You are here: Home » Special Features » Finding Signs Of Life In DAM: The Role Of Integrated Digital Asset Supply Chains

Finding Signs Of Life In DAM: The Role Of Integrated Digital Asset Supply Chains

This feature article has been contributed by Ralph Windsor, editor of DAM News and is the first article in a series of items with the title Finding Signs Of Life In DAM.

 

In the first part of this series, I attempted to diagnose the reason why DAM has gained less traction than other enterprise software markets and proposed that the key issue was the lack of relevant metadata which made it harder to isolate digital assets than it needs to be to permit the productivity benefits of DAM to be realised.  In the remaining articles, I will consider some possible solutions.

Understanding The Nature Of The Problem

At the heart of the problem of generating relevant metadata for digital assets is the disconnect between the digital asset record in a DAM solution and the available sources of metadata.  Metadata provides context to digital assets; the more there is, the easier they are to find.

At present, too much is expected of the users cataloguing digital assets in terms of supplying relevant metadata to achieve this goal.  There are three ways this could be improved:

  • Digital Asset Supply Chains
  • Interoperability
  • User Education

A digital asset supply chain means being able to make explicit the implicit connections between the sources and destinations of assets: in other words, where they came from and where they are going.  If there is greater awareness that digital assets are not just disconnected entities, but an appreciation that they will need to get ingested into DAM solutions (and used afterwards) then there is a greater opportunity to acquire contextual information about them which can be used later for metadata-related purposes.

DAM interoperability is the nuts and bolts of enabling integrated digital asset supply chains.  It means the conventions and standards which simplify the process of transferring assets from one node to another.  Interoperability often gets thought about in terms of integrations between separate DAM solutions and although that might be a requirement sometimes, a more common usage scenario is to allow a third party application or data source to send or receive assets with a degree of confidence that they will be presented in a standardised manner.

As described in the case study I presented in the first part of this series, where DAM users have prior knowledge of metadata concepts (e.g. because they have a job that involves it like an archivist or digital asset manager) then the probability of relevant metadata being applied increases because they understand why it is important.  Their training (plus professional experience) encourages them to think like someone who might need to search for an asset.  While it is too much to expect the same of all DAM users, education and training must play a role as well.

A fourth area is Machine Learning and Artificial Intelligence.  I do not believe the results of the current solutions are satisfactory, at present, but I will acknowledge they do have some potential and there are some developments in AI which are at a more advanced stage than others.  An article to address this topic will be included as part of this series as well, but in this piece, I will consider integrated digital asset supply chains.

Why Digital Asset Supply Chains?

Concepts like value chains (which are to all intents and purposes, the same thing) have been discussed on DAM News before but a more recent piece is What is the Digital Supply Chain? by Mark Leslie:

Digital Asset Management sounds like it’s all about technology and software, but I believe it’s more about getting work done. You’re familiar with our supply chain, right? How we source our goods and get them out to customers? That is our PHYSICAL supply chain. DAM deals with virtual goods – files, documents, assets – which represent our company’s products digitally. This is our DIGITAL supply chain – DAM.” [Read More]

Mark is exactly right in his analysis, DAM is entirely about getting work done.  There are two types of work that are absolutely essential to generating ROI from DAM:

  • Getting the right digital assets to the people who need them.
  • Organising digital asset operations so that the process is efficient and reliable.

When DAM is delivering on its ROI promise, it should be one of these technologies that seamlessly blends into the background to the extent that users might forget it even exists.  This is how other non-digital supply chains work, most people don’t usually think much about how a parcel gets delivered, that the light comes on when you press the light switch in your home/office or that consumer items are available to buy in shops – they just are.  Despite all the other topics that often get included in discussions about DAM, I believe this is what most users really want from DAM technology and it is the reason they buy into it to start with.  Their expectation is that having invested into a DAM solution, they will get immediate access to the digital assets they require to carry out a given task.

Most DAM vendors (and users) are aware of the digital asset lifecycle where digital assets are created by combining a binary essence (or ‘file’) with metadata.  From a supply chain perspective, however, this is too late into the process.   If DAM platforms had access to information about potential digital assets before they existed, there is some scope to harvest and predict relevant metadata, in advance.  For example, if an organisation decides to commission some photography or comes up with a brief to purchase stock media etc, it has to have been discussed.  If DAM solutions are able to get access to this kind of data (in digital form) they could offer a number of contextual hints which would enable the suggestion of potentially relevant metadata when the digital asset comes into being.

The way most users currently have to catalogue digital assets with metadata is to enter it by-hand into a form or questionnaire.   There are tools like data loaders which allow users to prepare the material with spreadsheets etc but those require some technical expertise and prior experience of data preparation and they do not really solve the metadata problem, they just make entering it faster.   In literary terms, what happens now with DAM solutions is that users are given the equivalent of a blank sheet of paper and asked to come up with a plot line, theme and dialogue to tell the story of every single digital asset they expect someone in their organisation to need to use at a later date.  Expressed in those terms, it is easier to see why the task is not always given fullest attention in every case.  Clearly a best-selling masterpiece is unnecessary on every occasion (and derivative or formulaic copy is going to be quite a common theme) but DAM users need more help than they get now from DAM software to achieve some approximation of useful prose.

The key concept to understand is that data exists before a digital asset comes into existence which could become metadata at the point it is identified.  A DAM which has some visibility of the digital asset supply chain could use this and (based on a combination of user direction and machine-driven inferences) propose metadata suggestions for the user, which they can take or leave as required.  It is unlikely any of this will become fully automated in a way that is entirely safe and reliable to be left alone, but the level of effort required to catalogue assets properly should reduce and therefore become easier as a result.

To start to manage digital asset supply chains with the objective of using them as a source of relevant metadata, two activities need to take place and they require both users and DAM software developers to collaborate with each other:

  • Users need to identify all the points in their existing business processes where a decision to potentially create (or acquire) some digital assets.
  • Developers need to devise techniques for capturing this unstructured data and simplifying the process of associating it with digital assets.

I will cover this subject later in the education section, but both sides will have to negotiate a mutually acceptable level of effort and sophistication which can be reasonably expected of the other.  Digital asset supply chains offer a means to broker an answer to this problem, but both will need to work in partnership to realise it.

Devising A Digital Asset Supply Chain Management Design

There are three elements to a digital asset supply chain design which could help yield relevant metadata in a more efficient fashion so the full responsibility is not left with users, these are:

  • Identify common upstream sources of digital assets.
  • Introduce techniques and working practices to associate metadata with digital assets.
  • Develop components to automatically derive metadata based on their sources.

Identify Common Upstream Sources Of Digital Assets

This involves analysing the origins of materials which will get introduced to the DAM but which are not yet digital assets (i.e. essences or media files which have not been uploaded nor associated with any significant metadata).

With most DAM initiatives there are specific events or sources which tend to cause a large number of digital assets to get generated.  From my experience, it is rare for them to get ingested in a linear fashion with a consistent throughput over a given period of time.  There are ‘pressure points’ when something like a re-brand or product launch causes a lot of new material to rapidly appear within a relatively short space of time.  In non-digital supply chains, these are referred to as ‘bottlenecks’, they are simultaneously the cause of many supply-related inefficiencies, but also where significant value tends to get added as well.

Here are some typical examples for a content-oriented DAM:

  • Photo assignments
  • Re-branding exercises
  • New product launches
  • Promotional or marketing campaigns
  • Corporate communications events (e.g. annual reports)
  • Key account sales proposals
  • Stock media purchases
  • Existing sources (e.g. staff photographs)

The list is far from exhaustive and is skewed towards corporate marketing use-cases, but most readers should be able to see at least some items which look familiar and think of others which are more relevant to their needs.  The important point is to try to discover where (and why) the majority of digital assets get created in the first place.  There will be a number of assets that are hard to rationalise in this fashion, but providing around 80-90% can be allocated, that is sufficient to start with.  DAM software developers have a role to play in this process, but a lot of the analysis needs to come from the users themselves.

Having conducted a thorough analysis, DAM users may find the final list of sources might be quite lengthy (especially in a large organisations) and it could become more like a taxonomy.  For example, a given product could have many marketing campaigns associated with it over a period of time and there are variations in the metadata that needs to be applied in each case.  What is required is more like a ‘map’ describing where assets exist in the context of the organisation.  As described previously, metadata is contextual data, so for an efficient digital supply chain, it must be possible to see the relationship between a prospective digital asset and what else is going on inside the organisation.

Introduce Techniques And Working Practices To Associate Metadata With Digital Assets

Once the sources of digital assets have been identified, the next requirement is to link or associate a candidate asset to one of these.  This is consistent with what happens in a conventional physical supply chain, the items (assets) that traverse it need to be directed to the right destination.

There are several different methods to achieve this, ranging from fully automated approaches through to manual techniques.  With a logistics supply chain (e.g. delivering letters or packages), there is an expectation that those who want to use it will either write the destination address or a reference number on the item.  Even though digital supply chains have obvious differences to physical ones, the same process needs to occur.  In other words, to help derive or associate metadata to digital assets more efficiently, either the users must be prepared to ‘address’ the assets properly or whoever supplies them must include this information via another means.

In terms of using a DAM solution, this means either users need to select one of these sources by-hand, or another method such as asset suppliers like photographers, video production companies etc must be required to embed metadata into assets, or provide digital manifests with them that allow the same to occur.  Companies that depend on highly optimised supply chains to gain a competitive advantage will mandate certain requirements to their suppliers as a pre-condition of working with them.  Depending on the scale of the repository and expected growth levels in terms of asset volumes, the same may needed for the digital supply chains that service DAM solutions.

Develop Components To Automatically Derive Metadata Based On Their Sources

Equipped with an overview of the different sources where most of the organisation’s digital assets will come from (and some clues about what assets should be assigned to them) it is a lot easier to start to define some rules about how to catalogue digital asserts.  This is where it becomes easier for DAM software developers to evaluate the process of automatically assigning metadata so they can write code to achieve this goal.

There are a variety of techniques for implementing this, none are perfect or risk-free, however, they do provide the opportunity to continuously improve the efficiency of the digital supply chain and make it incrementally less painful and time-consuming to catalogue assets with relevant metadata.  These are some examples:

  • Metadata templates.
  • Analysing metadata entered by other users.
  • Data mining related metadata sources.

Metadata Templates

If the source of the asset (i.e. the purpose or reason it got created) can be both identified and communicated to the DAM, one option is to use pre-defined templates with metadata which applies to all assets of the same type.  I have seen a number of systems where such metadata templates exist, but either they rely on the users selecting the right one (which can be difficult if there are many) or an existing asset must be specified (which means finding a suitable one first).  In the scenario I am describing, the DAM solution already proposes a template based on the source of the asset, so the effort required to identify and select one is avoided.  This kind of feature is relatively simple for most DAM software developers to implement.

This technique is not without trade-offs.  Some risk factors include:

  • Assets which might share the same context but have totally different metadata requirements
  • Users not adjusting or adding to the template suggestions so numerous assets all have identical metadata applied which makes them hard to find.
  • Failed identification further up the digital supply chain, causing the wrong template to be suggested.

There are few easy answers to all this, however, something like visual prompts to warn users that the metadata they have entered is identical to the template (or the other assets they have just uploaded) might offer one solution.  Alerts and notifications to administrators (possibly via a dashboard rather than an email) could be another method for administrative users to find out about misuse of these tools so the appropriate training and guidance can be given.  A common theme of any supply chain initiative is that there must be an acceptance that exceptions will occur and plans must be made in advance to mitigate them.

Analysing Metadata Entered By Other Users

Most search facilities in modern DAM solutions now offer suggestions that are derived from what other users have entered previously.  I cannot recall seeing the same for metadata entry.  Having the ability to see what colleagues who are supplying assets for the same type of assets entered might help to achieve more consistent cataloguing, especially if the metadata can be correlated with the context identified upstream in the supply chain.

As with metadata templates, there are risks and trade-offs here too:

  • Unsuitable suggestions (e.g. incorrect or old terminology for products or brand names) can get introduced and begin to ‘infect’ other digital assets when other users decide to copy them without thinking about it.
  • Entering metadata which is used to describe lots of other assets is something of a double-edged sword since although it is consistent, if there are many thousands of assets all with the same metadata, finding them becomes challenging (this is similar to the point made about using templates).
  • In DAM solutions with controlled access, the security aspect must be taken into account. For example, if users are offered suggested metadata from asset groups they are not permitted to view then the DAM could leak sensitive information.  If the suggestions use the contextual hints from the digital supply chain, the security issue is less likely to occur, but this needs to be managed and enforced by the application (i.e. explicitly blocked, not a case of hoping that the users don’t discover how to work around it).  This will be a particular concern for multi-tenant Cloud-based solutions where many organisations simultaneously use the same platform.
  • The user interface needs to be handled with great care and skill. If the suggestions are too intrusive, then they could become irritating and get in the way when users need to catalogue something which is more specialised.  If the feature is hidden away then users might not realise it exists and never use it.

The last point is a critical one.  A lot of DAM vendors claim their interfaces are easy to understand and cite how different they are to older DAM solutions, which historically had a reputation for being not very user friendly.  I question how many really have made significant improvements in this area, however, especially when it comes to entering metadata (which still gets called an ‘admin’ feature by many).  A good number of the DAM systems I get to see have poor digital asset cataloguing interfaces, even though they look clean and efficient on the search and selection areas that are accessed by most users.  As such, simply optimising and enhancing the metadata entry facilities offered on many DAM solutions (without anything more complex) might yield  some benefits as an exercise in its own right.  The description applied to this process is something of a misnomer, calling this ‘tagging’ or ‘keywording’ vastly underestimates the intellectual demands placed on users and encourages developers to skimp on the level of effort that is applied to the user interfaces for these key areas.

Data Mining Related Sources

I intend to cover the data mining topic in more detail in the later AI piece, however, one observation I will make in relation to digital asset supply chains is that the technologies for analysing text data are a lot more developed than those for image recognition.  From the perspective of generating a higher ROI yield, they could represent a better opportunity.  For example, spam-detection in emails and concept/keyword analysis is not considered cutting edge any longer and the results (while far from perfect) are far superior in terms of extracting meaningful metadata.  At present, most DAM software I have seen is not able to mine arbitrary text-based sources for cataloguing metadata without custom development work being required.

In the upstream source identification stage, I outlined why it was important for organisations to devise some kind of classification taxonomy (or even just a simple list).  As part of the process of commissioning photography, planning re-branding exercises etc, a substantial amount of emails are written, planning documents, presentations etc.  In some cases, the volume of text would be sufficient to fill an entire shelf of printed books (probably more, in some cases).  All of this contains potentially useful metadata which can be analysed relatively easily now and synthesised with many other characteristics like the asset type, upload user, date created etc.

To go one step further, plenty of meetings are held where the topics that form these events/asset sources are discussed.  Recording them and using text to speech to capture the data in a way that allows it to be mined to extract concepts and assimilate them would not be either technically complex nor onerous for users.  For each, the users would only have to designate what project(s) they were associated with.  This is the kind of digital supply chain management which requires both very little effort on the part of users and some relatively straightforward software but which could enhance DAM ROI and make the cataloguing task a lot easier than it is now.  I acknowledge that users might not want every single meeting or phone call about operational subjects to be recorded, but there certain important ones where few would object if they knew it was happening (and that it might help reduce the workload involved for any digital assets that might subsequently get produced).

The chief reason why DAM solutions do not access narrative data now (whether digitised or not) is because there is no method to associate text with the aforementioned sources, nor ability for DAM solutions to mine it.  A common theme is emerging here: both DAM users and vendors fail to acknowledge that there is a wider digital supply chain already in operation, so limited effort is made to make it more efficient and both parties are negatively impacted as a result.

Implications Of Integrated Digital Asset Supply Chains

I am fully aware that there will be drawbacks to every method I have described in this article.  I have presented ideas and sketches for answers, the detail of implementing them will be demanding and it is yet another problem in DAM where there are no silver bullets.  As with most supply chain initiatives, however, there is an opportunity for continuous improvement where incremental enhancements can be made.  This can only happen if everyone involved in them understands the nature of the activity they are participating in.

Most vendors have so far avoided the need to dissect and re-factor the DNA of their applications by simply bolting on third party technologies.  The less skilled operators (or those who have jumped on the DAM bandwagon) have also been able to merely copy what their competitors do.  Those who are able to come up with some plausible answers to this, the biggest problem in DAM, could be able to use that investment to gain a competitive advantage over others who have settled for the current unsatisfactory status quo.  As should be apparent, the advantage will also accrue to those DAM users who grasp it too, not just in terms of selling more DAM software solutions and services but in terms of greater operational efficiency and lower costs.  This is the bottom line reason why Digital Asset Management is considered a worthwhile undertaking by most DAM users, so that is what the focus of everyone’s attention should be on.

 

This feature article has been contributed by Ralph Windsor, editor of DAM News and is the first article in a series of items with the title Finding Signs Of Life In DAM.


Leave a Comment