Most of us have heard the old John Wanamaker quote, “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.” It’s been over a hundred years since those words were first uttered, yet for some reason the advertising industry still struggles with the same familiar challenge. For the past five years, Media Two has been deeply involved in the attribution game – fighting to define the true merits of marketing that drive sales for our clients. One thing I’ve learned through this adventure is that in most cases the data tends to over-complicate the circumstances, and it becomes easy to lose track of basic concepts like reach and frequency. With this in mind, I’d like to set the record straight with a high level overview of what attribution modeling is and is not, and why it’s a must-have in your analytics toolbox. First, I’ll address why it’s a must-have. Simply put, see the quote above. If a tool exists to fix this century-old problem, a marketer can’t afford not to have it – especially if it pays for itself in dividends. Now that we’ve covered the easy part, let’s jump into the meat and potatoes. So what is attribution modeling? It is:However, it is not:In order to fully understand attribution, you have to start with the concept of a conversion funnel – beginning with awareness, and leading to interest, consideration and ultimately purchase. In reality, every marketing touchpoint could fall somewhere within that funnel, and some have a greater impact at certain points than others. For this reason, we segment analysis into four basic areas within the funnel:Now let’s take this a step further. Realistically you can analyze every touchpoint within those areas, but you still need to define how to assign each a value. Depending on a client’s product or services, some models will fit better than others. In most cases it’s recommended that the greatest emphasis be placed on the Bayesian model – a learning-based model that essentially shows statistical probability. However, in their basic forms, there are six major models or methods: If you’ve made it this far down the page, that’s a wrap on Attribution 101. You’ve passed. No final exam required. Over the coming weeks, I’ll attempt to share my thoughts on some of the other nuances around attribution, the challenges and the results it can yield. If you can’t tell already, this is a passion and a driving force behind our services, but it’s also a complex and ever changing subject. Let us know if you’d like to learn more about how attribution modeling could work for you.
- A set of rules that determines how credit for conversions is assigned to various marketing touchpoints.
- Data modeling across a range of sources – paid, owned and earned media – not just advertising.
- The centralization of data allowing for analysis of conversion lag times and audience segmentation.
- A data set that tracks a single individual through the conversion process.
- A substitute for other tools such as Google Analytics.
- Origination – Representative of reach, this is the very top of the funnel, and the first thing with which a customer sees or engages.
- Bulk Frequency – Also referred to as the roster, these are the values falling within the middle of a conversion funnel.
- Sales Assist – This is essentially the next-to-last touch in the funnel.
- Converter – The last touch before a purchase is made. Often we see converters being either organic or brand search. For this reason, sales assists become important to segment from roster values, as they are an indication of what’s driving someone to the actual converting sources.