Everyone and no-one most likely. Everyone uses, or rather abuses attribution to their own gain. Quite naturally of course.
Did this ad work? How much money did I make from the money I spent?
These are questions marketers ask and hear often. Most people just go with the flow and believe what the industry says. But exactly what this is can be a difficult topic.
One of the benefits of digital marketing is that it is measurable down to the penny and you can work out exactly what your return on investment is. But with the amount of data there is out there to measure it’s easy to get muddled and this is where attribution seems to be at the moment – there is a lot of grey and a lot of shade, particularly of grey.
The common digital channels that lots of people use, such as Adwords, display, social and analytics could each claim a particular conversion. This is great, but what if they are all claiming the same conversion: which one was responsible? Or rather which one was ultimately responsible, even initially responsible? Unless the question you ask yourself is pinpoint specific, the answer could raise more questions.
The Facebook default for attribution is a one day view or 28 day click (i.e. if you saw an ad and converted that day, or clicked on an ad and then went onto convert within 28 days, Facebook claims this as down to them) – this can be changed of course but consider this:
This person actually heard about a site from a mate, then clicked on a social ad they saw followed with a pay-per-click ad, who then saw a prospecting display ad then went back via organic. All this within a week.
Chances are that Adwords, FB and the prospector all claimed the conversion as theirs, and yet got logged within analytics as an organic traffic conversion.
All this could also be logged as assisted conversions within Google Analytics but then what happens if it was conducted on multiple devices? The attribution becomes disjointed and riddled with gaps.
Complicated and that’s only the tip of the iceberg…literally… you only see 10% of an iceberg with the rest being underwater, also each iceberg is unique, furthermore it changes whichever angle you look at it. So you have to be more specific, take an angle and ask yourself “What is it that I am trying to attribute?”
To answer, you have to start with Tracking, ensure there is a process and decide what the attribution model you are working to is. This data, in theory, can then be used to steer your strategy and investment.
There are a few common models all of which provide differing viewpoints – the key here is deciding on that angle.
Last click attribution may seem robust (e.g. that ad made that sale etc.) but it’s a blinkered view, as it doesn’t take into account any steps down the sales funnel or other influences.
First click attribution also seems a good one (e.g. the one that first got you on the path to convert)
Linear attribution takes all points along the journey to conversion and equally distributes credit, but doesn’t take into account an important one – that which tipped the balance towards the conversion.
Position based attribution decides the most important places along the path and attributes towards them. This seems to me to be too arbitrary on the whole though.
Time Decay Attribution for me is a good one: an ad that was viewed 21 days ago couldn’t possibly have resulted in an associated conversion, but one which was viewed 10 minutes before a conversion seems to be too much of a coincidence not to have been the cause. But again this doesn’t give a full picture either, it only gives higher probability to cause of conversion.
Which ones you use, how many you look at to inform yourself should depend on what your objectives for the channels you use are and what the specific questions you want answered are. Did this get me noticed (awareness), did this ad drive interest? Did this ad make a sale? Etc.
There is no one size fits all, and in reality all the data should be manipulated to reduce assumption and duplication.
Any attribution model should be interpreted as “degrees of influencing” albeit to vastly varying degrees and molded into a bespoke plausible attribution model.
If it’s plausible, then your confidence will be greater. The greater the confidence, and in statistical lingo the greater chance of accuracy – meaning you are more likely to steer your strategy to success.
A Plausible Attribution Model should be made up of various scenarios and is about degrees of influence. In my view, a necessary complication.
The answer to the CEO’s question of “Did that poster work?” can still be YES, but their possible assumption that digital can answer exactly what sale was driven by exactly what channel can still currently only be partly and truthfully answered in a multi-channel campaign.
For a chat to help you determine what questions you want answered and therefore what your plausible attribution model is get in touch now or visit our Paid Media page for more information.