I think when understanding your user journey - if you're going to start out doing this, I would do it with your new and existing customers. Most of you - or all of you, I would have hoped - would have some sort of reliable way of knowing whether a customer is new or existing. So, you could recognise an existing customer when they purchase because you've stored something unique, probably an email address, where you've got a customer ID against someone, or 'identifying new' being a cross-check. I don't know if you have looked into how accurate it potentially is, but if you've got that view, you understand if you've come into some sort of contact with that customer before if they've purchased.

So when you're looking at new and existing customers, are you able to see if their behaviour is different when they're coming to the site? With the attribution view of that journey, the richness of your insight really depends on how engaged you are in feeding us data - Paul mentioned earlier that we have the capability to ingest all the data that we don't have tracking solutions for - so you begin to build out that user journey. Once you've got that view, you can start to see if there's any patterns in behaviour between the new and existing customers and the way they get to the site. You would expect that your new customers would probably be interacting with your display, your SEO, your social media more so than your existing customers that know your brand, which would probably interact with your PPC, your email and direct.

But another basic thing we do with the user journey, that one of my clients finds particularly interesting, is just being able to see when someone actually started to interact with your site. If you're looking at the last click, you have no idea when somebody began researching - you don't know that they were looking for a product that you offer three weeks ago, you don't have that visibility, you don't know where they started out. You don't know that they interacted with you on Facebook by chance and therefore came to the site, signed up for your emails and slowly, as payday came around, they made that purchase. So having that view alone means that you can start optimising your channels.

So, do you want to start sending out an email a particular number of days before you have a sale? Other things; like who does the most research, is a new customer more of an impulse purchaser than an existing customer? You really can't see these things unless you have that data. So when you do, you want to optimise your channels, feed your data out to anyone that's going to be beneficial to you, like your search agencies, to get your search keywords optimised, as this will reduce your spend and ultimately help increase your ROI.

So, a more advanced look at how segmentation overlaid with attribution would be... Targeting. So, knowing your customer types and targeting them differently, different messages at different times. Like I said, when a certain customer type will start looking for your product or service, or maybe it's something they do regularly. Reducing spend - so, stop marketing to people who don't react to your generic messaging - make sure it's as relevant as possible, so people are as fully engaged as they can be. Even playing around with things like revenue thresholds - do you recognise that your younger, more sporadic shoppers only buy from you once a year? But if you were able to recognise those users and target them with some sort of offer to get them to engage with you, do they turn into a regular shopper? Do they spend more with you? Are you making more money from people that generally wouldn't?

The other side of that would be to increase your PPC bids for those users who you know buy your higher margin products, so again, if you know your user types, you can start to target them differently. I mean overall, your conversion rates, as you start to put the right messaging to the right user types, will increase, and you're going to get more from your money.