All 18 respondents to the survey (table 1) have an ILS created by one of the three major vendors (Ex Libris, SirsiDynix, or Innovative Interfaces). The number of records in their respective ILSs varies from 1 million to more than 12 million, with the average close to 6 million records. Staff size of the responding institutions varied greatly, as did the relative proportion of librarians, staff, and hourly staff and student employees. The number of full-time equivalent (FTE) librarians ranged from 45 to 266 (with an average of 108), number of staff ranged from 60 to 488 (averaging 211), and the number of hourly staff and student employees ranged from 50 to 454 (averaging 166).

The number of FTE devoted to batchloading activities varied from 0.8 to 11.0. The number of librarians devoted to batchloading activities varied from 0.0 to 5.0 (averaging 1.7), with other professionals ranging from 0.0 to 4.5 (averaging 0.9), and support staff ranging from 0.0 to 5.0 (averaging 1.5). One institution reported 0.3 FTE “other staff” participating in batchloading efforts and no institution reported using student assistants. While some correlation exists between total staff and the number of employees devoted to batchloading activities, notable exceptions also were seen: the library that reported the largest number of staff devoted to batchloading (11) had 716 total staff, whereas a library with even more staff (750) reported only 6 FTE devoted to batchloading. Similarly, 1 library with 589 total staff had only 1.2 FTE devoted to batchloading, whereas another library with fewer total (513) had 7.0 FTE devoted to batchloading. The 6 private universities had an average of 666 total staff, whereas the private universities had an average of 434 total staff. However, both the public and private universities reported an average of almost 4.0 FTE devoted to batchloading activities.

An analysis of staff devoted to batchloading activities in comparison with total expenditures for e-resources does not indicate any correlation. For example, when the responding libraries were ranked according to expenditures for e-resources, the 6 libraries that spent the least on e-resources averaged 4.6 FTE staff devoted to batchloading, the middle 6 devoted 5.2 FTE to batchloading activities, and the 6 libraries that spent the most on e-resources devoted 2.6 FTE to batchloading activities.

All respondents reported that they had redefined existing positions to add responsibility for batchloading. Four respondents (22.2 percent) reported that they had redefined existing positions to be solely dedicated to batchloading. Another 4 reported that they had created new positions dedicated to batchloading.

Two-thirds of respondents (12) expected to devote more staff to batchloading in the next five years. Because two-thirds of the respondents anticipated growth in staff levels for batchloading, one could assume that the batchloading workload is increasing or growing more complex at most institutions polled. The institutions who responded that no increase in batchloading staff was projected may feel they have sufficient staff devoted to batchloading for the short-term or they may feel that current budgetary climate and competing priorities would not allow allocation of more staff to batchloading, even if doing so were deemed to be advisable. Future research might explore the perceived priority of batchloading in the matrix of other services provided by libraries’ technical services departments.

Only 23.5 percent of 17 respondents reported having a dedicated budget allocated for ongoing costs of batchloading activities. The amount of money spent each year on the purchase of bibliographic records varied widely. One should note that because many collections (especially e-content) include bibliographic records as part of the purchase, expenditures for records are not necessarily an indicator of number of records acquired. No institution reported spending less than $1,000 per year; 3 of 16 respondents (18.8 percent) reported spending between $1,000 and $5,000; 1 library spent between $5,000 and $10,000; 4 libraries (25 percent) spent between $10,000 and $50,000; 4 spent between $50,000 and $100,000; and another 4 spent more than $100,000 per year. Based on this rough snapshot, the cost of bibliographic records, although only a small percentage of acquisitions and collections budgets totaling $12,000,000 and more, is not insignificant for many institutions.

Funding for the purchase of bibliographic records came principally from collections and operations budgets. Of the 17 respondents to the question on sources of funding, 15 (88.2 percent) reported using monies from the collections budget; 11 (64.5 percent) reported using monies from the operations budget; 3 (17.6 percent) reported using special funds; 3 reported using endowments; and 1 reported using grant money. (Note that respondents could identify multiple funding sources if applicable.) One respondent specified that funding came from the operations portion of the technical services budget and another noted that the exact source of funding for bibliographic records is unknown because his or her library uses a decentralized funding model. Responses to this question make clear that institutions use various funding sources and the majority purchase their records using a combination of collections and operations budgets.

Nine respondents described their funding models in greater detail. Four respondents indicated that the costs of batchloading were covered ad hoc and distributed across job or organizational lines. Most respondents reported that costs were not explicitly tracked, aside from the costs associated with individual purchases of batch records. In one case, funds were taken from the supplies budget, which, at that institution, is distinct from the collections development budget. In two cases, a specific person (the assistant director for Technical Services and Collections Development officer; the head of Collections) was named as overseeing the costs. One respondent reported his or her library's process as “quite distributed, and not very efficient currently.”

The picture that emerges from the responses is of no single funding model, but an array of approaches. The relatively distributed nature of funding models likely makes tracking of resources and their assessment challenging. For example, assessing the cost-effectiveness of the purchase, loading, and ongoing maintenance of a set of records is more difficult if the costs associated with the workflow are distributed throughout an organization, with some costs being more transparent and easier to track than others.

No respondents reported batchloading fewer than 100,000 records into their ILS during the past three years. Of 17 responding libraries, 2 (11.8 percent) reported loading between 100,000 and 200,000 records; 3 libraries (17.6 percent) reported loading between 200,000 and 500,000 records; and 12 libraries (70.6 percent) reported loading more than 500,000 records. An analysis of the number of records batchloaded did not reveal a correlation between that number and the respective libraries’ total expenditures. For example, the 2 libraries that reported loading between 100,000 and 200,000 records within the last three years averaged $19,207,552 in total acquisitions budget. The 3 libraries that loaded between 200,000 and 500,000 records had an average total acquisitions budget of $19,235,052. The 12 libraries that reported batchloading more than 500,000 records had an average total acquisitions budget of $17,513,333, lower than the libraries that batchloaded far fewer records.

These numbers indicate that many libraries added a sizable percentage of the total number of records in their respective catalogs in a short timeframe. Although 8 (47.1 percent) of 17 respondents batchloaded less than 10 percent of the total number of records in their catalogs in the past three years, 6 (35.3 percent) reported that such records amounted to 10–20 percent of their catalog, and 1 library reported that they totaled 30–40 percent. In one noteworthy case, a library responded that recently batchloaded records represent more than 50 percent of the total records in its ILS.

All respondents reported deleting records in batches, some in large numbers. Only 3 (17.6 percent) of the 17 responding libraries reported deleting fewer than 1,000 records each year; 5 (29.4 percent) reported deleting between 1,000 and 5,000 records each year; 4 (23.5 percent) reported deleting between 5,000 and 10,000; 2 (11.8 percent) reported deleting between 10,000 and 50,000; and 2 reported deleting between 50,000 and 100,000. One library reported deleting more than 100,000 records per year.

The most common reason (reported by all 17 libraries that answered this question) for deleting records or suppressing records from public view was cancellation of a subscription to an online databases or collection. Nearly one-quarter (4 libraries; 23.5 percent) cited withdrawal of physical items, 8 (47.1 percent) cited invalid URLs, and another 4 cited errors found in records. Other reasons not explicitly listed in the survey question but identified by respondents included normal maintenance, e.g., the title was no longer available from a vendor or publisher, publishers were removed from EBL or ebrary profiles, and the contents of large sets and e-book packages changed. The number of responses to this multiple-choice question makes clear that most libraries have multiple reasons for batch deleting records from their ILSs.

The responses summarized above indicate that batchloaded records compose a remarkable proportion of records in many institutions’ ILSs and that batchloading is a dynamic process (or set of processes) that involve not only the batch adding of records but also batch removal. Online collections are perhaps more subject to change and in briefer timeframes than physical collections. When a substantial portion of an institution's collection is digital rather than physical, the fluidity of the catalog and thus the need to batchload and batch remove records is likely to increase dramatically. While monographic records for print materials and microforms are relatively static, records for monographic e-resources tend to be more subject to change. For catalogers with a purely monographic background, this fact may require change to a mind-set closer to that of a serials cataloger, who is accustomed to repeated edits of a single record. Given the current scope of batchloading (and its anticipated future), management and coordination of access to digital assets will no doubt be an important role for libraries technical services departments in the coming years.

Management of batchloading processes is commonly shared across participating libraries’ organizational units. The most frequently cited unit with responsibilities for batchloading activities is Cataloging, identified by 13 (76.5 percent) of the 17 respondents to this question. Additional units identified were IT or Systems (9; 52.9 percent), Acquisitions (7; 41.2 percent), Collection Development (3; 17.6 percent), and Public Services (1 library). Another 6 respondents (35.3 percent) reported additional units with responsibilities (E-Resources Management Section; E-Resources, Serials, and Database Management; Scholarly Resource Integration; and Knowledge Access and Resource Management Services), reflecting the diversity of organizational structures in place in responding institutions. Respondents were given the option of selecting more than one unit and had the option of providing their own response.

Respondents indicated that Cataloging held primary responsibility for managing batchloading activities in 12 (70.6 percent) of the 17 responding institutions. This is followed by IT or Systems (7; 41.2 percent), Acquisitions (2; 11.8 percent), and “other” (4; 23.5 percent), which includes E-Resources Management Section; E-Resources, Serials, and Database Management; Scholarly Resource Integration; and Knowledge Access and Resource Management Services. Since the responses to this question total more than 100 percent, the possibility exists that respondents had trouble deciding who was primarily responsible for managing batchloading activities. However, Cataloging and IT or Systems most frequently were identified as bearing primary responsibility for batchloading activities in the majority of libraries. None of the responding libraries reported that Public Services or Collection Development were primarily responsible.

Of 17 respondents to the question about timeliness of batchloading, 6 (35.3 percent) reported that loading records for large packages took longer than three months; 4 (23.5 percent) said such loads occurred within one to three months; another 4 libraries said they occurred within one month; 1 respondents said within two to three weeks; and 2 reported batchloads occurred within a week. Asked whether the reported turnaround time was acceptable, 11 respondents (64.7 percent) said no, while the rest said yes.

In addition to batchloading records in the local ILS, 10 (58.8 percent) of 17 respondents reported that their libraries used other methods, such as a metasearch engine, to ensure access to collections. One might assume that libraries providing access via alternative routes may feel less pressure to load records for all their e-book holdings into their ILS, although these alternatives to batchloading may have their own challenges and disadvantages. The alternatives mentioned include WorldCat Local using the “treat as held option,” SFX Find It deep linking for conferences and proceedings within article databases, Summon from SerialsSolutions, e-indexes and aggregated index services such as Primo Central, MetaLib, and ebrary's platform. One respondent reported bypassing their ILS and loading records directly into their discovery layer (Ex Libris's Primo), with the possibility of moving the record sets to the ILS (Aleph) at a later date. The survey did not capture data that might have allowed the authors to assess the relative merits of these approaches. Further research is warranted in this area.

Respondents were asked if their libraries ensure access to titles in Google Books via the ILS. Of the 17 respondents to this question, 11 (64.7 percent) do so. Of libraries that do ensure access to Google Books, the vast majority (10 of the 11) ensure access only to Google Book titles for which their library holds a print version. Only 1 library reported that it ensured access to titles not held by the library. Ten libraries used the Google API to ensure access to Google Books through their ILS and 4 reported using other methods, such as a script written by Systems staff, persistent links for books scanned for the Google Books project (presumably the links are pushed into the corresponding print records), an OpenURL resolver (Umlaut, an open-source link resolver, in conjunction with SFX's knowledgebase), and selective searching of the Google Books database by subject. No respondents reported batchloading records into the ILS to ensure access to titles in Google Books. Because Google requires delivery of a bibliographic record for every volume a library sends for digitization, one can assume that libraries always have records for the titles they themselves have supplied to the Google Books project.

Six (35.3 percent) of 17 responding libraries reported that they ensure access to titles in HathiTrust via their ILS. Only 1 of those 6 reported that such access includes titles for which the library holds no print version. Five libraries reported using the HathiTrust API to ensure access, and 1 library reported batchloading records into the ILS to ensure access. Four libraries reported using other methods, including a cataloger manually adding links to print records and using an OpenURL resolver (Umlaut + SFX).

When asked if their libraries batchoaded records for freely available e-resources in addition to those available via Google Books and HathiTrust, 9 (56.3 percent) of 16 respondents reported batchloading records for these resources. Eight of these 9 reported that such titles are selected by subject specialists and bibliographers. One respondent said titles are loaded on the basis of patron suggestion. Four reported the titles are selected by vendors. Four respondents provided other details. One respondent reported that free titles are selected by the Digital Library Program and library administration, 1 reported that the majority of free e-resources loaded are U.S. government documents, and 1 reported loading MARCIVE records for U.S. government publications but was considering batchloading records for Open Access journals (via SFX and MarcIt) and for National Academies Press titles (all of which were made freely downloadable in PDF format in June 2011).

Workflow is directly affected by high-level policy questions, many of which remain to be answered either by individual libraries, or perhaps more usefully, in a coordinated fashion by consortia or professional library organizations. What is the role of the online catalog in a world increasingly shaped by cloud computing and network-level resource discovery? Should an institution's catalog include records for free resources? If so, which ones and how should they be selected? To what extent should an institution's online catalog replicate access to resources in HathiTrust and Google Books? Batchloading workflows (and workloads) will certainly change as these questions are addressed.

A majority (14; 82.4 percent) of 17 responding libraries reported that the use of vendor-supplied metadata for digital resources has lowered their library's quality standards for bibliographic data; the rest said these metadata had not changed their standards. No respondents reported that use of vendor-supplied metadata had caused them to raise their standards.

Seventeen respondents answered the question about how they assess the quality of vendor-supplied records. All 17 reported that they used visual review by catalogers or other staff to assess the quality of vendor-supplied records, and 9 (52.9 percent) also reported using automated validation with MarcEdit or other software. (Given the size of many record sets, one might reasonably assume that any visual review by catalogers is of a sampling of records rather than of every record.) Two of the 17 responding libraries reported using other methods, specifically data analysis, in-house validation, and loading the records into a development system and evaluating them in the user interface. Ten respondents described the tools and applications they use in greater detail. Six mentioned MarcEdit; others mentioned Excel, the programming language Ruby, and locally devised Perl scripts or C++ programs to assess records. In one case, a respondent drew an important distinction between data integrity and record quality. One respondent reported using processes that are part of the ILS to identify invalid tags, indicators, and so on in external files before loading. One respondent reported loading the file of records into a test region on the local ILS for review.

Seventeen libraries reported using a variety of methods to address quality issues in the bibliographic records they load. Respondents were given several options and were able to select more than one. Sixteen reported that they deal with different levels of quality and fullness from different vendors by editing records using MarcEdit or locally devised scripts to meet local or national standards. Five (29.4 percent) reported that they also edit records manually to meet local or national standards. Nine (52.9 percent) reported that they sometimes accept and load records “as is,” and 2 reported that they have rejected records that they determined were unacceptable. One respondent said the library tried to pressure vendors to supply improved data. From these responses one could assume that different approaches are taken with different record sets, which is not unexpected given the varying levels of quality reported.

Thirteen (76.5 percent) of 17 respondents said they had rejected sets of bibliographic records because of quality issues. The respondents who had rejected records cited many reasons:

• lack of authority control or subject access
• bad data that would have been difficult or impossible to clean up by automated means
• incomplete title fields
• character encoding errors
• right-to-left text orientation errors
• records lacking unique identifiers
• nonstandard cataloging practices, such as cataloging groups of unrelated material as a whole or creating unexpected analytic records
• concerns from public service librarians that too many nonspecific subject headings would have overloaded the subject areas and made books and journals too hard to find
• technical limitations of the ILS (unable to match incoming records to existing records)
• the possibility of pulling together better sets of records for the same resources
• invalid URLs
• overlap or duplication with print records

Batchloading vendor-supplied records for electronic resources can result in multiple records in the ILS for the same title. Thirteen (76.5 percent) of 17 respondents reported that they accepted multiple records for the same title, 5 (29.4 percent) followed a single-record approach and attempted to describe all instances of an e-resource with one record (but keeping print separate), and 4 (23.5 percent) followed a single-record approach and attempted to describe all instances of a resource with one record (print and electronic). Four respondents provided more detailed answers. One said they take a single-record approach for serials but do not deduplicate for monographs; they also do not yet have a plan for how they will implement the provider-neutral standard in the ILS with batchloaded records coming from a variety of sources. Another reported accepting multiple records when print and electronic records cannot be easily matched and accepting multiple records for e-resources when they are available from multiple vendors. The third respondent reported using a single record for print and electronic serials, but separate print and electronic records for monographs and attempting to use a single record for all vendor iterations of electronic access to a title. The final respondent reported that discussions of a single-record approach were ongoing, particularly regarding serials but also regarding monographs.

One can assume that no institution has the resources to originally catalog or upgrade records for all the digital assets to which it has access, so the need to batchload records is unavoidable. As reported above, 82.4 percent of 17 responding libraries reported that the use of vendor-supplied metadata for digital resources had lowered local quality standards for bibliographic data. One might intuit that respondents perceive the quality of their catalogs as having declined as a result of batchloading. While some vendors have made efforts to improve their metadata to conform more closely to national standards, others have not. The authors believe that many public service librarians would agree that some access to resources, even if imperfect, is preferable to none, but this leaves the question of bibliographic quality standards unanswered. Technical services departments, and especially catalogers, may be increasingly called on to justify maintaining current quality standards in light of shrinking budgets and a ballooning universe of digital resources. The fundamental question will remain: are users able to find what they are seeking quickly and easily? If not, how can quality standards be modified to better serve users’ wants and needs?

Many respondents reported collaborative efforts to create bibliographic record sets. Two approaches were identified: collaborating with other libraries to address resources owned in common and collaborating with vendors and bibliographic utilities. Seven (38.9 percent) of 18 survey participants reported that they had collaborated with other libraries. Notably, all of these were public institutions, with 58.3 percent of the 12 public institutions reporting some collaboration with other libraries. Two libraries reported contributing to the CONSER (Cooperative Online Serials) project to catalog serials in the Directory of Open Access Journals (www.doaj.org). Two libraries mentioned that they had contributed to Committee on Institutional Cooperation (CIC) efforts to catalog collections purchased jointly, such as the Springer e-book collection. Some cataloging efforts were coordinated by OhioLINK and the University of California's Shared Cataloging Project.

Similarly, 7 (38.9 percent) of 18 survey respondents reported collaborating with vendors or utilities to create bibliographic record sets for electronic or microform collections. In this case, 3 of the 6 private institutions reported collaborative efforts with vendors, whereas only 4 of the 12 public libraries collaborated with vendors. Some libraries reported specific collaborative projects in which they participated. These fall into two categories. In the first, libraries provided feedback to vendors to help improve their ongoing services. An example of this included a library that collaborated with a vendor to develop and deliver records for e-books as part of a patron-driven acquisitions pilot project. Another library reported working through CONSER to improve the quality of serials records that are subsequently distributed through a vendor, in this case SerialsSolutions. The second type of collaboration with vendors included the cataloging of discrete collections of materials, such as The Making of Modern Law: Primary Sources, 1620–1926 collection (gdc.gale.com/products/the-making-of-modern-law-primary-sources-1620-1926) or the EEBO collections.

The survey findings demonstrate that some collaboration is taking place, but it is opportunistic rather than methodical or programmatic. Some collaboration arose from consortial purchases leading to consortial efforts to improve or create records for resources purchased jointly. Others resulted from individual libraries’ efforts to work with vendors to improve their products. This area has much room for improvement. More broadly based collaborative efforts could benefit more libraries, increase efficiency, and reduce costs.

Of the 17 respondents to this question, 8 (47.1 percent) reported using MarcEdit as part of the batchloading process; 3 (17.6 percent) reported using locally devised scripts; 10 (58.8 percent) reported using a combination of both MarcEdit and locally devised scripts; 4 (23.5 percent) said they use other software and scripts; 2 reported using the Millennium system load tables; 1 reported using UltraEdit and writing preprocessing programs specific to each load stream; and 1 reported using a suite of resources, including XSLT, SQL and Data Warehousing, Ruby, Perl, Primo Normalization Rules, Unix, Bash, Awk, and Excel.

A significant majority (14; 82.4 percent) of 17 respondents said that IT support is necessary to maintain batchloading at their institutions. Respondents described their IT needs in a variety of ways, including the following:

• customizing records
• programming
• working on special projects or “tricky” problems
• scripting for data transformation and automation
• adapting of load tables for new record sets
• troubleshooting
• running system reports or load programs
• batch deleting records
• creating or changing authorizations for staff access to servers
• creating FTP scripts to retrieve files from external sources and move them into the correct location on servers

These responses indicate a wide range of approaches, probably dependent on organizational structure, job descriptions, and local expertise; 15 (88.2 percent) of 17 respondents reported that their current ILS presents technical obstacles to managing batchloads. The 15 respondents who answered yes to this question described the following obstacles:

• an inability to mark records for deletion
• lack of sophistication and customizability of ILS loaders
• lack of levels of granularity to support multi-campus holdings structure
• ILS software upgrades that cause previously devised processes to fail
• inability to effectively match and replace or overlay records
• limit of loading a certain (relatively small) number of records at one time
• extremely limited global edit options
• unsafe batch deletion features
• lack of a unique record ID in every record that is usable as a match point
• “all-or-nothing” security features that preclude authorizing staff to perform certain essential functions without giving them access to “everything”
• inability to make batch updates to MARC holdings records
• limited system resources that make scheduling a large number of loads challenging
• absence of a deduplication utility
• cryptic error messages that make data correction before load difficult or impossible.

Some common themes that emerge from these responses are the challenges of record matching, batch deletion, MARC holdings management, system resources (i.e., scheduling loads including the number of records that can be loaded in a day, week, and so on), and global editing of records. No ILS is perfect, but the authors suggest the possibility of encouraging ILS vendors to build certain features deemed to be desirable for batchloading into future releases of their products. Following up with the 2 respondents who reported their ILSs present no obstacles to batchloading would be instructive. The question also arises as to whether locally maintained ILSs are the future of library asset management, or whether more and more data and database maintenance will be shifted to the cloud and third-party vendors. In cases where responsibility for batchloading is distributed across technical services and IT departments, coordination, thorough and up-to-date documentation, and effective communication are assumed to be highly desirable, if not essential.

Survey respondents reported various methods of assessing the quality and impact of their batchloading efforts. Nine (52.9 percent) of 17 responding libraries reported using usage data as an assessment tool. End user feedback was used in 7 libraries, while only 1 library used formal end user testing as an assessment tool. Two libraries used faculty review, and 1 conducted focus groups as an assessment tool. No libraries reported the use of an end user survey. Other assessment activities reported included review by libraries-wide task forces, review by bibliographers, quality assurance testing, staff review, and error reports. One indicated that his or her library did not conduct assessment on a regular basis.

Two libraries made changes to their policies and procedures on the basis of the results of their assessment activities. In both cases, usage data revealed a notable increase in the use of resources after records for the resources were batchloaded into the catalog. This resulted in an increase in the staff resources devoted to batchloading activities to minimize the time between when the resource was available and when the records appeared in the online catalog.

All but 1 of the 17 responding libraries informed one or more constituents when a batchload was completed. Multiple answers were possible. Twelve informed all library staff, while 9 specifically informed subject specialists and selectors. Three informed academic department faculty who might be interested in particular loads. None of the responding libraries informed students or the general public about completed batchloads. One respondent's library used a wiki to post information about batchloads, 1 reported notifying the staff at the requesting library, and another indicated relying on “various informal notification channels within the Libraries.” Four libraries indicated that they relied on selectors to inform their respective communities. Some notification occurs at most libraries, but it is not consistent and often relies on the liaison role of subject specialists and selectors.

Survey respondents were asked to select what they felt were the biggest challenges related to batchloading activities facing their libraries. They could select more than one answer, but they were not asked to rank them, and they also could supply their own response. Of the 17 respondents, 14 (82.4 percent) considered inconsistent record quality to be one of the biggest batchloading challenges facing their respective libraries. This was followed by staffing (13; 76.5 percent), ongoing maintenance (10; 58.8 percent), vendor technical support (9; 52.9 percent), local technical support (8; 47.1 percent), and funding the purchase of records (6; 35.3 percent). Other challenges identified in the comments section include the need for good records from developing countries, time needed for initial analysis, managing record-quality expectations, achieving ongoing confidence in any particular source, and the proliferation of potential record sources, even for the same resource.

Respondents’ expectations for the future were very similar to their view of current challenges with batchloading. They were asked what they felt would be the biggest challenges they would face in the next five years. As in the previous question, they could select more than one answer, but were not asked to rank them, and they also could supply their own response. Fourteen (82.4 percent) of 17 responding libraries expected inconsistent record quality to continue to be their biggest challenge. Staffing was considered to be a future challenge by 14 libraries, and vendor technical support, ongoing maintenance, and funding records purchases were each identified by 11 (64.7 percent) libraries as additional challenges. Eight libraries (47.1 percent) identified local technical support as a challenge. Additional concerns identified by respondents included the future of the ILS and the number of resources with no records available at all. While inconsistent record quality was identified as the biggest current concern of respondents, the challenges libraries most expected to face in the next five years were funding, staffing, vendor technical support, and maintenance.

Responding libraries reported using several methods for maintenance of batchloaded records. The most frequently cited methods of the 17 responding libraries (and the percent who reported using these approaches) were the following:

• notifications from vendors that titles have been deleted or added, or that new or updated MARC record sets are available (14 libraries; 82.4 percent)
• feedback from patrons (13 libraries; 76.5 percent)
• feedback from subject specialists (12 libraries; 70.6 percent)
• regular review by catalogers or other staff (6 libraries; 35.3 percent)

Other methods reported were monthly reloading of batchloaded records, subscribing to OCLC WorldCat Collection Sets standing orders, setting up reminders in the library's ILS, running a URL checker report within the ILS, and running scripts to check with vendors for new or updated files.

Notifications about invalid links or other errors from patrons, faculty, staff, and other library users can take several forms. A website for reporting functionality problems is used in 15 (88.2 percent) of the 17 responding libraries. E-mail (used by 14; 82.4 percent) and telephone (used by 8; 47.1 percent) reports of inaccuracies also are common. Other options for reporting problems included QuestionPoint (www.questionpoint.org), chat reference, the online catalog's feedback form, in-person reporting to a staff member at a service point, and notifications sent to an error-reporting electronic discussion list (OhioLINK).

URLs are seldom checked in the respondents’ online catalogs. More than half (10 libraries; 62.5 percent) of the 16 that answered this question reported never checking the URLs in their catalogs, 3 (18.8 percent) checked URLs irregularly, 1 library checked them quarterly, and another checked monthly. One library reported running a link checker on resources included in a separate database, but never against the entire ILS. Another library reported using the SFX link resolver service whenever possible to avoid bad links.