How Auto Brands Can Predict Campaign Success with AI Tech

By marrying user and campaign data, marketers are able to build a predictive model that will anticipate success of future campaigns.

Today, marketers generate engagement as customers at several stages of the purchase path consume content. This allows them to plan and place sequenced content from brand discovery to purchase, by tracking each customer’s content consumption.

Through personalization; the ability to deliver more relevant content and experiences based on a real-time understanding of an individual’s needs, preferences, and actions, marketing teams will gain a deeper understanding of what a successful campaign looks like over time.

Leveraging Watson's NLP and Machine Learning Capabilities to Predict Success - Faster, Better, Smarter

Watson does not aggregate data; it understands how to process and analyze data and undergoes general learning for more accurate predictive analytics in the future. Therefore, your data will not be accessible to Watson users in the future. Opentopic uses Watson to analyze specific parts of text, images, and videos - Watson does not store any data.

Using Natural Language Processing and cognitive APIs like Concept Expansion, Keyword Extraction, Entity Identification, and Sentiment Analysis, Opentopic will analyze marketing emails to expose cognitive elements. Such analysis will also use machine learning and models trained with the data prepared in initial step.

• After gathering a small batch of results, they are used to retrain models and calculate validation scores, which improves processing of next emails, filling up the review queue.

• This review/labelling process is seamlessly integrated in a standard results/analysis workflow, making it as effortless as possible for all involved users.
• Periodic retraining of models constantly increases number of correct annotations and predictions, adapting the system to possible changes.
• Review input from the client is not required, but when the client sees incorrect result he/she may quickly mark it – such marks makes the system adapt more quickly to specific data.

Example:

Based on a 360 degree view of our consumer segments (geo-demographics + psychoanalytics), we can predict that 50% of 65 year-old males in Arlington, VA will purchase a purple convertible within the $100-200K price range within the next 12 months after receiving a series of marketing emails that feature rainy images and talk about retirement. Additionally, we can reduce missed service visits by 10% by scheduling an aggressive cadence of service reminders to this consumer segment post-purchase.

Check out how our customers increased lead generation by 200% in 6 months....