with data synced from Salesforce for lead gen businesses) as it has all the data of website visitors AND touchpoints after the onsite conversion. For lead gen, GA4 will only show you data until visitors have converted on the site but won’t have information about the lead quality or what happens next with sales whereas HubSpot/Salesforce do.

Marketing attribution by platform

Most analytics platforms will offer attribution modelling. The most used is Google Analytics but it’s limited to data linked to your website. HubSpot and Salesforce can be used to analyse data once someone has converted and is handed over to sales. For lead gen businesses, it will provide valuable information on deals and revenue to help you calculate ROI and cost per lead.

You can also use attribution models in advertising platforms like Google Ads, Facebook Ads, LinkedIn Ads, etc. As previously mentioned, be careful with what they say as it’s always in their interest to show you good results!

Attribution in HubSpot

HubSpot provides non-anonymous data as it tracks activities, touchpoints, and engagement per individual, their company, and the revenue generated from any associated deal. Therefore, it can provide a much more detailed picture than Google Analytics.

It’s great for both lead gen and ecommerce as you can look at attribution based on leads created, deals created or revenue (the last two need an Enterprise license). For its attribution reports, you can look at assets, interactions, channels and sources, and campaigns. This means you can create quite detailed reports which are more accurate when the content is hosted within HubSpot i.e., website pages or landing pages.

However, don’t get carried away with all these possibilities. Attribution works best when looking at the overall picture rather than going granular. Too often we see clients wanting to dig deeper into individual contacts which defeats the purpose of the exercise and can turn into a rabbit hole. HubSpot attribution models are best used to understand the overall picture of how channels or assets are performing.

Another caveat is that, depending on the model you’ve chosen, revenue might not always add up to the actual total you’ve received. This is because HubSpot is crediting various touchpoints following a certain pattern which might not always equate to 100% exactly. Using these models helps you to understand which channels or assets are performing best overall, but, as we’ve said before, it won’t provide you with a 100% accurate picture.

Finally, remember that some B2B buying journeys are extremely complex with many people involved in the decision making process and a lot of back and forth that can last for months, if not years! Deal attribution is therefore complex by nature for these businesses. Imagine multiple contacts associated to a deal, each with different touchpoints that will report differently depending on the model you choose.

For example, first interaction would give 100% of the credit to the first interaction chronologically amongst the contacts, which depending on the structure of the business, might not be a fair representation of the importance of that touchpoint.

Get expert marketing analysis for your business

If you use the right platforms with clean data, attribution modelling can provide essential insights for CMOs and marketing managers to make the right decisions. When you apply the right models to solve your problems, you can gain competitive advantage by discovering insights that will help maximise your budget.

However, remember that although digital provides a lot of data, it isn’t bulletproof and there will always be gaps that just cannot be filled. From privacy issues to real life interactions, some things just cannot be tracked accurately. As a marketer, you will sometimes need to use your gut to make some decisions or take some risks to test new hypotheses where data just isn’t available.

Innovation Visual has a team full of bright marketers. Whether you need help with collecting accurate data through tracking or setting up insightful reports that use the appropriate attribution models for your business, we can help! Contact our team of experts today to help you make better decisions and maximise your budget.