Adobe’s Digital Marketing solutions are now opening doors to predictive marketing optimizations that were previously only available to huge companies that spend millions of dollars constructing complex economic models to explain the reach of their marketing spend.  These large models typically include surveys, studies, panels, and hosts of other expensive outside resources to deliver statistical optimization to your marketing budget that will maximize your return on spend.

With Adobe’s recent acquisition of Efficient Frontier and recent advancements in predictive consulting, SiteCatalyst customers are able to unleash the full potential of their web analytics data in ways that have not been possible before.

One particular customer Adobe Consulting recently engaged with wanted to determine how to best allocate their digital marketing budget based on several years of historical performance.  Their digital marketing channels included SEO, SEM, affiliates, media advertising, social media, network partners, and email campaigns.  Because the performance of these channels had been thoroughly recorded across several years, Adobe Consulting was able to use this data to form statistical performance models around each of these marketing channels.

Accomplishing this analysis required several important steps:

1.  Building the right attribution model that incorporates cross-channel interactions

Attribution can be a very challenging problem in its own right.  Part of the problem that companies face (including our customer) is that the total online revenue is not usually completely due to digital sources, so attributing all online revenue to digital sources can be a big mistake that will lead to overestimation in a company’s return on spend.

To tackle this problem, Adobe Consulting Services created a special dataset based on SiteCatalyst data that combined several years of user interactions across each campaign channel with the total revenue that each user generated in the same time period.  Using this special dataset and several statistical models, we were able to assign the measured revenue to 3 categories: revenue that was derived from non-digital sources, revenue that was derived from non-paid “earned” digital sources, and revenue that was derived from paid digital sources.

2.  Assigning marketing spend amounts to each channel and campaign

This step might seem the most obvious, but it is actually quite challenging.  Many organizations do not have a central location to look up the marketing spend on each campaign channel, so this information has to be hunted down from various groups across the company.  Additionally, some marketing channels do not have an immediately apparent cost such as certain social campaigns or SEO.  With help and additional data from our customer, we were able to finally assign a cost to each channel or campaign in a way that the customer felt was most logical.

3.  Building meaningful models to characterize what each dollar spent means to each marketing channel

After each channel has been assigned a cost and return, we built appropriate models for each channel that map spend within a channel to a certain return based on each channel’s individual campaign performance, search keyword performance, or ad impression performance.  These models describe the theoretical return of any given campaign spend.

4.  Combining all of these models and forming an optimum marketing blend

Finally, once each channel has been given an optimal performance model, a simulation is run that will locate the optimal media mix to maximize an aggregate marketing return using an overall budget that reaches across all of the marketing channels together.

After the entire process had been completed for our customer, the model found that there was a larger opportunity in SEM than they had previously thought.  We also found that media advertisements were being weighted too heavily and had a much smaller ROI than the customer initially believed.  This was incredibly useful information to our customer who had already suspected that this was the case but just needed some solid analysis to validate his claims to upper management.  Most importantly, Adobe Consulting found that by reallocating the marketing budget, our customer was able to make an additional $600k in revenue that would have been left on the table otherwise.

Using SiteCatalyst to construct media mix models is an awesome way to take full advantage of the massive amount of data that you may already be collecting.  The result is not (as Jared Lees has already pointed out) a magical output conjured from a crystal ball, but more of a guiding compass to inform and direct a digital marketer in how improvements can likely be made.

If you’re interested in building your own media mix model with your SiteCatalyst data, Adobe consulting can help you set up, create, and maximize your marketing spend without having to rely on a multi-million dollar product.  Contact your Adobe sales rep or account manager to learn more!

0 comments