Master Data Management with Sepia Alterra

Multi-Shop and Multi-Site Content from a single source

After we have been using the Internet for more than two decades, applications to coordinate and store communication and to appointments have made particularly great progress. Examples of such applications are social networking platforms in the private area, and groupware applications for enterprise use.

In parallel to this rapidly progressing development, many businesses are not making any progress in information structuring and consolidation. Thus, in particular companies that wish to make their data available on the Internet come to realize that product information – even for simple products – are only available in different versions, in some cases in various languages and in others only in one, are incompletely specified and poorly tailored to the target market.

In addition to this, there’s the vexed issue of "Searching and finding” – rehashed over and over again under new marketing terms ... the currently used buzzwords are: “Big Data”, formerly known as “Knowledge Management” and later also “Enterprise Search”.

Poor data quality

Enhancing the quality of information for the target market often requires immense effort in a company’s product marketing and sales departments. Even if the information visible from the outside seems complete and conclusive, it is sometimes surprising to see the tortuous paths that data must take to reach potential consumers in a sufficiently high quality. A particularly annoying thing: Often, the quality of the information about a product is diametrically opposed to its own outstanding quality and the sophisticated processes used for its production.

 

Excessive use of MS Office

One of the main reasons for the stagnation of the development in this area is the fact that the handling of this information in companies is following patterns that slipped in during the 1990ies and then entrenched themselves. Very popular handling techniques include exchanging Word and Excel files without versioning, using Excel as a database, and an impenetrable mesh of MS Access / MS SQL Server databases created by internal programmers and administrators. In this context, we should consider that software companies such as notably Microsoft might have no interest in eliminating these abuses as this would contradict the paradigm to sell software licenses in as large numbers as possible; in other words: the more distributed and inconsistent data sets are managed in parallel, the better for the business.

Legal aspects

It is rather the rule than the exception that a chaotic data extraction process eventually results in an incorrect description of a product – which is not always only cosmetic, but might as well bring about serious legal issues. If a customer buys, for instance, a refrigerator with a noise level, which has been mistakenly stated to be 40 dB by the manufacturer, but later realizes that the device works at a noise level of 42 dB this has automatically effects on the return and exchange rights. At this point, it is, of course, always important to look at the total number of cases that are relevant for the seller’s turnover – and finally have a negative effect on a company’s reputation; keyword: negative word-of-mouth advertising.

Time to market

Against the background that the factor "time to market” is becoming increasingly crucial for many products, it is mandatory that in addition to product development and the optimization of the supply chain, also the processes in product marketing are optimized.
This said, we have to bear in mind that the handling of product information from a planning and coordination perspective is much more difficult than the recording and forwarding of communication fragments in social media or groupware applications. Managing master data requires a) planning, b) coordination and c) a flexible database to store and find the information.

Solution approach: Master Data Management (MDM)

2 scenarios deliver solution approaches for strategic planning: One approach considers only information that concerns product marketing and uses Product Information Management (PIM) on the software side. A broader approach gets down to the root of the problems and starts to consolidate and synchronize master data across the enterprise – which can be implemented on the software side via Master data management (MDM).

What aspects should be considered when planning and building a sustainable Master Data Management? It is generally assumed that a master data management system should be easy to configure, easy to be filled with existing data and feature a high performance for processing queries. Also desirable are interfaces to other systems that are easy to set up, preferably to an Enterprise Service Bus or for integration into a Service Cloud.

Aspect: Data acquisition

During the acquisition of data, we assume two partial aspects: Firstly, it must be easy to migrate existing data to the MDM system. A keyword in this context is data mapping. Mapping should be possible from both 2-dimensional data sources (tables) and from sources with more dimensions such as, e.g., XML trees. Secondly, a migration should be controllable via ETL tools (Pentaho or similar). Timed synchronization or "real-time" updates from other applications should be also an option.

Support features in Alterra: 

• Import Wizard

• Flexible query language for CRUD functions (Create, Update and Delete per SQL and XPath) *

• Possibility of using ETL tools (Extract Transform Load)

Aspect: Data quality and integrity

If a master data pool is maintained, it must be made sure during initial filling and during ongoing operation that the data quality and integrity are guaranteed and, above all, remain so. To ensure this, an MDM application should include functions for data cleansing. In addition, the planning of an MDM data pool should consider that one of the most effective means to avoid data inconsistencies is hierarchical inheritance.

Support features in Alterra:

• Data Cleansing Assistent*

• Query Designer

• PHP-API Java-API*, .Net-API*

Aspect: Maintenance of subsets and contexts

Another aspect that shouldn’t be underestimated is the break-up of information blocks into fine granular parts of information that can be combined with each other in different ways in different contexts (data splitting). In this way, texts usually summarized in a document can be broken down to text modules, for example.

Support features in Alterra:

• Alterra is a multidimensional data store.

• Configuration of property objects

• Configuration of text modules (XHTML formatted, non-formatted)

• Configuration o dependencies between properties

• Dynamic linking of properties with products

Aspect: Content-related requirements of the target markets

The information existing in granular text units can now be quite easily combined to longer texts with different priorities assigned to the text modules – and different information as might be relevant for each target market. This allows to adapt product descriptions to regional conditions.

Support features in Alterra:

• Country and language-specific values

• Customer-specific values

• Platform-specific values

Aspect: Technical specification

Describing master data in a differentiated way and making it efficiently searchable, retrievable and comparable can be achieved by product classification. Classification is also useful when master data should be synchronized with partners or customers. Only classifying allows you to offer customers directly comparable alternative products – or to launch the suitable successor product immediately after the discontinuation of a product. In the retail business, for example, the classification of products facilitates the performance category management that allows to analyze and plan product ranges or portfolios. To be able to technically implement this, an MDM system should allow the creation of an unlimited number of attributes as well as taxonomies of any required depth.

Support features in Alterra:

• Classification of products

• Creation of own classification systems

• Creation of own attributes

• Various preconfigured data types

• Standards such as UNSPSC, GTIN, eClass, ETIM

Aspect: Languages

Where multilingual information is necessary, it must be possible to use the information in question explicitly in the target language.
If multilingualism is not required, a standard data set must be retrieved from the data pool. As far as possible, the data pool shall be configured in a way that ensures that the same terms are never translated twice. In daily business, information is way too often translated several times and stored. The result is that whenever changes occur to the source data set – that is in the language in which the manufacturer designs its product – cannot be communicated directly through to the target language (information chain interception).

Support features in Alterra:

• Definition of any language possible

• Differentiation of translation of individual properties

• Definition of language area and country intersections

• Translation Memory system as a translation aid

 

Aspect: Data exchange with target platforms

For the data exchange with other marketing and sales-relevant systems, mainly the following should be considered:

• Shop systems

• Content management systems

• Apps (formerly "programs")

• Portals (Amazon, Mercateo, etc.)

Generally, an MDM system should provide common standards for synchronous and asynchronous data exchange. Standard technologies are XML export / import, CSV export / import, SOAP interface, web services and RFC / RPC. Ideally, an MDM system already offers standardized interfaces for common systems. For webshops, the major OpenSource systems should be covered; for CMS also the top 5 of the OpenSource universe. Apps are usually used to keep the data volume to be transferred as small as possible via differentiated SOAP or JSON calls.

One of the most important issues in this context are trustees such as Amazon, B2B eCommerce portals such as Mercateo, or syndication platforms. At this point, standardization via XML formats such as BMEcat and industry-specific classification systems are helpful. The real-time output of these formats should come as standard.

Support features in Alterra:

• Event-based real-time updates of third-party systems

• XML views in real time

• Extensions and plugins for various OpenSource systems such as Magento, Tpo3, Joomla as well as Enterprise CRM / eCommerce systems

Aspect: Formats and standards for data exchange

This leads us directly to the issue of formats for product data exchange. At this point, it is recommended to focus on the most popular standards. With respect to data exchange this is clearly XML (Extensible Markup Language) in its various versions and with its high flexibility in terms of defining and schematization (DTD, XML schemes). In addition, it is highly recommended to use UTF-8 as preferred encoding for internationalization. To specify the language, the ISO 639-1 standard should be applied, and for the country code, ISO-3166-1 is to be used. The data exchange should be running over the HTTP Internet protocol; data should be transferred compressed to reduce the network load (compression: ZIP, GIP), the same applies to graphics – they should also be exchanged via compressed formats such as PNG, JPEG of GIF.

Supported formats in Alterra:

• XML (DTD, XML schema for exchanging catalog data with UTF8)

• All common image and video formats and automatic compression / conversion (PNG, JPEG, GIF, MPEG, SWF ...)

• File compression with ZIP, GZIP

• HTTP and HTTPS as protocols for data exchange

• Use of ISO 3166-1 for countries and ISO 639-1 for languages

 

Aspect: Performance (especially when synchronizing)

When exchanging data with other systems, 3 variations are possible:

• Synchronous exchange = obtaining information in real time (queries for smaller amounts of data directly in “live” mode.

• Asynchronous exchange = exchange of information at time intervals (queries as batch process for large data sets).

• Asynchronous exchange via event triggering (usually in combination with ESB = Enterprise Service Bus or Service Clouds).

 

Support features in Alterra:

• High-performance MDM data pool with up to 10 million objects (maximum access time 0.2 seconds, tested in the benchmark)

• XML output in near real time

• Definition and creation of snapshots*

• Definition of cron jobs for batch transfer

• Event triggering

Aspect: Usability

Despite all the fine technical features, an easy and especially convenient handling of the user interface of the MDM system is of utmost importance. The systems currently on the market usually offer a flexible configuration and have elaborately designed programming interfaces; however, in daily business, any manual intervention or manual data capture must be as easy and smoothly as possible. Borrowing from well-known application patterns such as editing mass data in the grid (as in Excel) come as standard with newly designed web applications such as Alterra.

Figure: Flexible web-based editing of tables with Excel functions (here: drag cells" in Alterra 2.8 with FlexGrid plugin)
 

Outlook

With the go-live of Alterra 3 that is scheduled for 2013, customers with high demands in terms of flexibility, configurability, performance and standardization for MDM projects are addressed. Alterra 3 is designed as enterprise software for medium-sized and large companies. The required JAVA and .NET connectors as well as the scripting support via PHP are integral parts of the new release.

Alterra 3 in numbers:

• Lines of code (up to 1/2013): 0.4 million

• Support: 3rd-Level Enterprise Support

• Publication: 2013

Learn more about Master Data Management and Product Information Management...