Article:
Five questions to ask before implementing
multi-touch attribution

Call us on 0845 234 0391

9am-5pm Monday-Friday

Marie Myles, Director of Consulting, Experian Marketing Services

‘For the last three or four years digital marketers have been discussing attribution in forums and conferences but now at last it seems everyone is rolling up their sleeves and implementing attribution to varying degrees of sophistication. Indeed the buzz now seems to be around phrases like ‘beyond attribution’. This article will look at the 5 key questions any digital marketer should ask before embarking on setting up an attribution model and then ask what next? First though a brief overview of what is meant by attribution and how it has evolved over the last ten or so years.’

In the beginning there was traffic or visits and the big questions were:

1 - How many visits do I have to my website?
2 - How many sales have I made?

Then web analytics emerged and the digital marketer no longer had to rely on the ‘Webmaster’ supplying web logs for translation. It was now possible to measure unique visits and conversion and examine customer journeys. The analytics tools also provided a view of where the traffic was coming from. The digital marketer could start to measure the effectiveness of his advertising.

However as web analytics tools were designed to measure the effectiveness of websites and the behaviour of customers on those websites the journey the customer took to arrive at the website was not measured. Hence the emergence of the first attribution model the ‘last click wins’.

This is still the most common form of attribution. The problem with last click is that it favours search brand terms and, depending on the vertical, affiliates.

Other media such as display advertising, email and social are down weighted. In effect if you optimise to last click you run the risk of optimising your campaign to such an extent that although ROI improves you lose volume and future sales and/or leads. You still need to generate demand or interest and for that you need channels such as display and none-branded search. These channels all have their own measurement tools and for many marketers the first forays into ‘multi-touch’ attribution will have been through their search bid management tool.

So what is ‘multi-touch’ attribution and what are the key questions a marketer should ask before setting up an attribution model?

‘Multi-touch’ attribution is measuring the effectiveness of your media spend by modelling data and desired outcome in terms of sales or response and then optimising the channels accordingly to deliver improved performance and feeds into the customer engagement strategy.

So what are the five questions you should ask when implementing ‘multi-touch’ attribution?

1. Do I need a ‘multi-touch’ attribution model?

2. What channels are you measuring and should this include offline?

3. Do you have the tracking in place to measure all touch points?

4. Do you need real time attribution or can modelling retrospectively suffice?

5. Are the skills available for analysis of the data and implementation of the model?

So looking at these in turn...

1. The first thing to consider is do you actually need a ‘multi-touch’ attribution model? Reasons you may not might be a lack of advertising. You may only advertise through a small number of keywords on Google and last click may suffice. Another reason could be your sales cycle. If your sales cycle is relatively short, hours perhaps a day, then again last click will probably suffice. However if you have any of the following - a significant advertising budget, multiple ad sources, sales cycles are longer than a day or you have different sales cycles depending on the product - then ‘multi- touch’ attribution is a more obvious choice.

2. Once you have decided on a more advanced attribution than last click you must consider which channels you are going to measure and whether this can this include offline. Ideally you should be measuring all your digital activity and offline wherever possible. Typically offline can be tracked through unique URL’s, unique product codes or unique phone numbers. One reason you may exclude a channel from your model may be tracking.

3. Tracking is key to a successful ‘multi-touch’ attribution model. Ideally you want all activity tracked through the same source usually a container tag solution. If this is out of scope then tracking will be through existing web analytics tools and third party tools such as ad serving technology and bid management tools. If this is the route taken then the model will not be real time but applied retrospectively.

4. Real time attribution has a number of advantages not least the fact the data is not out of date. There is also the added advantage with the container tag solutions that you can de-dupe in real time. Based on your model you can decide not to fire tag requests; for examples if a cashback affiliate is the last click (goal hanging) you can choose not to fire the affiliate tag. Another advantage is that if the tracking is set up correctly and you have a sizeable paid search campaign you have a real time percentage of how your non-branded ad groups are supporting your brand terms.

5. Finally and most importantly it is one thing having all the data for the attribution model but the right skills will be needed to build the attribution model and put a rigorous testing plan in place. These skills will usually sit within the insight department of the larger companies or within external agencies or consultants.

Once you have your attribution model in place you can start to look ‘beyond attribution’. If it’s generally acknowledged that we have now moved from last click to a ‘multi-touch’ attribution modelling then the opportunity must surely be to push data back from the customer database into the model. The next logical step is to optimise not to a sale or lead but optimise to customer behaviour or life time value.

About the author:

Marie Myles
Director of Consulting
Experian Marketing Services

Marie has a wealth of client and agency experience in delivering data and insight driven actions and business benefits. She is a senior marketer with over 12 years experience managing and delivering large scale sales and marketing plans in the utilities and telecommunications sectors. This is backed by over 7 years agency and supplier side experience, ranging from a Senior Planner in a direct marketing agency to leading a consultancy team in the analytics sector. Marie backs these marketing credentials with a strong commercial background and qualifications ranging from a degree in Economics to diplomas from the CIM and IDM.

  • © 2014 Experian. All rights reserved