Attribution and ROI
"It’s a near certain fact that much of a marketer’s time is spent on analysing specific campaign ROI, or put another way, how much marketing spend was needed to generate a particular sale. It does seem to make sense that if a brand is employing a vast number of channels to reach a customer – PPC, social, email, display – it will want to know how to spend its limited resources, in order to optimise future campaigns. However, I am not convinced that campaign level ROI is the best metric for measuring campaign effectiveness."
The classic ROI (Return on Investment) model was developed to measure the effectiveness of the use of capital. For example, a company wants to build a new plant and it is relatively straight forward calculation to calculate the return on this investment: the revenue generated by the plant-the cost of the plant which is then divided by the cost of the plant to get a percentage. In the interest of completeness it should be noted that the calculation has to be adjusted as the revenues and costs are spread over a number of years.
Return on marketing investment is not as straight forward. First in the example above, the money used to build a plant is a capital expense and therefore does not immediately impact the P&L whereas all marketing spend is an operating expense and therefore hits the P&L. Second, measuring the output of a plant and therefor the resulting revenue is easy to measure. The revenue generated from your marketing on the other hand, is substantially harder to measure because it is difficult to attribute specific revenue to specific marketing spend.
eConsultancy recently published a blog on marketers’ views of marketing attribution in multi-channel marketing. The general view is that it is a massively convoluted issue with no good solution. Similarly, Google’s Avinash Kaushik goes into the challenges of attribution and using ROI as a metric of email effectiveness at great length in his recent blog , discussing the pros and (mostly) cons of each attribution model – I highly recommend that you read both of these articles.
It’s a near certain fact that much of a marketer’s time is spent on analysing specific campaign ROI, or put another way, how much marketing spend was needed to generate a particular sale. It does seem to make sense that if a brand is employing a vast number of channels to reach a customer – PPC, social, email, display – it will want to know how to spend its limited resources, in order to optimise future campaigns. However, I am not convinced that campaign level ROI is the best metric for measuring campaign effectiveness.
In an ideal world, all marketing would happen in a vacuum – a campaign either delivers immediately or is instantly forgotten. In this Utopia, it would be much easier to decide which marketing tactics did not work and not do that again. Attribution would also not be a problem because each campaign would stand on its own and would neither be impacted by all marketing that went before nor have any impact on campaigns to come. We know of course that this is not the case but we get tied up in knots trying to assign a portion of the revenue back to each of the multiple touches leading up to a purchase – but how far back do you go?
This approach de-humanises customers from a marketer’s standpoint, reducing them to merely a statistic. Every marketing campaign has a direct impact on a consumer. These are all brand touch points, each with their own historic effects on brand perception and future effects on conversion. On top of that, the journey that each customer takes could be different giving you almost infinite combinations of paths to analyse and measure. Given that is the case, the model of tracking campaign ROI is a fool’s errand.
So if not Campaign ROI then what?
Instead of focusing on the marketing spend needed to generate a specific sale, focus on the value that each marketing channel brought over a period of time. Very simply pick your period of measurement and look at the total revenue by customer over that period. Next group your customers by the channels through which they interact with you. The sum of the total revenue for the customers in that group gives you the portion of your total revenue driven by that group of channels. By comparing the different groups you can then unpick the value driven by each channel.
Okay, maybe not so simple but much easier than trying to determine the value of a PPC click vs. opening an email. It also better matches the revenue with the marketing spend that led up to that sale, and it makes more sense from an accounting standpoint as you begin to amortise the investment in the cumulative effect of your marketing across all of the sales for a period.
Instead of looking at all of your customer journeys and modelling the value driven by each touch point, we should try and employ – as best we can – a common sense model, where we look at the revenue driven by the customers using each channel mix and determining the value driven by each. As Kaushik says, all we can really do is try to be ‘less wrong’.