Data-as-a-Service: Solution Architecture Part II

Customer Capabilities: Part II on DaaS Solution Architecture

"Firms that embrace big data concepts, open data, and adopt new adaptive intelligence approaches are creating next generation smart systems that overcome limitations and create disruptive business innovations.”
- Forrester Research: "Top Technology Trends for 2014 & Beyond"

The 2nd layer of a robust commercial Data-as-a-Service (DaaS) solution architecture comprises a number of key data management capabilities that are bringing to life the powerful value of DaaS for enterprise customers - deep, accurate data within the key applications used daily, driving deeper insights and competitive agility. As I noted in my earlier piece on the arrival of DaaS as the next major enterprise Cloud category, these features utilize many of the best well-established Master Data Management (MDM) practices that have, for many decades, advanced reconciling, classifying and enhancing the quality of data sets company-wide.

DaaS Platform Layer II - Customer Features

DaaS data management services build upon and tightly integrate with the DaaS Layer I Feature Set for External Data Providers that I outlined in a previous post. Flexible data provider on-boarding & update tools, configuration & mapping for a consistent data dictionary and an advanced matching engine are all part of this "Layer I".  These customer-focused Layer II capabilities include but are not limited to the following:

• Data Standardization
  To achieve this essential data consistency & trustworthiness, each standardization service needs to specifically tuned to the data domain it's addressing - name, address (and its many sub-elements) and financial data, to name just a few.  This attunement requires specific referenced dictionaries, as well as robust fuzzy matching & data cleaning.  Done well, this lays the groundwork for then achieving higher match rates between the existing and DaaS-provided data sources.
  
  
• Search & User Interface Framework(s)
  Search, which is invoked by the users tied to the requesting marketing, sales, service & analytics applications, must be optimized for both the canonical data category taxonomy for all hosted data source providers I noted for DaaS Layer I for Data Providers, as well as retrieving the data values within.  And to serve key business personas well, the DaaS UI must provide intuitive, rich faceted filtering controls to provide back exactly the targeted data sets these time pressed users are seeking.
  
  
• Match, Append & Clean - The True Power of DaaS, Unchained
  These are the customer services that invoke the robust matching engine (mix of probabilistic controls & deterministic algorithms) that lies within the Layer I of a commercial DaaS offering.
  
  It's critical that these services provide the interfaces & controls which ensure that consuming application users adhere strictly to their respective Data Governance policies & rules:
  What are the matching confidence level thresholds (Example: 1-10 scale) at which cleaning updates of existing customer data with the invoked DaaS sourced data is allowed?  What about allowing the appending of these existing sets for enrichment purposes? These services need to provide an intuitive, "staged" review of results for these actions before they are executed, again in adherence to the Data Governance controls.
  
  In addition to these interface controls here, DaaS should have well-defined APIs whose structured calls & retrievals that mirror these controls, serving larger enterprise clients who prefer a more tightly integrated external data source consumption model.
  
  
• Duplicate Awareness Management
  Many of the existing customer data sets include full or partial duplicates that the DaaS matching engine (with fuzzy matching) can detect & flag here as results, pre-clean update or append action. 
  Resolving these duplicates is tricky, and I'm reluctant to say that the DaaS offering should perform this task - I welcome some broader discussion & input on this front. Perhaps DaaS should be closely complemented here by other MDM tools for true existing duplication management (handling resolution on "primary" record here, as well as any data record dependencies, before executing DaaS matching actions. 
  
  
• Usage Visibility (and Notifications)
  Tied closely to the defined subscription usage model (high level subscription model; per record) for DaaS, each customer must be able to "dashboard" monitor and track their usage of DaaS data record services.  Notification controls, as subscription are per-record consumption milestones are approaching, should be provided here. These services are critical for managing customer awareness and consumption satisfaction closely, supporting and driving recurring subscription-type consumption.

DaaS Solution Architecture Recap

In two parts I've now covered, admittedly at a very high level, the key solution architecture capabilities that a strong commercial multi-external sourced data master offering must have to deliver on DaaS, serving both the vast ecosystem of data providers as well as consuming enterprise application customers. This ecosystem of high value data providers runs the gamut, from D&B global account data to consumer segment demographics (Example: Acxiom) to even more real-time newsfeeds on companies ("New product line introduction announce") or contacts ("Individual X just promoted to new executive role").

The bottom line - business managers in marketing, sales & service today want the right, rich data in context, in real-time for true business agility and personalized customer and partner experience delivery.  Look for several commercial DaaS vendors (large and small) to vie in the coming months to deliver the next generation of intelligent, data-driven enterprise applications, particularly those for major verticals such as Retail, CPG, Financial Services, Health Care and industrial areas such as Oil & Gas.

Upcoming Pieces

1. DaaS MDM - Essential for PoweringToday's Analytics
2. Managing Data Privacy in the Land of DaaS