Creating websites that foster patron engagement, enhance civic pride, and position public libraries as vital cultural hubs

Preserving privacy in libraries is essential for safeguarding freedom of inquiry, and we rely on tools and systems that prioritize visitor privacy and data protection. Good librarianship has us dismantling Google Analytics, Meta Pixels and sales tracking, and instead using open source tools where the data is kept safe and only used by the library

Privacy is a vital tenet for libraries, and Article VII of the ALA’s Library Bill of Rights protects true freedom of inquiry. ALA’s guidelines say that “when users recognize or fear that their privacy or confidentiality is compromised,” that freedom of inquiry is lost.

Libraries need to collect visitor data to see usage trends, make cases for funding, quantify partnerships, and report numbers to boards, staff and municipalities. Tracking usage is technically easy, but leaves patron data identifiable, and the numbers can be wildly off as it’s difficult to detect which traffic is bot-driven and what is human. So libraries usually rely on vendors for analytics, most of the time from Google Analytics, which is free and makes the tracking data non-identifiable.

But then we have a Google problem. In the last year alone, California, Arizona, Illinois and then a 40-state coalition have reached class action settlements with Google on direct violations of people’s privacy. As the world understands today, “if you’re not paying for a product then you’re the product”, and Google’s cash cow is identifying people and showing them ads. Even if Google isn’t being evil with the library’s data, people perceive it as a privacy failure, and that falls within the “fear that their privacy…is compromised” from the ADA guidelines.

Libraries have had issues with vendors intentionally or accidentally giving identifiable information away. In 2014, some libraries were using an Adobe ebook lending product and researchers found Adobe had collected and sent identifiable and personal information, fully unencrypted, which Adobe tried to explain away as license requirements. This becomes a problem when anyone in a coffee shop could scan the WIFI and find exactly what somebody was reading. Should this person be a bad actor, this has some real world implications for the reader, be it stalking or phishing, etc. The stakes go up when the reader is a journalist or community leader, and the bad actor can use library data to silence her voice.

For libraries that use vendors, it’s crucial for libraries to be fully aware of the information vendors collect, and audit that the information isn’t being misused. The better solution is to use on-site or self-hosted, open source analytics suites like Matamo or Umami, or paid providers who have clear privacy policies and use aggregated data. These involve an investment to set up and fine tune, and may need staff time to interpret the analytics these tools provide. The advantage is that we’ll have a library-owned system that tracks better information and truly respects the visitor’s privacy.

The lack of universal search on websites, even modern ones, limits the ability to connect users with valuable library resources beyond the catalog. Despite technical challenges and institutional changes required, universal search should be a primary focus for new library websites to enhance user experience and expand access to diverse information.

It’s very common for websites, even ones built today, to have a pattern where an online patron needs to choose where they want to search, and often defaulting to the catalog. While the catalog is the primary use case, we’re missing so many opportunities to connect people with the library by sending them, by default, off to the OPEC and never to return.

The other choices for search vary by library, but often includes the site, collections and sometimes even events. If someone looks up “Judy Bloom” or “Goodnight Moon” or “science fiction”, wouldn’t they also want to know about the teen events, story times, or science fiction book clubs? Wouldn’t they also want to see book lists that cover coming of age or bedtime or cyberpunk novels? Not only would the patron love this, it fits into the promotional efforts of any library’s marketing efforts.

So if both patrons and the library want it, why isn’t universal search common? The first reason is that it’s hard. Some ILS vendors don’t have API capabilities. Some library systems are afraid ILS data would include patron details and therefore break privacy. It’s also technically difficult to actually collect that information and then make sense of it all. Ranking systems that work for people have to be created, and search has to be made fast and include features like autocomplete and spell checking.

The second reason is that it’s an institutional change that requires institutional investment. It requires working with vendors, testing the data, re-thinking patron assumptions and tinkering with the algorithms. And universal search takes a commitment to make it better over time, to find search terms that areoutdated, to test with patrons and to sift through analytics to find meaning.

When we first saw a library’s universal search go live a decade ago, we saw it as a great feat and one of the biggest aha moments in online libraries. But it’s been a decade, and despite the technical and human roadblocks, universal search should really be the first order of business on a new website

Byte has been on the forefront of the open data / licensing movements, is a founding member of the Open Data Initiative, and has presented open data concepts to both tech groups and civic organizations, including co-presenting with American Civil Liberties Union’s national senior policy advisor on technology at the University of Wisconsin - Milwaukee.

The Walters Art Museum in Baltimore holds a plethora of collections data, including a growing digitization of their world-renowned manuscript collection. We helped show them how their open licensing and data could create a manuscript exploration website that went on to win a gold MUSE award and was presented at the Best in Heritage conference in Dubrovnik.

We also helped the University of Pennsylvania use many other institutions' manuscript data, including The Walters’, to build a new manuscript portal that included a quire explorer that helps scholars, researchers and artists understand the inner workings of manuscript binding.

We help organizations manage connections between open databases and their own databases. For example, we helped Mount Vernon create their ARGO Revolutionary War maps exploration tool by mixing open data and their own custom curatorial data to create better storytelling and curations, and include subjects and other curatorial tagging that was beyond the scope of the original open database.

When possible, we’ll both recommend how a client’s data could be useful and help them set up a repository of their own open data. We also help with integrations across many sets of data, open and closed, and have that data be useful in ways that extend the website’s functionalities and further our clients missions.

We’ve also created our own content management system with integral features around open data and licensing. For example we built our document asset management tool that takes the place of a standard file management system that includes open licensing as part of the storehouse and the search features.

In an era where websites are often designed for short-term use and quick turnover, we take a different approach by providing indefinite support to our clients, prioritizing long-term management, growth, and data extensibility through simple, modular tables, targeted frameworks, and proven architectural advantages.

Tim Berners-Lee helped create the web using open standards, and his code, HTML, also made the internet accessible. The HTML code itself has semantic meaning, thus a navigation, an image and caption, or an article all have their own tags. That way a screen reader, search engine or web browser can understand its inherent meaning.

A missed opportunity

Unfortunately, despite widely available standards, most sites are not accessible. It takes developer expertise and content mitigation training, and most website managers are fine with leaving what they assume is a very small population behind. Yet, accessibility needs are not rare at all, around 25% of people have some sort of need at any giventime. Accessibility is also situational and includes people holding a baby in one arm, people who can’t find their glasses, or are drunk, just waking up, or simply distracted. When accessibility is done right on a website, everyone benefits.

Snake Oil

Website managers are starting to understand the value of accessibility, so shady companies are selling the idea that they can fill the gap with “software overlays” that “solve” accessibility. But that doesn’t work. Accessibility needs to be baked into the fabric of the design and code, and into the practice of content management. These overlay companies have been called out as snake oil by librarians and accessibility advocates as being a net negative on actual accessibility.

An investment over time

Of course public libraries are built on open standards and accessibility, and their websites have to go the extra mile, but accessibility isn’t a binary and there is no panacea. A website can always be built to be more accessible, but any inherited content like blogs or events will take time to mitigate, like adding alt tags and making headlines and other content work in an accessible way. We create or find content management tools that help train administrators and contributors practice accessibility, help identify and fix issues, and see their progress over time. Accessibility is a team effort

Our features include

• Keyboard & Screen Reader Navigation: features that enable individuals to navigate and interact with websites using only the keyboard or screen reader software.
• WAI-ARIA tags: Backup plans for when the semantics don’t work in a situation
• Alternative Text: Alt tags are used to explain what is happening or what an image represents. That gives more information to search engines (SEO pretty much requires alt tags).But more importantly, it helps users that require a screen reading digest your content.
• Color Contrast: for text readability
• Dark Mode: reduces amount of light emitted from your screen and is extremely helpful for low-light situations for users and their environment
• Highlight Links: for users with poor vision, this feature adds additional styles which offer a better opportunity for perceiving hyperlinks.
• Font Sizes: Larger text size available for those who need larger letters
• Desaturate: Color can not be the only means of conveying information

In an era where websites are often designed for short-term use and quick turnover, we take a different approach by providing indefinite support to our clients, prioritizing long-term management, growth, and data extensibility through simple, modular tables, targeted frameworks, and proven architectural advantages.

It’s common for companies to build websites to have a 3-5 year life span, and not build for long term management, growth or data extensibility. The case for planned obsolescence is that there is too much entropy in tech and things need to be rebuilt often anyway. Also, more websites means more repeated business.

We have a different relationship with our clients: we support our work indefinitely. This gives us a different business perspective, one where no matter what happens with a site, we’re going to fix it.

In data, it’s common for website builders to merge various types of data into one grand data source, merging schema and relationships into settings and letting the software slice it in ways that can be displayed. Wordpress famously considers all data as basically one data table which is basically impossible to read as a human. When we create a site, we build all the data sources in discrete, simple, modular tables with clear normalization patterns. This makes reading it easy without special tools, and so adding features later on are simpler.

When it comes to code, many developers use the most modern languages, frameworks and practices possible, so-called “green field development.” Not long ago everything was object-oriented, then came object-relational-modeling, then functional programming, etc. More recently there’s the continuous integration and “dev-ops” deployment cycles, and build cycles that compress and “transpile” readable code into unreadable compiled code.

When it comes to our code, we keep things simple and architecture-specific. Every abstraction, every layer, every practice creates a multiplication effect that the future developer or owner will have to sift through. Instead of “modern” practices and frameworks, we use targeted practices and frameworks that are built for the environment the website is going to live in. We need to be able to make direct changes without a complete rebuild of the site. We won’t use frameworks for back-end programming, and we’ll use non-invasive frameworks for the front end like Vue, versus React or Angular.

We use modern processes all the time, but only when there’s a proven architectural advantage. For example, websites are great data providers, and using “JAMstack” is the natural way web servers and browsers work. Websites are also great at holding content, so the modern practice of content publishing makes sense versus content management.

Our process, from content to design to our content management system, is centered around patron-needs.

A library website is for all its patrons, so the right mix of information and content is important. One common technique in modern design is to wildly simplify the content and information. We find that for a library, it often goes too far, and the trend of oversimplification and a preference for “lowest common denominator” aesthetics can actually work against good UX and messaging goals.

An oversimplified site intentionality leaves visitors with less information, but leaves people thinking there’s little worth exploring. A site that isn’t interesting enough to foster engagement won’t accurately show the value of the library. We call this an anti-pattern as this works against the mission.

A case in point: this image is an example of a new materials list on a library home page:

While the design is quite compelling, the visitor is missing details that would improve the experience. The visitor won’t see the media format, recommended age groups, what branch they’d be found at, or if they’re available. The lack of information puts the weight on the four books themselves, which makes it seem like a librarian recommendation when they’re not recommended at all. Oversimplified sites promote less of what The visitor won’t see the media format, recommended age groups, what branch they’d be found at, or if they’re available. The lack of information puts the weight on the four books themselves, which makes it seem like a librarian recommendation when they’re not recommended at all. Oversimplified sites promote less of what the library wants to share, despite how pleasing it is.

Oversimplified design also takes away from the uniqueness of the library. The current design trend is to pare down everything and then apply a thin brand veneer, but we find it tells the patrons that the library is generic and bland, and that’s not the library we know. When between generic and interesting, let’s err on the side of interesting.

There’s an accessibility case to be made for oversimplification. Some people are going to have a hard time with too much information. Since we want everyone to feel welcome, we can also build in a ‘zen mode’ that’s a lot less dense. We would build the choice into an accessibility or preference panel so people can choose their level of density, like “compact”, “cozy” or “comfortable”. That allows the best of both worlds, low- and higher-density designs.

Our best work is where we show the library’s value to its citizens and expand library use. There are plenty of people who think the library is a pre-Google relic and others who think there’s nothing for them at the library (and still others who think the library is political). Better design and clearer communication of the library’s value is the way to change hearts and minds, and that’s the biggest challenge we invite.

While this is our treatise on information density, each website we create will be based on patron needs, library objectives and patron testing. We want to help visitors tap into their curiosity and find the unexpected, and also ensure everyone can find a comfortable and clear way to explore their library.

Search systems are continuously being molded as we’re all learning, building and testing ways to better serve site visitors. Rather than acquiesce to the search systems that exist in today's world, we pose libraries to think of the future possibilities of search and be a part of shaping it.

When visitors are looking for something in the library, they often end up talking to a librarian who helps figure out the need behind the query, corrects assumptions, and formats a context around what’s available and where it is. Yet, search on most library websites is surprisingly limited – often separating the catalog entirely from the site results (See: Universal Search) For site searches, the results are a long list of text matches based on keywords more than meaning. These often make the searcher feel that there are no matches, even if there’s a shiny needle deep in the haystack.

We need to go beyond just “search” as it’s not the right pattern for a visitor of a large site. Instead, we need to employ a “search and discovery” pattern to help the visitor make sense of all the potential results.

What can a search and discovery system look like?

• Merging the catalog, website and other data into one federated search
• Auto-complete common searches
• Correct spelling errors
• Include “did you mean…” or “people also search for…” nearby searches
• Group results by result type, like events, blog posts, etc.
• Group catalog items by media or availability
• Employ common site tags to help illustrate or group results beyond type
• Include catalog item details, like branches, availability, internal mapping
• Make clear visual distinctions between types
• Place a “direct hit” above searches with an obvious answer, like “hours” or “branches”
• Use site tags with deeper subjects to include starting places or common searches or subjects, akin to finding aids

An effective search and discovery platform is not a product that one can just plug into a site. It requires an investment in the technology, and expertise from the technologists. It requires site administrators to invest time and domain expertise to help manage and update the data and algorithms behind search. It’s a system that needs to be studied, understood and updated over time.

Search is still a nascent technology. Google and Bing only recently added knowledge graph and other discovery pieces in the last few years. Search is continuing to be molded as we’re all learning, building and testing ways to better serve site visitors. When we work with libraries, we don’t want to use the patterns of today, our job is to help libraries imagine the possibilities of search and discovery of tomorrow