Predictive Analytics are Ready for Early Adopters: Super Cool Sneaks from the Adobe Summit

Adobe announced their embedded forecasting and predictive models into Adobe Social, Discover and SiteCatalyst at their Annual Summit today.

I blogged about predictive analytics in my annual prediction for marketing trends entry for 2012, where I indicated that this would be the year for the early adopters to begin bringing these capabilities to digital marketing.



It’s here.



There are too many instances to mention, and many more that I expect are coming as the Product Development Teams begin to insert the Efficient Frontier (EF) models underneath the platform as underpinnings to nearly all of the products.  EF, in case you aren’t familiar with them, used Nobel-prize-winning Henry Markowtiz models called the Modern Portfolio Theory (MPT) to apply the models that help financial engineers balance risk and returns on the stock market (which is essentially an auction) to the similar problem of risk (as defined by volatility and conversion metrics) and the price of keywords, which again is essentially an auction.  They’ve been around for over about a decade, and employ seven PhDs in the field.  I attended their workshops, and began to assess how these models will be embedded into the underpinnings of the Online Marketing Platform from Adobe.



So here are three things that impressed me, and keep in mind that since the EF acquisition only happened six weeks ago, none of the models have been incorporated into the platform at this time.



The Adobe SiteCatalyst product will be releasing The Navigator feature, there are new predictive analytics capabilities in the Discover 3.0 product and Abode Social shines the headlights around the corner to predict sentiment and help marketers more carefully use words that will increase engagement.

Navigator feature in SiteCatalyst runs a model against all KPIs and metrics and identifies all trends that are outside their statistical norms based on previous history.  Hit the button again on a given anomaly, and the system performs a very deep dive to assess all the factors that had high correlations relative to the variance.  See the trending variance for forecasted and expected results on any KPI, and see where the actual KPIs occurred.  Make your models smarter and tighten the variances.  This will save hours of trying to figure out why we had a 40% increase in sales on Tuesday, because marketers can avoid having to dive deep into Search, SEO, Display, email and affiliate campaigns.  The system does that for you.


Discover 3 allows for super intense visualizations of cross-session browsing and path analysis.


Adobe Social – I have heard of PhD candidates that study “Mathematical Grammar”.  I got to see that in action where in the platform a marketer can enter their tweet and instantly, as they write, three bars on the right appear indicating the Sentiment,  Interactions and half-life of the tweet.  The amazing part is that the words in the tweet change colors (either red or green) to indicate the strength of the sentiment and its associated direction of sentiment; big red for bad and big green for good.  Then the magic happens.  Adobe Social offers suggestions using a synonym algorithm.  Change “Beast” to “Creature” and see that the sentiment goes from 7 out of 10 to 8, the interactions are forecasted to grow from 30 to 40 and the half-life of the tweet is extended another 10 hours.

Leave a Comment

Your email address will not be published.