Indexing, the often underestimated field of Marketing Resource Management

Undervalued and rarely used to best effect by companies, good media indexing is nevertheless critical to guaranteeing the success of your MRM media library project. What are the main pitfalls to avoid? The challenges ahead? The Wedia indexing team provides some answers.

“In companies, everyone wants to be able to search the media library with the same ease as a Google search. Nevertheless, obtaining relevant results requires rigorous preparatory work, and that’s the problem in a nutshell…” summarizes Marion Quantin, head of Wedia’s indexing department. In fact, sketchy content indexing constitutes a serious handicap for any MRM (Marketing Resources Management) project. Without quality indexing, employees are disappointed with the results of their searches and lose faith in what should be the go-to reference for their content. The down side is well known: the inability to reuse images, videos, PDFs and other documents which seriously undermines the return on investment for content production.

A legitimate question: Why is content indexing not made easier or even automated by MRM solutions? For one simple reason: there is no generic classification to suit all companies. Each brand has its own unique history, products, glossary, business vocabulary and… varying employee profiles. “This is the number one challenge for indexing,” says Quantin. “A media library is a resource for several user profiles, each of which uses its own lexicon. As a result, an image with a simple generic caption is unlikely to match user requests. In fact, you don’t choose keywords in the same way you would write a caption…”

As an example, images of people whose metadata does not distinguish between men, women and children are very unlikely to help a marketing manager looking for an image representing a family. “The flip side of this, which is in fact quite frequent with commercial image agencies, is ‘over-indexing’” warns Marion Quantin, “or linking too many keywords to a piece of content.” For users, this over-indexing results in a lot of ‘documentary static’, lists of results that are quite remote from the subject of their research. Unfortunately, this is the fate of many media libraries: they become unusable because of this ‘documentary static,’ or the opposite pitfall, ‘documentary silence,’ resulting from insufficient metadata linked to content.

Can the use of artificial intelligence make this task any easier? “Customers expect AI to automate indexing and these expectations are understandable: machine learning and deep learning algorithms show great promise for progress in this area. Yet once again, the importance of one factor is underestimated: training. To use AI properly, you have to train it, and determine what it recognizes and with what level of probability, just to make it capable of carrying out pre-processing. For instance, by default, for the AI, a lit window in an image is primarily interpreted as a screen.”

In practice, making good use of AI for indexing requires two preliminary steps. The first step is preparing the data required for learning. In practical terms, this means ensuring that large volumes of media are run through the algorithms, with metadata that have been edited and verified enough to constitute a reliable learning base. The second step is creating reference dictionaries. These are essential for weighting, filtering and refining AI results. The “simple” detection and recognition of text in an image requires filtering any misspelled text and overweighting the tags corresponding to the reference lexicon (product names, business terms, names of people, etc.).

The AI’s performance therefore depends first and foremost on the availability of sufficient data to support the learning phase. Otherwise, how can companies ensure high-quality indexing of their resources? Malaïka Fauveau, Information Manager in Wedia’s indexing team, reminds us that: “We do know the answer to that. It’s a matter of having this work carried out by reference librarians. Nevertheless, in practice, managing the database is often assigned to a member of the communication or marketing department who has other responsibilities and little or no expertise in the field.” However, this is not the worst of the possible situations. “The most complex case is when several dozen contributors work on the media library without any shared guidelines.” Another source of complexity is the sheer volume of assets. “We have customers with a database of more than a million pieces of content. It’s hard to index that kind of volume without dedicated resources.”

For many clients, awareness that good content indexing requires specific expertise dawns after 6 to 12 months. Often called upon to help in this kind of situation, Wedia’s indexing team gets involved in several operations: defining the documentary structure, actually collecting the assets, integrating them into the database and, finally, indexing content via metadata editing. These are the kinds of tasks that, by default, are frequently underestimated. Of course, sorting out a repository that has already had content uploaded to it for a few months requires an audit of the database. “Another subject to bear in mind,” cautions Malaïka Fauveau, “is training the media library contributors. An MRM solution may be well designed and efficient, but it requires staff trained in good indexing practices to realize its potential.”

Implementing media library procedures, calling upon the skills of reference librarians, training, etc. “Where is the return on investment (ROI)?” The question is bound to be asked, and the answer takes many forms. Well-indexed assets are automatically re-used more, thus increasing their ROI. More successful research by employees means obvious productivity gains. Finally, a structured database enables the brand to manage the evolution and life cycle of its assets better. The benefits of professional indexing are therefore not evaluated over a single quarter, but are clearly demonstrated over the long term. On a daily basis, they are in fact the glue that holds an MRM project together. Who wants to risk managing without it?


Want to be updated with
our latest news?


  • i

    Why this form? As a software company, we know how important your privacy is. Our customers know and trust us, they handle their own confidential data in the platforms we operate for them. They know they are protected by our strict respect for their information, and by our ISO and TISAX certifications.

    Your personal data is therefore secure and we are transparent about how it is handled.

    Why this form? Because we would like to reserve the possibility to contact you in order to send you information, whether to propose you other interesting downloads, or to understand if you intend to launch a project soon... In no case, we will transmit your data to third parties.

    And of course, you can unsubscribe from our newsletter at any time, or ask not to be contacted. Finally, you can request to view and/or delete your personal data at any time.

    To learn more, read our « privacy policy ».


Follow us on social networks