In March 1998 I was approached by the Centre for Indigenous Environmental Resources, in Winnipeg, to automate its collection. The collection's focus was Aboriginal environmental material. The material was in a number of formats - books, serials, CD-ROMs, audio-visual material, maps, newspaper clippings and article offprints. Two thousand government documents, unpublished material compiled by First Nations' communities and mainstream items competed for my attention. There was no classification system, no cataloguing and no automation system. In the last year I have classified and catalogued the material as well as implemented an automation system. I created a web page of automation resources on completing the project (www.escape.ca/~automate/resource.html) The following are some of lessons I learnt and the tools I found helpful.

On Selecting an Automation System

Although it may seem self-evident, the first thing to do when faced with automation is to procure evaluation software from system vendors. At the time I was reviewing evaluation software one sales representative was unable to provide me with evaluation software. The CD-ROM the company sent me was simply a presentation introducing the software. Another representative was unwilling to allow me to install the software and to demo it myself. Needless to say, I could not consider these systems.

It is necessary to test, test and test again the evaluation software you receive. Make sure the system can do what the literature says it can do and that what is being promised is not "vapourware." Assume the evaluation software is your new system and import and export records, edit them, print labels (if the software allows), do different kinds of searches, print them out, connect to web URL's etc.

Once you've installed the demo software getting answers to your questions requires persistence. Try to get hold of technical support staff for technical questions, rather than relying on the sales rep assigned to you. Try to speak to the second tier staff in the TS departments. One customer service rep I called was paging through the manual as I asked him questions I knew were not covered by it. Similarly, be alert to long pauses when speaking to support staff on the phone - the staff member is probably going through the online help screens/manual as you speak. Ask to speak to a cataloguer about cataloguing questions (this will reveal whether there are cataloguers or even librarians on staff.) Too often the TS and Customer service staff do not know the product well enough or have insufficient library background.

Your experience with the customer and technical service division will give you some idea of the kind of support you can expect. Ask for a list of customers and call them to find out about their experiences with the system and the company's technical support.

My own experience of all the evaluation software for small libraries was disappointing. Each package had glitches or shortcomings that one wouldn't expect — no Boolean searching for example, or no access to certain MARC tags (even though the system imported and exported MARC it was not possible to edit certain tags, and sometimes most of them). In the light of the shortcomings of small library software it is very important to know what you really need and what you are willing to compromise on.

In my case there were a three main criteria, along with a number of technical specifications I established. Technically, the system needed to be windows based and, if possible, developed for an NT platform. It needed to have client-server architecture and be a 32 bit application. It had to be Year 2000 compliant and the vendor willing to sign a document taking responsibility for costs incurred should the system not prove to be Year 2000 compliant.

My main criteria as a user of the system were that it should be MARC compliant, and allow for editing of all MARC fields, have a user-friendly search interface that allowed for Boolean searching and have a web pac that was not too expensive. I did not require a serials or acquisitions module.

Even once you've decided what to compromise on, you discover other less crucial but annoying shortcomings in the software — sometimes based on the fact that professional, experienced librarians truly do not seem a big part of the development of these systems (despite what the sales reps may say). The reports may not report on what you need. The system may not import imperfect MARC records, because it is MARC compliant (which you want, but which is annoying when you find a good record with a small error in it and the software doesn't allow you to edit the record before importing it.) Try to impress upon the sales representative that you understand all systems to have glitches and that you would like to know of the glitches before purchasing the system. One way of doing this is to ask about upgrades that are being planned for new releases. Nasty surprises seem to be par for the course, however.

On Cataloguing and Retrospective Conversion (Recon)

If you have an automation system that is MARC compatible/compliant I suggest that you outsource the cataloguing or retrospective conversion if you have over a thousand records (consider doing this even if you have fewer than a thousand records).

The timeframe for the "recon" will depend on the turn around time provided by the vendor you use. There is bound to be a minimum of 6 weeks turn around for even the smallest collection and for larger collections it will be a lot longer. You should be careful about choosing the vendor. Make sure their customer base consists of academic or special libraries (as opposed to school libraries) and ask them what their sources are for the records. Decent cataloguing is a must.

The original cataloguing can also be really time consuming whether you use a vendor to do this (often a pricey undertaking) or do it yourself. Part of the first step in automating an uncatalogued collection is to ascertain how much of the material is more mainstream (has an ISBN or CIP) and how much is "grey material" for which you (or a vendor) will have to do original cataloguing.

If you have fewer than a thousand records you could get a subscription to a bibliographic utility like AMICUS, OCLC, or RLIN. I would suggest employing an assistant to do the copy cataloguing. You will not need very high level assistance - it is usually as simple as typing in an ISBN to pull up the matching record.

Another option for very small collections is to use a Z39.50 client like Bookwhere 2000 (if your automation system does not have an inbuilt one). The Z39.50 protocol is best known for allowing users to search remote databases using the interface of their own OPAC. Bookwhere 2000 is software that provides an interface with which to search library catalogues worldwide over the Internet. It enables users to download records from a library catalogue in MARC format. If your automation system is MARC compatible/compliant you should be able to import most records into your system once you've downloaded them. It is a painless exercise.

A good source of cataloguing is the Library of Congress but it is difficult to get a good connection to its Z39.50 server. There are many other academic and public libraries from which you can procure cataloguing - including libraries in the countries in which foreign books have been published.

Last year I used Bookwhere 2000 to catalogue 2,000 items in the course of setting up a library. I decided to do the cataloguing myself as there was not much else I could do during the 6 weeks the vendor required to do it. The upside of my decision is that I now have great records because I chose the best records from universities across North America and then had my assistant add tables of content and summaries for those items I felt required them. I also have a familiarity with the collection that I would not otherwise have.

The downside of doing the cataloguing/recon oneself is that it is very time consuming and expensive. I pursued it because the subject matter with which I was dealing - Aboriginal environmental issues - is inadequately treated by LCSH and LC classification. I did not want cookie-cutter records for the items - I wanted to see how other libraries were dealing with the material.

It took me six months to choose the software, install it and procure copy cataloguing of about 2,000 records. I did most of the work myself - and the work was very intense. It is now a year later and there is still a bit of a backlog. It has taken a further six months to set up acquisitions and serials accounts and processes and to create a web page with web resources for the library.

Automating is demanding work. The selection of a system entails a fastidious attention to detail and a sceptical eye. The cataloguing/recon work is intense and time-consuming even if it consists of working with a vendor to get it done correctly. There are numerous rewards for the first-time automater, however, the greatest of which consists of witnessing the satisfaction of patrons as they learn the skills required to search online catalogues and their success in finding relevant material more efficiently.