Intel - Analytics and Insights
Intel provides insights into time-series data such as clickstream, transactional or telemetry data. Intel powers the analytics and computed attributes on entities such as Customer, Products, Categories, Stores, Campaigns, Personalization Rules, etc. Time series data can be ingested into Intel from unlimited upstream data sources including transactions, data warehouses, social media, etc on a batch schedule, in real-time or in streaming modes. Conscia's Track API also captures real-time signals and pushes them into Intel for further processing.
Intel supports both DX Graph and DX Engine to provide various types of insights, for example:
- Content Analytics
- Product Analytics
- Customer Traits
- Campaign Analytics
Calculating Computed Fields
Intel can compute user profile attributes such as customer's lifetime value, lifetime orders, category affinity, etc, as well as campaign and content level analytics such as click-through rates, conversions, most viewed by segment, etc. These computed fields are calculated using the clickstream/transactional event data such as purchase transactions, views, clicks, etc.
The following table lists the various types of calculations that can be computed and used to enrich any entity.
| Metric type | Data Type | Computation Template | Parameters Required | Filter Parameters | Sample Computations |
|---|---|---|---|---|---|
| Time Since Last | Number | {Time Since} Last {Event Name} | Event Name, Time Format | Time Window, Context Field | Days since last {Purchase, Add to Cart, Add to Wish List, Visit, Page View, etc} |
| Time Since First | Number | {Time Since} First {Event Name} | Event Name, Time Format | Time Window, Context Field | Days since first {Purchase, Add to Cart, Add to Wish List, Visit, Page View, etc} |
| Count | Number | {Count of Events} of {Event Name} | Event Name | Time Window, Context Field | Number of {purchases, page views, etc} |
| Count, Grouped | Array of Key/Value Pairs | {Count of Events} of {Event Name}, grouped by {Context Field} | Event Name, Group By Context Field | Time Window, Context Field | Number of {purchases, page views, etc} by {Channel, Device, Location, Browser Type, Product Category, Brand} |
| Context from Last | String | {Context Field} for Last/First {Event Name} | Event Name, Context Field | Time Window, Context Field | {Product Category, Brand, Price, Device, Location} of Last {Product Purchased, Product Viewed, etc} |
| Context from First | String | {Context Field } for Last/First {Event Name} | Event Name, Context Field | Time Window, Context Field | {Product Category, Brand, Price, Device, Location} of First {Product Purchased, Product Viewed, etc} |
| Unique List with Counts | Array of Key/Value Pairs | {Unique List} {Context Field}{Sorted by} | Context Field, Sort Type | Time Window, Context Field | Unique List of {Channels, Device, Locations, Brands, Topics, etc} |
| Aggregation (Sum/Min/Max/Avg) | Number | {Aggregation} for {Event Name} | Aggregation Type, Event Name | Time Window, Context Field | Total Spend, Average Spend |
| Aggregation, Duration | Number | {Aggregation} Duration of {Event Name} | Event Name, Context Fields | Total Minutes Watched, Minimum Minutes Watched, Average Minutes Watched | |
| Aggregation (Sum/Min/Max/Avg), Grouped | Array of Key/Value Pairs | {Aggregation} for {Event Name} grouped by {Context Field } | Event Name, Context Fields | Time Window, Context Field | Total Revenue by Brand, Category, Device, etc |
Scheduling Computations
During onboarding, the computation schedule is established. For some use cases, it may make sense to run computations multiple times per day while in other cases, once a day may be sufficient. Please note that the total compute power required for your instance will be based on the number of compute operations and the storage requirements.
Number of Compute Operations = Number of Unique Computed Fields x Number of Times Calculated
Generating Predictive Scores
Conscia provides configurable Machine Learning models to create different types of predictive scores such as customer's propensity - how likely or unlikely people are to take an action, such as purchase, or churn. This allows our customers to create optimized offers, promotions, discounts and more based on thousands of behavioral data points in no time.
For more information on this capability, please speak with your Account Representative.
Intel - Data Flow
The following demonstrated the overall data flow starting from events received in real-time or batch to computed values and traits enriching the customer profile or any other entity being managed within Conscia.
Details on how to send event data into Intel are here.