Attribution Explained in 5 Minutes


Listen as Paul Cruise from Rakuten Attribution gives a quick recap on the traditional view of attribution, covering the basics in less than 5 minutes.

This video is a clip from "Attribution and Segmentation: Why combining data is essential for marketing success" - watch the webcast on-demand.

[Paul Cruise:] So, let's talk about attribution. I'll try and do this, as I said, in 5 minutes - Martin, keep an eye on the timer and see if we can do this. I want to do a quick recap on the traditional view of attribution, and I think this is going to be familiar to most of the people on the call, but on the off-chance that it is less familiar to some of you, I think that it's important we go through it.

So, attribution at the highest level is the process of identifying a set of user's actions - events, or touchpoints - that contribute to a desired outcome, and typically that's a sale, and then it's assigning a value to each of those events. Now, the starting point, when we start working with customers, is to give them the most comprehensive view possible of how their customers are interacting with their brand, and the traditional starting point for that is around tracking, so I'm going to talk about some of the actual build phase of putting a successful attribution project together.

Initially, we'll place tracking across all of the advertiser's web properties, so it's on every page of their website, it's about putting tracking on any apps that they may have, and we're tracking all the things that your customers do as they interact with your website - so, all the searches they've made, the items they've added to the basket but didn't buy, for example - and we'll track multiple conversion and sale types. We then want to add tracking to all of the marketing channels that the business is using. We integrate directly with Google and Bing, accounts with paid search, we'll append tracking to display creative, we'll append tracking to affiliate links, et cetera - you get the idea. Importantly thought, we're tracking both converting and non-converting journeys, and we do do that to the most granular level, so for paid search, it's down to a keyword level, we track display impressions, so it's down to a creative level for display, down to a publisher level for affiliate.

But there will be data points that are important to have as part of the user journey for the purposes of attribution which we can't track via our own technology, and when that's the case, we'll ingest that data. So, these are just a few examples: cost data, so all the associated cost data for your various marketing activities; it could be that we could ingest some of the trading side data that you have, so it might be margin of particular sku's, or levels of returns for different sku's; and then there'll be, for many businesses, conversions that occur offline, so that might be on the high street, or via call centres, so we'll ingest that transaction data.

Now, people want to buy from you using different devices at different times, so we'll seek to join those devices at a user level wherever we can, and that gives the advertiser a more accurate, cross-device view of the customer journey. Similarly, as I've just mentioned, for online to offline, if we can stitch together those different pieces of the journey to give the advertiser a good view about how their customers are interacting online and then purchasing offline, that's something that we'll do.

So once we have that really rich, cross-channel, cross-device, online to offline view of the user journey, then we'll take that raw data and we'll start to put it into structures that are more useful for the business, so we can apply attribution. Of course, attribution and the way that we do that, the algorithms et cetera, is a whole webinar in itself, and it's not really the topic that we're concentrating on today, so I think the headline is that we help advertisers move away from traditional first- or last-click models to an attribution model that helps them power better data-driven marketing decisions.

So once we have that build phase complete and we've applied modelling, the traditional output to the advertiser is about budget optimisation, so, it helps them understand if there are channels and activities that they've been overvaluing in the past. Conversely, are there channels and activities that they've been undervaluing, and different attribution providers will help you take action from that in different ways. We have a strong service element to what we do, so we actually work with our customers and make recommendations as to what they should be doing differently.

So, that, hopefully, is my 5 minute introduction to attribution.

[Martin Reed:] Exactly 5 minutes, Paul.

[Paul Cruise:] Perfect.