Jump to:
Disclaimer: These are opinions based on 18 years in library tech. Your situation may vary. Always talk to your IT director and legal team before signing anything.
Discovery Systems
Discovery systems are supposed to make your collection searchable. They do that. But are they worth what vendors charge? Let's talk.
Primo (Ex Libris)
The market leader. Primo works because it\'s been around forever and integrates with Alma (their ILS). Search results are solid, faceting is decent. Problem: you're paying $40K+ annually for a discovery layer, and your patrons will use Google instead.
Worth it if You're already deep in the Alma ecosystem and have the budget. Your patrons actually use it.
Vendor lock-in is the entire business model. Your data lives in their cloud. Switching means migrating everything.
Summon (ProQuest)
ProQuest\'s competitor to Primo. Similar pricing, similar integration headaches. Works fine technically. But ProQuest\'s acquisition strategy means you're paying for their portfolio bloat. They own everything now.
Careful ProQuest keeps buying smaller vendors. Your contract might include forced upgrades to tools you don't want.
Their analytics are opaque. You won't know which modules are actually being used until you audit it yourself.
VuFind (Open Source)
Free, open-source discovery platform. Works great if you have IT staff who can maintain it. Indexing is flexible, you control the data. The catch: you need developers. Not for everyone, but for librarians who can code or hire developers, it's a massive cost savings.
Worth it if You have technical capacity. You want data sovereignty. You're tired of vendor pricing.
Requires ongoing maintenance. If your IT team leaves, you might be stuck. Make sure knowledge is documented.
Evergreen
Open-source ILS with integrated discovery. Solid for smaller library systems and consortia. The learning curve is real, but once you're in, you're not paying license fees. Used by public library consortia nationwide.
Worth it if You're in a consortium or smaller system. You can invest in training. You want long-term cost stability.
ILS / Integrated Library Systems
Your ILS is the backbone. Choose wrong and you'll regret it for 10+ years. These decisions matter.
Alma (Ex Libris)
The new standard. Cloud-based, modern API, plays well with other tools. If you're starting fresh, Alma is probably your best bet. But it\'s expensive ($100K+ first year for mid-size libraries) and implementation is complex. Ex Libris knows they're essential now.
Worth it if You have budget, you're starting fresh, you want modern infrastructure.
Careful if You're migrating from an older system. The migration alone can take 18 months and $50K.
They\'ve deprecated features people rely on. Read your contract carefully about what\'s included in your license tier.
Polaris (Innovative Interfaces)
Still used by major library systems. It works. It\'s stable. But it\'s also aging. Innovative has been slow to modernize. You'll spend the same money as Alma but get older technology. Not the best choice for new implementations.
Careful Good for maintaining existing systems. Bad for new deployments. The vendor is still profitable but seems to be coasting on existing contracts.
Sierra (Innovative Interfaces)
Innovative\'s cloud product. Better than Polaris, not as modern as Alma. Some large systems are on it and seem reasonably satisfied. Middle ground, middle pricing. If you're happy with Innovative, this is the upgrade path.
Careful Not much differentiation from Alma. You're paying nearly the same for an older product.
Millennium/Symphony (Innovative)
The legacy system that refuses to die. Thousands of libraries are still running this. If you're on it, don\'t rush to migrate unless you have budget and pain points. If you're evaluating it for new deployment, walk away. This is a dead end.
Skip For new implementations. The vendor has clearly moved resources to Polaris/Sierra.
Data migration out of this system is an absolute nightmare. Lock-in is real. Plan escape routes before you're forced to stay.
Koha (Open Source)
Totally free, open-source ILS. Used by public libraries, academic libraries, even some large systems. The catch: you need IT support. Implementation and customization require developers. But if you have that capacity, you save six figures over 5 years.
Worth it if You have technical staff. You want data sovereignty. You're in a consortium supporting it together.
Skip if You need vendor hand-holding. Your IT team can't dedicate time to this.
The community is smaller than commercial vendors. Finding specialists can be harder. Document everything.
Evergreen (Open Source)
Another open-source option. Stronger in consortia environments (it was built for Georgia\'s PINES system). Good if you're in a consortium or planning to be. Less suitable for standalone systems.
Worth it if You're in or planning a consortium. You want to pool resources for shared infrastructure.
Migration Reality Check
Migrating ILS is one of the most painful things a library can do. Budget 12-18 months, $100K+, and expect your staff to hate you for half of it. Your data is messier than you think. Clean it first or regret it forever.
Big red flag: If a vendor says migration will be "quick and painless," they\'re lying. It won\'t be. Budget accordingly.
E-Resource Management
Managing digital subscriptions is a nightmare. Here\'s who\'s making it worse and who's actually helping.
Serials Solutions / ProQuest Central
Market dominant. They manage most academic library e-resources. Integration with other vendors is their job, so it\'s usually decent. But pricing is opaque and ProQuest uses their market position to upsell aggressively. You\'re not negotiating - you're accepting.
Careful Necessary for academic libraries, but don\'t expect negotiating power. They know you're stuck with them.
Their pricing model is tied to your collection. As your collection grows, so does your bill. There\'s no cap. They know exactly how much you\'ll pay.
EBSCO ERM
EBSCO\'s competitor to Serials Solutions. Similar market position, similar dominance. Integration works but feels less seamless. EBSCO has been in the game longer and their systems reflect that (they\'re older).
Careful Another necessary vendor for many libraries. Same lock-in issues as Serials Solutions.
Alma (for ERM)
If you're on Alma for ILS, their ERM module is integrated. That\'s the main selling point. Standalone, it\'s not compelling. Some libraries manage e-resources with spreadsheets and manual processes better than Alma\'s ERM module.
Careful Only consider if you're already on Alma. Even then, audit whether you're actually using it.
LibLynx
Smaller player trying to solve authentication and license management. Works well for what it does. Not a complete ERM but handles access management elegantly. Good if you want to escape one of the big vendor stacks.
Worth it if You want to reduce vendor consolidation. You have technical capacity for integration.
DIY Approach
Some libraries are managing e-resources with internal databases and careful spreadsheet work. It's labor-intensive but gives you control and costs almost nothing. Consider this if you have the staff time.
AI Tools for Libraries
Everyone\'s suddenly talking about AI. Here\'s what actually helps libraries and what's just vendor hype.
ChatGPT (OpenAI)
Cheaper than hiring a consultant. Your patrons are already using it. Staff training on this takes 2 hours, not 2 weeks. For general reference questions, answering patron email, drafting policies - it works. The downside: your data goes to OpenAI. Read their terms carefully before using it with patron information.
Worth it if You're using it for non-patron-data tasks. Your IT and legal team have approved it.
Skip if You can't avoid sending patron data through the system. Privacy concerns are real.
OpenAI\'s usage terms have changed multiple times. They claim they don\'t train on your prompts anymore, but verify with legal before trusting that.
Copilot / Claude / Other LLMs
Alternative to ChatGPT. Claude is more careful about content policy (better for libraries). Copilot is fine if you're in the Microsoft ecosystem. Pick one and use it the same way - reference desk support, policy drafting, staff training material generation.
Worth it if You want alternatives to ChatGPT. You're comfortable with different vendor terms.
Library-Specific AI Tools
Several startups are building "AI for libraries." Most are solving problems that don't exist at scale yet. Automated cataloging, patron recommendation engines, smart collection development - the technology is real but adoption is tiny. Pricing is startup pricing (high, with promises of ROI).
Skip for now The space is too new. Let others pay the early-adopter tax.
Startups die. Your data goes with them or gets sold. Ask about data ownership and exit clauses before signing anything.
Automated Subject Heading Assignment
Multiple vendors now offer AI-based cataloging assistance. Accuracy on subject headings is around 26-30% according to recent research. That\'s not good enough to trust fully. Best used as a suggestion layer that librarians review. Don\'t let this replace human catalogers unless you want a degraded catalog.
Careful Use as augmentation, not replacement. Always have a librarian review.
These tools often can't distinguish between similar concepts or handle interdisciplinary materials well. Your special collections and niche materials will break it.
Patron Recommendation Engines
"AI-powered recommendations" sound good. In practice, most libraries don\'t see meaningful adoption. Patrons use library discovery systems inconsistently. Recommendation algorithms need data to work. If you don\'t have good usage data, recommendations will be generic. Test this rigorously before paying for it.
Careful Demand proof of ROI. Don't buy based on demo hype.
The AI Reality for Libraries
AI isn\'t transforming libraries yet. It\'s reducing some specific labor (writing emails, drafting forms) but not solving the core problems (understaffing, underfunding, vendor lock-in). Be skeptical of vendors who claim AI is your salvation. It's not.
Staff Training Platforms
Vendor Training (Ex Libris, ProQuest, Innovative)
Most library vendors include training as part of their contract. It\'s usually available - webinars, documentation, online courses. Quality varies wildly. Some vendors" training is excellent. Others is useless. You'll learn a lot from your users and your IT team, not from the vendor.
Worth it if It's included in your contract anyway. Use it as supplement, not primary training.
LinkedIn Learning
Cheap, broad library and tech training. Your institution probably has a subscription. Not specific to your system, but good for general professional development. Staff can learn at their own pace.
Worth it if Your institution already has access. It's fine for general training.
ALA Training
The American Library Association offers webinars and courses. Variable quality. Some are excellent, some are mediocre. Pricey for what you get. Good for specific professional topics but not for system-specific training.
Careful Check the specific course before paying. Read reviews from other libraries.
DIY Training
Build your own training for your systems. Use screen recordings, documentation, internal wikis. Takes time upfront but perfect for your context and much cheaper. Your knowledgeable staff can train newer staff. This creates institutional knowledge.
Worth it if You have a trainer/coordinator role. Long-term cost savings are huge.
If your trainer leaves, knowledge walks out the door. Document everything.
Emerging Tech: Hype vs. Reality
AI Cataloging
The promise: AI reads your books and automatically creates catalog records. The reality: AI can extract basic metadata (title, author) from covers and title pages. Subject analysis is weak. Authority control is messy. Not ready for production use at scale.
Wait The technology is 2-3 years away from being actually useful. Don't pay for early versions.
Predictive Analytics for Collection Development
"Our AI will tell you what to buy based on patron behavior." Good idea, hard to execute. Most libraries don't have consistent, clean usage data. The model is only as good as your data. Most implementations produce generic recommendations (more popular fiction, more audiobooks).
Wait Not mature enough. Your collection development librarian probably outthinks the algorithm anyway.
Automated Deaccessioning
AI that tells you which books to remove. Sounds useful if you're drowning in collection debt. Reality: algorithms can\'t understand your community context. That unused book about local history matters differently than unused fiction. Let humans make these calls.
Skip This requires human judgment about community value. AI can't understand that.
Natural Language Search for Catalogs
Ask your catalog a question like you're talking to a person. "Books about climate change for teenagers." This actually works better than faceted search in some cases. Several vendors are adding this. Early implementations are promising but still need refinement.
Watch this one Not ready to pay for yet, but worth testing when free/cheap beta versions appear.
Automated Holdings Verification
AI that checks your physical collection against your catalog without manual scanning. Could save huge amounts of staff time if it works. Still early, but worth watching. Image recognition + AI could make this real in 2-3 years.
Watch this one Not ready yet but solve a real problem.
Detailed Setup Tutorials
If you're considering open-source solutions, here are step-by-step implementation guides:
Koha ILS Setup for Small Libraries
Step-by-step walkthrough of installing and configuring Koha. Prerequisites, installation, initial setup, MARC record import, and troubleshooting.
Read guide →VuFind Discovery Layer Setup
Self-hosted discovery platform as an alternative to vendor solutions. Installation on Ubuntu/Debian, index configuration, search facet customization, and ILS integration.
Read guide →Evergreen for Consortium ILS
Shared catalog setup for multi-library systems. Architecture, multi-library configuration, governance, member onboarding, and scaling considerations.
Read guide →The Bottom Line
For ILS: If you're starting fresh, Alma or Koha. If you're already somewhere else, stay put unless there's a specific pain point. Migrations cost too much to do recreationally.
For Discovery: Don't overspend. Patrons will use Google. A decent discovery system is table stakes, but paying six figures for marginal improvements is wasteful.
For ERM: You're stuck with a big vendor. Negotiate hard on pricing because your data is locked in. Consider partial DIY approaches where you can.
For AI: Use commercial LLMs for staff productivity (ChatGPT, Claude). Skip library-specific AI tools unless you have a specific, painful problem. Most AI in libraries is still hype. Wait 2-3 years for real solutions.
For everything: Read your contract. Know your exit clauses. Don\'t let vendors lock you in without knowing the cost. And remember: you're not buying technology - you're buying a partnership. Make sure it\'s a partnership worth the cost.
Questions about your tech stack? Let\'s talk /consulting.html or get in touch. I can help you evaluate vendors, plan migrations, or audit what you're paying for.