As XML workflows become more established within the publishing industry, the need has arisen within publishing organisations to develop electronic systems for storing and processing their digital content holdings, so called content management systems (CMSs).

Yet though the underlying technologies of XML and other digital file formats are themselves well-established, the selection, installation and use of a CMS for managing them are still activities that can be fraught with technical and commercial uncertainty.

A recent trend within publishing is the 'self-build' CMS, in which a publisher decides to assemble a bespoke system out of available (often Open Source) components. Based on real-world experience of a number of publishers' plans and experiences, this paper will examine the reasons for this trend and discuss the commercial and technical arguments for and against such initiatives.

One of the most immediate barriers for publishers is a mismatch between what is expected of a CMS in publishing, and in many other environments. In the wider world, a CMS can sometimes be seen as little more than back-end to a web site; yet for a publisher a CMS is a system marshalling large amounts of content through complicated workflows. Though the CMS marketplace is a crowded one, there are good commercial reasons why CMS vendors can eschew the publishing market, as simpler CMSs can sell better into less demanding (and richer) markets.

Despite the rise of XML, published information is still much more than just structured data. A CMS for publishers must offer good control of unstructured data as well as XML, particularly in the difficult area of digital right management (DRM). Again, like CMSs themselves, DRM often means something different to the publishing world, than to the non-publishing world - in particular any publishing CMS should be aware of the issues surrounding permissions management.

Though technically intriguing, the ability of many CMSs to explode XML data into fragments for storage is little needed in publishing, where data transactions are typically applied to larger units of content relatively infrequently. The expense of acquiring and deploying a 'true' fragmenting XML repository is one which publishers can often avoid, particularly when they often already have effective storage systems within their businesses.

Though avoidance of 'vendor lock-in' has long been one of the promises of structured markup technologies, this means little if a publishing workflow becomes practically dependent on a particular commercial CMS. However, the costs of maintaining a bespoke system are not negligible - how do they stack up against the costs and benefits of acquiring a CMS from a commercial vendor?

The paper will conclude by suggesting that for the first time the crystallisation of a number of information standards, together with a number of high-quality commercial and Open Source components, allows publishers the ability to allow a 'self-build' Open Source based CMS room for consideration when they are acquiring such a system.