Hierarchical Completeness within Product Information Management

Some years ago I wrote a blog post called Hierarchical Completeness. This post also had some excellent comments and David Loshin made a good follow up post called Hierarchy Data Completeness and Semantic Convergence.

HierarchyThe importance of hierarchical completeness, not at least within Product Information Management (PIM), has become close to me again.

It is a numbers game. Often having an advanced PIM solution on board is based by the fact that you have many products to manage. Too many products for a single data steward to control. Add to that today’s challenges of doing multi-channel business and tomorrows challenges of embracing social media engagement. This means a lot more attributes and digital assets per product and perhaps more products to manage as told in the post called Social PIM.

All products aren’t equal. The one size fits all term doesn’t apply to selling shoes or any other range of products. The attributes and assets needed differ per product categorization and so does the performance measures and expectations for each product.

Bookmark and Share

5 thoughts on “Hierarchical Completeness within Product Information Management

  1. John Owens 30th September 2014 / 06:47

    Hi Henrik

    One thing to make sure of with any Product MDN solution is that it does NOT represent products in hierarchical form. This is because as hierarchies are a primitive and archaic structure and create a huge amount of duplication; enough to can sink an enterprise.

    All flexible and effective product structures are networks, with each product element capable of being part of many other product elements. When it comes to analytics, network structures also allow each element to be made part of any number of ‘analytical’ hierarchies, allowing the data to be ‘sliced and diced’ to cater for all sorts of ‘what if’ analysis. Hierarchies, on the other hand, are a major barrier to this.

    It is product experts in the business, rather than data stewards, that ought to know and manage product data. I think that data stewards ought to be banned from all enterprises who are really serious about data and information quality. Data and information quality is only real and sustainable when it occurs as a result of business experts doing their day to day business activities.

    Kind regards

    • Henrik Liliendahl Sørensen 30th September 2014 / 08:01

      Good points John. I guess many people, like me right now, have to balance between how the perfect world would look like and the harsh realities of the current state of things. We must definitely change the tasks of data stewards from filling in the blanks to supporting experts in doing so. Hierarchies, and relational databases, are not a true picture of the real world and we await new feasible technologies to enter the enterprise systems landscapes.

      • John Owens 30th September 2014 / 23:23

        Hi Henrik

        Hierarchies are definitely a limiting structure. However, there is no commercial structure that cannot be effectively represented in a relational database. There is, sadly, a severe shortage of people who can do high quality logical data modelling in order to enable these databases to be built. The technologies are already here; it is the skills that are missing.

  2. Kalle 30th September 2014 / 13:52

    Every attribute value that exists more than once is actually a classification element (can be used in reporting/ views etc.) so I also don’t think that hierarchies are the best way to operate here. BUT because a shoe is not a tire, there should at least be somekind of dynamic attribute, product group or something, that determines the attributes that an item has.

  3. John Owens 30th September 2014 / 23:38

    @Kalle In general it would be relationships to domain entities that would be used to categorise products for performance analysis. A good product management system would enable the structure (attributes and relationships) for different different types of product to be dynamically defined. This is where the power of the Logical Data Model comes into play when designing and building these systems. If they are not properly modelled they will not properly designed and built.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s