The wide adoption of recommendation systems is relegating search to a second priority for many customers. Today customers expect experiences that are designed with intuitive navigation and interactive engagement, something that is only made possible by platforms that use recommendations technology (or engines). In these recommendation-driven systems, ‘Search’ only comes later when a recommendation system fails to provide the information the customer is looking for.
Just as much as the customer, marketers are finding great value in also placing a higher preference in implementing recommendation models with predictive abilities over a mere search feature. As a result, leading Customer Data Platform (CDP) vendors are implementing advanced recommendation capabilities as part of their core product offering.
So, what do Predictive recommendation models in a CDP enable? These models leverage algorithms and data mining to figure out what the customer is most likely looking for. And, place it right in front of her with utmost precision possible.
Predictive recommendation provides intelligent exposure to meaningful choices that a customer translates into purchase, usage or consumption. HBR quotes that “Recommendation engines (or recommenders) force organizations to fundamentally rethink how to get greater value from their data while creating greater value for their customers.”
That brings us to understand how a CDP applies predictive recommendation models that help marketers in improving customer experience while integrating fragmented sources of data and creating a unique single customer view.
This two-part topic helps marketers understand how predictive models are used for generating content recommendations in a Customer Data Platform (CDP) in the first part. And, the second article provides a list of use cases that could be explored using predictive models deployed within a CDP.
How does a CDP provide recommendations?
Data is drawn into a CDP in different forms and formats. Content is a bucket of data that is ingested into a CDP during the data ingestion process.
Each content is tagged with what is called the ‘data labels’. Data labels are applied to every content based on the type of content each customer responds to—humorous or serious, excited about offers or brand equity, and so on.
For new customers, with little or no historical data, FirstHive’s algorithm creates a look-alike using the first-party data. Then, it creates a cold recommendation based on behavior, demographics and other parameters used to create an identity. Using this, the platform indicates recommendations about the type of content, at a given time, via the available channels that lead to high conversions.
Beyond this, the platform recommends to the marketer about which among the available channels would reap a high probability of response or conversion. The high-converting channels are further mapped to the best time for response and conversion, the optimal messaging and content for a given high-converting customer.
What can a marketer infer as insights for campaign activation?
Using predictive models, FirstHive has been able to deliver a sector-agnostic approach with content recommendations. This means the platform is capable of customizing recommendations and personalize communication for a variety of sectors such as retail and CPG, hospitality-travel-tourism, healthcare, high-tech, manufacturing, etc.
The simplest version of your FirstHive dashboard that describes recommendations would look like this.
Recommendations for each dimension would include the following:
Channel: Channel optimization is based on the content tag that works best for a segment of customers or a single customer. It recommends the best timings that suit a given customer segment.
Time: Using the best time-response recommendations, ads and marketing budgets can be optimized.
Content: Content tags are marked to each content added to a campaign. Using historical content tag performance for a given customer segment, content and personalized communication recommendations are made.
Customer: For the given campaign criteria, the platform helps to find High-Value Customer and customer segments that translate into optimized conversions.
Some other cases would include recommendations similar to the contents listed in the table below.
|Website||Country time zone||Text:image ratio||Look-alike customers|
|Social Media||Day of week||High-response messages||High-probability|
|Phone / Voice calling||Seasonal periods||A/B testing results||Customer segments|
|SMS & Mobile||Time of the day||Copy that has worked in the past|
Added intelligence to Marketers
Including human intelligence, marketers can use platform recommendations for the best ROI. These recommendations that are system-driven remain unadulterated by human bias, assumptions, and emotional cues. It is purely data-driven which makes a marketer smarter.
Should you want to gain access to similar intelligence, reach out to email@example.com