Some call it multi-touch attribution, others call it engagement mapping. Google calls it Multi-Channel Funnels and it has to be one of the biggest talking points in online marketing measurement circles today.

Whatever you want to call it, it is now live in limited pilot for some lucky Google Analytics customers.

A quick note: Google has let us know that the feature called Multi-Channel Funnels discussed in this blog post is in limited pilot. That means that Google is testing the feature and its usefulness to a small group of trusted testers, and has not made any plans or a timeline for a full launch.

To be clear, what we’re talking about here is the end of last-click-only attribution in probably the most widely used web analytics platform in the world (if launched to everyone). This is a feature which until now has been available only to those with big annual web analytics budgets.

It will allow the committed digital marketer to answer those burning questions they’ve been trying to get their heads around for a long time now:

“Should I be advertising on the head or focussing on the tail?”

“How does search and display interact?”

“Has anything good ever come of Tweeting?”


Anyone familiar with the Google AdWords platform will know that Google launched AdWords Search Funnels back in March 2010. This launch will take things a huge leap further and allow analysis of all online marketing channels through to conversion. Below I am presenting the four aspects that I find most useful in ascending order of usefulness. 

A final note of caution before we commence: I am using highly sensitive data from our own GA account so any competitors should look away now.

Understanding how many interactions lead to conversion

The first question I asked myself was similar to the above. Essentially; is the analysis worthwhile at all? The Path Length Report helped me to decide that.  

Using our data we can see that 28% of all conversions involved more than one interaction prior to conversion. This seemed substantial enough to warrant further investigation. 

Path Length Funnels

Notice also that 57% of all conversion value occurred after more than one interaction (below). This leads to the most important addition to this feature from previous iterations: You can now look at path lengths for different goals individually.  

For example, here we’re looking at the path lengths for a new phone call goal tracking initiative. Its early days yet but it seems users are more likely to have two or more interactions prior to calling than for all goals rolled into one.

Path Length Call Tracking

Weeding out your sowers and harvesters

In addition to seeing the final interactions we can also see the first interactions and intermediate or ‘assist interactions’ using the Assisted Conversions report. To simplify things in my own mind I have split interactions into two camps – Sowing and Harvesting:

  • Sowing channels are those at the top of the funnel. These are the channels involved in the early stages of the buying cycle, sowing the seeds to be harvested later. Understanding which channels these are and how to optimise their delivery is crucial to the medium/long term health of our online business.  
  • Harvesting channels we know from old, they are the last-click converters. These have always been visible to us but now we can understand how they interact with their seed-sowing neglected cousins.

Here I have chosen to focus on the sowing channels which were responsible for the first interaction that converting users had with the site. Firstly in raw numbers i.e. which channel/s seeded the most users who went on to convert? 

First Interaction Conversions

That is fine by itself – but what if those channels are also major harvesters? We won’t have learned much that we didn’t already know. Google recognised this and gave us a ratio to play with. See the right column above for the ratio of first to last interactions or seed-sowers to harvesters:

  • A ratio below one means that the channel is more likely to be a harvester.
  • Above one and the channel is more likely to be sowing the seed for you to harvest later. 


Default and custom traffic grouping

A spot of multi-coloured visualisation never did a tired analyst any harm. With Default Traffic Groups you can start to compare your cornflour-blue (my wife’s words) paid advertising with your light green organic search. 

A very nice, straightforward way of grouping channels as a whole, but that is not where the genius ends……

Custom Groupings

Using the excellent Google Analytics segmentation capabilities we can even create Custom Groupings of our own and test and retest different grouping models, for example analysing brand vs generic search across multiple mediums and platforms.

Finding lost keyword & channel interaction data

I hope I’m not being too dramatic when I say that the Top Conversion Paths report answers that most famous of marketing questions, here paraphrased for precision and efficiency: 

“Which half of my online budget am I wasting”? 

Exactly how we answer that, we’ll see in a moment. What I’d like to point out first is all the stuff you thought you were wasting time and money on but may not be after all.

I’ve taken two out of many possible illustrative examples here which I’d be confident you could all find equivalents of in your own data. 

Example 1: Organic search interaction with ‘ppc management’ leads to a paid search interaction on the same keyword which then leads to a brand search conversion.

Hurrah! Suddenly the £3 per click and hours of SEO work we spent on the keyword ‘ppc management’ is all worthwhile! Well done on the SEO, Jimmy.

Multi Keyword Funnels - PPC

Example 2: Twitter interaction leads to paid search brand term interaction conversion. Who says Twitter isn’t a direct response tool? Keep up the good tweets Jeremy.

Multi Keyword Funnels - Twitter

This last dataset is going to find many fans out there, particularly amongst those of us who own ‘I Heart Spreadsheet’ mugs.

A word of caution to the rest of us: It will require analysis. The answers will not jump out of this report and turn your bid management dials for you.

So – which is the wasted marketing spend? Quite simply it’s the bit that isn’t showing up in any of these reports. If a traffic source or keyword isn’t visible in this report then it isn’t driving users to convert. It is neither seed sower nor harvester. One could argue that’s wasteful.

The future of Google Analytics

Multi-Channel Funnels are a feature of the Google Analytics V5 Update which saw the introduction of some other nice new features. There is no indication that Google is going to slow down the introduction of new features so there may be more to look forward to.

Finally I know that this subject might be frustrating for those who don’t yet have access to the Funnels so I’ll try and anticipate some questions here:

  • No it doesn’t currently allow custom attribution modelling.
  • The window for the conversion data is 30 days.
  • Once the feature goes live all you will need goal data to get stuck in. Get started collecting the data now as there is a good chance it will be retrospectively available on launch.
  • No there is currently no mechanism for including offline marketing attribution.
  • Yes you do need to tag your marketing campaigns comprehensively and consistently.

Any more questions please feel free to check out the Google video for a functional description of the feature or, preferably, leave a comment and I’ll respond.