Multi Channel Funnels in Google Analytics


Google Analytics – like the majority of web analytics software – currently reports using last-click attribution. If someone first encounters your site by searching for “mobile phones”, browses around and then leaves and later returns a week later searching for “Your Brand Name”, clicks a PPC ad and then buys a phone, Google will class that as a sale that’s attributable to PPC.

If you like to filter for brand and non-brand sales, you’ll also see that as a branded sale.

The data you get here most likely informs the decisions you make. Based on just that data, you might choose to bid heavily on “Your Brand Name”.

I’m not saying this is necessarily wrong – this isn’t an argument about last-click versus first-click. It’s an argument about making decisions using only a limited view of data.

The problem with the example above is that you’re not able to see the full picture – it’s not immediately clear that the non-brand keyword “mobile phones” helped to bring in that sale, so when you’re making decisions about which keywords to try to rank for, or how much money you invest in PPC vs SEO (vs display, vs social etc), you may be making that decision for the wrong reasons.

A short while ago, Google Analytics began to beta test a feature they call “Multi-channel funnels” – their version of multi-touch analytics – which helps to reveal that bigger picture. In the previous example, where someone searched for “mobile phones” and then converted on “Your Brand Name”, multi-channel funnels will be able to show how that non-brand keyword assisted in the sale.

While there’s a few reports that it brings up, my favourite is the Top Conversion Paths report, which looks like this:

The report above shows the most common paths that led to a sale (for sales that required the customer to visit more than once). It also shows the number of conversions that path brought in and the amount in sales that it’s affected (which I’ve hidden). The most common path in the example above is a visitor arriving twice, both through organic search. The second most common, however, is someone arriving through organic search and then returning to buy at a later time, by going direct. (It’s worth mentioning that, in cases where the last visit was direct, the non-multi-touch version of Google Analytics would most likely attribute that sale to organic search anyway – but only because of how Google Analytics deals with “direct” visits, you can find out more on that from this excellent post by Wil Reynolds).

What’s more useful about this report is the ability to change the segments that Google uses. If you want to drill down to individual keywords, then you can. If you want to split it between brand & non-brand, then you can (and I’d definitely recommend it). If you set-up your own categories (using your own defined rules, which are nice & easy to set-up), you can get reports that look like this:

The beautiful thing about this is that it allows you to make more informed decisions. If you discover that people that visit via Facebook and Twitter later return (via branded PPC, SEO or direct) and buy, then you might decide that investing a bit more in social is a good decision. If you discover that non-brand SEO helps to bring in a larger amount of sales than you previously thought, you might be more confident with investing more in SEO.

Those two examples are based in part on an Econsultancy post by Tagman, which suggests that both SEO & social are hugely undervalued using just last-click attribution (from what I’ve seen – it’s true that they’re undervalued, but not to the extent that the Econsultancy post shows, sadly. Although it may simply be that it’s very different from retailer to retailer).

The ease in which you can set up custom segments with Google’s multi-channel funnels is amazing. There’s a few interesting tips & tricks that you can use, too – while the set-up that I use most often involves breaking down the segments into:

- Brand SEO
- Non-Brand SEO
- Brand PPC
- Non-Brand PPC
- Direct
- Social

I also like to segment results by “intent”, to help get a better view of who is browsing around in general, versus who is actually looking to arrive at the site. You can do this by creating a group called “Branded Intent” (i.e. people that specifically want to arrive at your site) and adding Branded PPC, Branded SEO and direct traffic to it.

That will allow you to see paths that look like this:

Another interesting & useful method for segmenting is to filter channels by informational queries versus commercial queries, so you could see whether people searching for things like (for example) “can iPhone 4 be used as a wifi hotspot” later end up buying an iPhone 4 from your site, at a later date. (Hat-tip to Mark Edmondson for that idea). If you knew that search terms like that later resulted in sales, you could have a case for building out informational content on your site.

The important thing to remember about Google’s multi-channel funnels is that it gives you the ability to make more informed decisions. There’s no word yet on when (or even if) it will roll out for everyone, but when it does I recommend jumping straight into it.

Multi Channel Funnels in Google Analytics is a guest post penned by Dave Peiris (@SharkSEO) – a freelance SEO based in the UK.

Image credits:
Sausages for Supper


  1. Robert

    Just in time.

    I was about to pause a number of costly ‘review’ keyword phrases to test their impact on conversions, but now I’ll wait for my beta access.

    Cheers Dave!

  2. Liz at OneResult

    This is a really exciting feature and I’m looking forward to being able to have a look and show our clients more about the behaviour of their users. Please let us know if you hear more about beta or full access rolling out across more accounts

  3. 8 Gram Gorilla

    An excellent feature, for sure, and it looks like a great way to really pinpoint the path to sales across multiple visits which, as you say, has traditionally been the tough thing to measure.

  4. Jason Hong

    Wow, is this really going to do what I think it says? Tracking the origin of a sale has always been difficult, unless you implemented cookie sharing or something of that sort. But reading this article, it makes me believe that a click from Twitter one day, then a direct purchase a couple days later can be tracked? Can we get data on how many days since the original click the customer purchased via direct? Awesome read. Very informative.

  5. Vincent

    This is a really exciting feature for brand exclusion I am using utm_nooveride in adwords (branded campaigns) + gaq.push(['_addIgnoredOrganic', '']); but this new feature is just exactly what where all waiting for to take better decisions.

  6. paul morris

    sharkseo – nice detailed article.

    Nice to see that Google MCF is out of beta and free for all. It’s still not ideal e.g. basic reporting and only allows for a 30 day cookie prior to the sale (hence not ideal for longer sales cycles such as mortgages) however it is certainly a welcome tracking addition.

    my post details how it will impact the wider digital marketing industry – – however to save the click here is my summary: Now that MCF is out of beta it will certainly impact the way that small-medium sized business’ track click value/ sales, it will shake up the tracking industry (e.g. the big tracking boys have got to be nervous and the affiliate marketing channel may start to move away from predominantly relying on last click attribution) and the majority of businesses should start to value the other channels that play a part in the sales funnel but are often not the last click e.g. more generic ppc clicks and social media.

  7. Adam Beaumont

    Good tips , not used this feature but will try and use for some of my e-commerce clients, hopefully get some interesting results.