Attribution, marketing attribution management, sales attribution, attribution analysis. The attribution jargon list goes on.
But despite this confusion, getting attribution modeling right could be the difference between focusing your team’s attention in the wrong direction, or achieving sophisticated, data-driven marketing planning.
We’ve identified seven key marketing attribution models below, and explained the pros and cons for you v-e-r-y s-i-m-p-l-y, so you can be more clever. Or is it cleverer?
Anyway, here they are so you can consider which one is right for you. And if you’re looking for even more attribution fun you can read our Attribution Toolkit here. For FREE.
1. Last touch
Last touch gives 100% attribution to the final touchpoint before a conversion purchase. Pretty simple.
It’s dead easy to set up and can provide quick, actionable insights. In analytics, this is simply a case of setting up a conversion goal, then checking the channels used to get there.
It’s misleading; it doesn’t take into account any activity that drove awareness or previous customer interaction that may have helped inform their decision.
2. First touch
Yes, no s*it Sherlock, ‘first touch’ attributes 100% of success to the first time the customer comes into contact with the brand or product.
Again, it’s simple to set up and to interpret. First touch works mainly on the concept of awareness, so regardless of the journey to buy, the assumption is that the first channel was the one that got them interested. This is particularly useful for monitoring the impact of ‘awareness activity’, like PR.
Like last touch, first touch is pretty one-track minded. It ignores the complexity of many buyer journeys, attributing 100% ROI to one single interaction.
With the ‘linear’ model, every step of the buyer journey is equally responsible for the sale. Got it?
It’s still simple, yet slightly more sophisticated than first and last touch attribution, recognizing that the buyer journey is typically more complex than a single interaction.
Results can be skewed. If every step is weighted equally, yet a particular channel is visited more than once, it will be credited more than once (e.g. if PPC results in a single visit, yet the visitor arrived twice via organic search, organic would receive double the credit).
It also ignores objectives. If your aim is to drive sales for example, social media and PR would be given equal credit, despite the fact their role focuses more on building awareness.
The positional model recognizes the fact that the buyer journey is rarely straightforward and that effort must go into both acquiring, and converting the customer.
Combining elements of first touch, last touch and linear models, it commonly attributes the most weight to the beginning and end of the journey, with every step in between credited at a lower level. Over time, as the buyer journey becomes clearer, the positional model can be customized to give heavier weighting to other stages.
It’s particularly useful for multi-faceted campaigns; for example, brand and product marketing, for which first touch and purchase are equally important, but the steps in between are critical for conversion.
While taking the entire customer journey into account, it doesn’t reveal timescales. The first touch may take place many months before conversion, over which time its value would diminish, yet it would still be valued more highly than some of the steps in the middle, which could actually be more responsible for pushing the customer to purchase.
5. Time decay
This sounds quite disgusting, but it’s not. The ‘time decay’ model presumes that the closer a touchpoint is to purchase, the more credit it should be given for conversion.
Unlike the positional model, time is the key factor here– the theory being that as the buyer journey accelerates towards the end, the more critical each channel is in persuading the customer to convert.
Particularly effective in understanding which tactics lead to the biggest step changes in the path to purchase.
Specifically, it helps in monitoring the impact of calls to action, as well as the effectiveness of different tactics in expediting a sale.
With this model, there’s a risk the touchpoints most recent to the purchase are given greater credit for a sale, when in fact, the decision was made much earlier in the buying cycle, during the research phase. As a result, it may not be suitable for markets with a longer lead-time.
A customized model allows you to take the most relevant bits of standard attribution models and set your own rules. It’s ideally suited to multi-faceted campaigns and complex buyer journeys.
It can be more relevant to your market, your product and your typical buying cycle. Time decay, channel-spend, and the weighting of individual touchpoints can all be adjusted according to your own reality, and objectives.
It can take time to set up and adjust to your individual campaigns. It also requires a high level of human understanding and analysis to get it right.
Algorithmic modeling is the most accurate method of attribution. Unlike rules-based approaches, it’s entirely data-driven, providing a comprehensive understanding of the impact of every interaction on the end goal.
Implemented correctly, it offers a much clearer perspective on consumer behavior and enables decisions to be made based on verified information.
Algorithmic modeling removes guesswork and gut instinct from marketing decision-making.
Success with algorithmic modeling depends on the richness of the data and the quality of analysis of the results.
In short, if you aren’t ingesting all of your data into the same platform, your results will be flawed.
For more tips for getting started with Attribution Modelling, download the Attribution Toolkit free and get with the program.