AEM Sites (+ Guides) Metadata: How to taxonomize evolving products?

[Eric_Amble]

Hi everybody,

I'm an AEM (Sites and Guides, cloud version) user with a group of companies, with a long history, developing and supporting decades worth of consumer products.

I've taken on the responsibility of coming up with a way of classifying all our products, assets, and deliverables throughout history for these companies. The products in particular can be classified according to the parent-child relationships you might find in a DAM's folder structure, such as:

• Brand
  • Family
    • Group
      • Model

...but there's one other aspect I'm having trouble with: these products can vary over their lifespans. For example, a product with the same name might have a modification or internal design improvement that affects its technical documentation, and the service manuals etc. for such products would all need to be available on Sites, whether new or old, so that servicers can search for a product based on its serial number, and get the appropriate documentation, for any vintage of product.

In the past, we've dealt with this evolution of products according to serial number breaks. For example, products with a serial number prior to 123456 have one coil with a certain resistance value, and products whose serials start with 123456 have an improved coil with a different value. Our convention is to describe these serial number breaks as "PTS <serial number>" and "SWS <serial number>"  

Of course, many of these serial number breaks overlap within individual products. A product with a certain serial number could contain a variety of parts whose parameters or procedures are punctuated by serial number breaks.

So, I would like to ask: in AEM, what's the best way to describe/locate content that can vary according to serial number? I was thinking there might be a way to use a piece of metadata (a field or a sort of metadata tag with a variable in it) in which we could enter "PTS 123456" or "SWS 123456", and then an AEM DAM search (or possibly in our case, an API call from elsewhere) could search on a serial number value and do a bit of math, such as "show me the items with these tags, and whose serial number metadata is less than "PTS <serial number>" and greater than "SWS <serial number>". I don't know whether that would take some kind of development effort on our part.

Has anyone had any experience with this kind of twist on very big taxonomies with long histories, and has anyone settled on a best practice for it?

Thanks, I look forward to hearing what you think!

Eric

[Harwinder-singh]

@Eric_Amble  A couple of questions before I can comment on the design.

• Are you storing the actual product data in AEM or somewhere externally in a PIM ?
• What are the business use cases that you are trying to solve via taxonomy? beyond the one that you have highlighted.
• How big is your current product inventory ?
• Are you using any search solution (SOLR, Algolia etc) for your search requirements ?
• What about product related documents (IFUs, Datasheets etc)? Where are they stored ?

If the requirement is just to have a way to classify and identify Products based on a specific serial number range, you can use AEM tags and apply appropriate tag names (PTS, SWS as you mentioned) and if this data is stored in AEM DAM, then also just have these products stored in different content trees and apply the appropriate serial number type Tag at the top of the content tree. 

[Eric_Amble]

Hi, thanks for looking into this with me.

The data is not stored in PIMs that we can use for this. There have been incomplete implementations of 'lightweight PIMs', but they are not comprehensive enough to help us with these problems.

The use cases are the usual stuff: we want to harmonize all our assets (at least, communications-type assets) in a CCMS that our authors, from many departments across the companies, can onboard onto and use. To do that, we need this taxonomy.

I have no idea what those search things are, so I would say not.

Product-related documents are stored in various places: some are stored in another CMS, some are stored in AEM, some are stored on network drives, and it's kind of the Wild West... like a lot of companies.
I want to bring all this disparate content under one roof, and manage it with a single system--the taxonomy is necessary for this, but I also need to separate it by serial number break.
With three companies and thousands of products (past, present, and future) to harmonize, could you realistically set up, say, a thousand tags reading "SWS 123456", "SWS 134578", "SWS 258740" etc. etc.? There must be a simpler way of dealing with this. It wouldn't be some kind of extra metadata field or customizable tag containing a variable value? Someone must have thought of this already.

[DivrajSingh]

@Eric_Amble : looking at the problem statement this definitely needs deep diving into the aspects of the products which needs to be:

• classified for end users accessing the products
• represented for showing details of parts and products
• searched on different criteria (you mentioned one use case where a range based on serial number can be an option)
• shows with varied details of parts based on old series vs new series
• stored based on access permissions or category of products (you gave a sample folder structure)
• strategized for its storage, persistence and archival (to keep the unwanted assets out of search criteria)

However, on a high level understanding the problem statement you can definitely achieve the requirement by breaking down such metadata into two parts:

1. A tag metadata that identifies the product series (PTS, SWS)
2. And a metadata field that defines the serial number 

Where:

• The product series (tag) can be auto assigned based on folder structure (if that is a clear categorization) using metadata profiles
• And the metadata field (for serial number) can be a multi-field as I can imagine the parts can be applicable for multiple product lines

Search based on tag+metadata can work to meet the scenarios you mentioned. And as @Harwinder-singh indicated - the volume of content to search on can be handled with indexing and capabilities of the search engine (SOLR, Lucene etc.)

An opinion on the solution approach you take is also dependent on:

1. Effort to setup metadata: dependent on how is the information available to populate the metadata to the assets - which are already in system - using bulk update or possibility to manually assign the metadata
2. Operational complexity: what are the different ways in which this metadata can/will be populated/updated; by users or by scripts

Some useful references:

• Metadata management best practices
• AEM Assets Search best practices

Please reach out to your Adobe team if you want to discuss the use case and gain insights into above information

[kautuk_sahni]

@Eric_Amble Did you find the suggestions helpful? Please let us know if you require more information. Otherwise, please mark the answer as correct for posterity. If you've discovered a solution yourself, we would appreciate it if you could share it with the community. Thank you!