Those who are relatively new to digital analytics might not be familiar with the term “assisted conversions”, so I’ll start this post with a brief explanation of the concept.  

People don’t always buy a product the first time they’re exposed to it. It sometimes takes a great deal of exposure (either through friends’ recommendations or through advertising campaigns, or both) to awaken our interest and finally convince us to make a purchase. 

 

As far back as the brief history of digital marketing goes, the last click in the user’s journey was traditionally given all the credit for the conversion. This obviously resulted in inaccurate return on investment measurements, as certain digital channels are good at raising awareness, but aren’t very effective in getting visitors to ultimately convert. 

 

Monitoring assisted conversions is possible nowadays with tools such as Google Analytics, although GA’s default method for measuring conversions still is the last click attribution model (or rather the Last Non-Direct attribution model, but we’ll get back to that later). Probably rightly so, since it’s fair to assume that the click who ultimately got a user to purchase a product deserves most of the credit. And that’s precisely the point I’d like to tackle with this post: last clicks deserve most of the credit, but not all of it. 

 

 

Assisted conversions

 

Assisted conversions are, according to Google, conversions to which a certain channel contributed but that were ultimately completed through a different channel. 

 

Google Analytics’ Top Conversion Path feature allows you to see your assisted conversions broken down as unique conversion paths. Here is an example of a conversion that was initiated by an organic search, where the user came back to the site through a direct access (possibly a bookmark), and later on came back through a paid ad, to ultimately convert through a referral site.

 

The figure below gives us an aggregated overview of different digital channels and their corresponding assisted conversions.

 

In this figure, we can see that Paid Search acted as the final touch point for a total of 8.756 conversions. But wait ! The same channel also assisted 1.799 conversions, although these were ultimately closed by a different channel. Which means that the money you’ve been spending over Paid Search indirectly helped generate an additional 20% conversions through other channels ! Now that is something your boss might be happy to hear about…

 

The ratio Assisted/Direct Conversions in the far right column tells you how good a channel is at assisting conversions. High ratio values indicate that a channel is good at assisting conversions but not so good at closing them, and low ratios indicate channels that typically act as the final touch point in the conversion process but that aren’t very good at raising awareness about your products.

 

 

Attribution models

 

According to Google, an attribution model is the rule, or set of rules, that determines how credit for conversions is assigned to touchpoints in conversion paths. Google Analytics offers 7 types of attribution models by default, but also allows you to customize your own attribution models. Agencies who invest significant amounts of money in paid advertising for their clients usually love attribution models, because the collateral conversions caused by display and PPC ads translate into an overall return on investment that is higher than what is recorded in the clients’ CRM platform.

 

Why your CRM isn’t telling the whole truth

If you happen to be working in digital marketing, you may have noticed a difference between the leads recorded in your CRM for a specific channel and the conversions displayed by Google Analytics for the same channel. 

 

Typically GA will display a higher number of conversions than your CRM for a specific channel. The main reason behind that gap is because Google Analytics counts Last Non-Direct clicks as conversions. This means that, if you’ve clicked on an Adwords ad, didn’t convert and came back later on to complete the conversion through direct traffic, Google Analytics will by default attribute that conversion to Paid Search instead of Direct Traffic. 

 

 

Your CRM on the other hand probably wouldn’t be able to assign that conversion to Paid Search, as it is most likely configured to record conversion channels based on your browser’s URL parameters which can only record Last Interactions (unless you’ve set up your CRM to collect cookies instead of URL parameters, as I explain in this post)..

 

Last Interactions versus Last Non-Direct Interactions 

The following figure shows us a comparison between Last Interactions and Last Non-Direct Interactions. Paid Search generated 1.275 last interaction conversions, which will most likely match the data in your CRM, as Last Interaction is the primary method through which CRMs collect URL parameters in order to attribute channel sources to your leads. 

However, we see that 1.327 conversions were recorded as Last-Non Direct interactions. The 4.08% difference between the two are leads that Google Analytics counted as Paid Search conversions, but that your CRM is counting as Direct Traffic conversions. The same applies to that 5.27% difference under Referral conversions. And finally, the third row tells us that about 70% of what our CRM counted as Direct Traffic actually results from other channels (Paid Search and Referral in this case). 

 

The following figure taken from the Multi Channels Overview tab in Google Analytics is another great way to show your boss how most of the conversions that are being labeled as Direct Traffic in your CRM actually come from Paid Search and Referrals. 

 

 

 

Attribution Models’ shortcomings 

For most companies, attribution models mostly serve one purpose, which is to raise awareness as to the importance of collateral conversions that aren’t visible in your CRM system in order to help you justify your Paid Traffic investments. However, keep in mind that the attribution data displayed in most web Analytics platform (including Google Analytics) is far from being accurate.

 

The main reason for this is because Attribution Models work with browser cookies and therefore don’t work across different devices. Many people nowadays use more than one device for the different steps of the conversion process, and these people’s behaviour can’t be monitored with Attribution models. 

 

My recommendation is to make sure your top management understands this, because it actually means that, taking into account cross-device conversions, the collateral effects of assisted conversions is even greater than what you can gather from Google Analytics, which again is likely to help you justify your Paid Traffic investments.