Most ecommerce businesses invest in a range of digital marketing channels, so working out the exact attribution and ROI can be incredibly complex.
For example, the importance of search can often be overstated, as that tends to be the last step on the path to conversion.
To try to develop a better understanding of its marketing attribution, Air New Zealand began using a tag management system two years ago.
The ecommerce team found that the assumptions and investments that it made based on a last-click model were hugely inaccurate, particularly when it came to display.
To find out more about how tag management impacted Air New Zealand’s attribution model, I spoke to UK and continental Europe online channel manager Chris Allison...
Can you give us some background on Air New Zealand and the importance of ecommerce for the business?
Five years ago we probably got about 10% of our business through ecommerce, but that has grown quite significantly and it now represents around one quarter of our total business in the UK and continental Europe.
So it’s become one of our most strategically important revenue channels, which I guess is no different from any other travel supplier out there.
And we are seeing continued growth in ecommerce at the moment and have fairly aggressive growth plans for the channel considering the size of our brand in this market.
So what is your specific role within the company and what are your objectives?
As online channel manager for UK and continental Europe my remit here is essentially on acquisition, conversion and retention from our .com sites.
We currently operate four sites here in Europe: a UK English site, a continental Europe English site, a French language site and a German language site. All of which are managed by my team here in London.
And we have an objective of increasing the channel’s contribution to total region revenue, specifically focusing on acquisition and conversion.
What are the most important digital marketing channels for the business?
The importance of various marketing channels has changed quite a bit in the past two years.
We had a traditional reliance on paid search as our core, bread and butter conversion channel, but we’ve been developing our data model quite significantly and also use email as a cost effective and efficient way of communicating and converting customers.
One of the things we’ve been looking at over the past 18 months is the importance of the display channel, which has been driven mainly by our relationship with Tagman over the past two years.
We’re probably in a position now where we place display on a par with where we considered paid search to be two or three years ago, and that’s thanks to the technology and data that we now have access to as a result of working with Tagman.
So previously what was the relationship between the different channels? Was each channel siloed?
That’s absolutely right. If we go back two years we understood the individual contributions of each channel quite clearly but we weren’t necessarily looking at how the channels worked together. We very much analysed our data in a siloed manner.
And in the position we’re in where you need to drive up the ROI on every marketing dollar that you have, it’s particularly important to understand the interrelationship between the channels and see how one can feed the other, which you wouldn’t really get from the traditional last-click model.
As a result of the work that we’ve done now, I think that it’s probably quite dangerous to look at your marketing on a last-click model as you’d be making entirely different decisions altogether and not really investing in areas that contribute to conversions.
So what was it that initially made you want to move beyond a last-click model?
We wanted to try to get a better understanding of the attribution from display, particularly when you are trying to justify spending in that area when it isn’t necessarily showing a positive ROI versus other channels on a pure last-click attribution model.
Anecdotally we always felt display was important but couldn’t really prove it based on our previous model, or lack of one, so the real key for us was opening up the attribution right through the purchase funnel so we could see how each channel interacted with each other and at which point during the purchase cycle.
Ultimately we needed to adjust our planning and investment accordingly across each channel to try and optimise the overall mix of channels more efficiently.
Were there any political or technical challenges around implementing a tag management system?
I guess with more ecommerce businesses there is a fear within IT teams about giving up control of the systems to marketing professionals, but the way we sold it in was trying to reduce the workload for that team from a purely mechanical perspective of adding and removing tags.
So in terms of implementing the tag management system itself, the conversations were very much around improving efficiency for the IT teams while they were nervous about giving up control.
But the tag management system has also given them access to a lot of rich data, so they can build test models and look more fully into the attribution model that we’ve started to build.
Was attribution the prime reason for the tag management system then?
To be perfectly honest, initially it wasn’t. Initially the prime motivation was very much around the efficiency of our tag management.
But while most tag management systems deliver that efficiency, the additional benefit we found from Tagman is that it opens up our other channel technology so we can get them talking to each other and pull all that data into one place so we’re not looking at channels in silos any longer.
And that’s probably something we didn’t fully understand initially, but is now one of the key functions of the tool that we’re now starting to unlock and use to drive some of the more complex initiatives in our marketing plan.
So how did you model change specifically as a result of using a tag management system?
When we used a last-click model display was only showing as contributing to 1% of conversions, which would make you question your investment in that channel.
So we wanted to understand what actually contributed to that last-click and influenced them prior to the conversion.
This came down to two key questions: which channels were bringing customers into the purchase funnel, and which channels actually pulled customers through to conversion?
And when we actually looked at the data in that way by utilising the tag management system we saw a very different result.
So regarding display for instance, we actually saw that it contributed to 9% of our entry path to conversion, which is hugely different to our last-click model which said it only impacted 1% of sales.
Similarly, we found that display was present in 30% of mid-funnel activity, which was a real eye-opener and showed us the impact that display has in pulling customers through the funnel.
But tag management also gives you other useful data on top of that, such as which publishers or networks were having which causal relationships through the funnel.
So we could identify which publishers were grabbing a particular customer using display, which publishers were pulling that customer through the funnel and where they were pushing the customer through to convert.
It was these sorts of insights that helped us to redesign our attribution model and optimise our budget without necessarily spending any more money.
So our display budget has gone from being tactically or campaign-driven to a much more rounded, always-on model as we’re more aware of its value. As a result, sales in 2012 increased by 15% and the cost per sale decreased by 20%.