Using A/B testing to improve library email click-through rates

Email remains one of the most direct ways for Australian libraries to reach patrons with timely information. A well-designed newsletter can alert readers to new books, digital resources, workshops, exhibitions and local events, while a cluttered or poorly targeted message can disappear unnoticed in a crowded inbox. Testing different versions of that message helps library teams make evidence-led improvements without relying on guesswork.

Public library services in Australia range from large metropolitan networks in Sydney and Melbourne to locally operated branches in regional towns. Patrons may receive several library emails each week alongside council updates, school communications and retail offers. Relevance matters: an email that reflects a reader’s branch, interests and current activities is more likely to earn attention and produce meaningful clicks than a generic campaign sent to every subscriber.

Establish a measurable baseline

Begin by defining what the email is intended to achieve. A promotion for a children’s storytelling session might aim for event bookings, while a collection update could encourage catalogue searches. The primary click-through rate is calculated by dividing the number of unique recipients who clicked by the number of emails delivered, then multiplying by 100. Record the source, campaign date, audience segment and tracked links so results remain comparable.

Establish a baseline from several recent campaigns rather than a single send. Look at delivery, clicks, unsubscribes and downstream actions such as reservations or event registrations. Delivery problems can depress click-through rates just as surely as weak wording, so confirm that addresses are valid and that recent messages reached the inbox instead of the spam folder.

Decide which metric is the main measure before testing begins. Click-through rate is useful for comparing creative approaches, but a high click rate is not valuable if nobody books a place or opens a database. Keep a small group of supporting measures, including conversions, unsubscribes and spam complaints, to reveal whether a change benefits the library as a whole.

Design variants around patron behaviour

Segmentation usually produces stronger results than trying to optimise one message for an entire database. Separate audiences by branch, membership type, age-related programs, digital-resource use or expressed interests. For example, a computer skills session in Parramatta may appeal to one audience, while a writing group in Geelong needs different language and imagery. Behavioural data should be collected proportionately and used to improve relevance, not to make assumptions about people’s backgrounds.

In a large city, many patrons read email on a phone during a train, tram or bus journey, or between household tasks. Australian library websites and newsletters should therefore present the main action early, keep buttons large and use concise copy. On a hot afternoon in Adelaide, an invitation to free indoor activities may also perform better if the email reaches a subscriber before they make other weekend plans. Local weather and public holidays can influence timing, though they should be considered alongside each audience’s usual reading habits.

Subject lines, preview text and calls to action can each form part of a test. “Reserve your place” gives a clearer expectation than “Find out more,” while preview text can explain what appears after the subject line. Before preparing variants, the Library Marketing Toolkit’s newsletter design guide offers useful guidance on readable structure and mobile presentation. Change one element at a time so the result reveals what caused the improvement.

Run a disciplined split test

Write a specific hypothesis for every comparison. For instance: “A subject line naming the event location will generate more click-throughs than one referring only to the event.” State the audience, version A, version B, primary measure and expected effect before sending. This prevents the team from changing subjects, artwork, links and timing together and then crediting the wrong feature.

Randomly divide eligible recipients into two groups with similar proportions. Each person should receive only one version during the test, and both versions should use the same sender name, send time, tracking method and content apart from the selected variable. Remove staff accounts and duplicate addresses, and avoid repeatedly sending a winning variant to the same recipient during the measurement period. Platforms such as Mailchimp, Campaign Monitor or the mailing tools provided by some library management systems can handle random assignment, although small teams can create the groups themselves.

Allow the test to run long enough for each version to receive a fair opportunity. At least one full Australian business day is often more realistic than checking after an hour, especially for audiences spread across eastern, central and western time zones. Avoid stopping early because one group appears to be ahead. With a modest local subscriber list, a small percentage difference may be random variation; document the uncertainty instead of declaring a winner based on one click.

Read the results beyond the click

Compare delivery, click-through and conversion rates across the two groups. A result is persuasive when the difference is large enough to matter operationally and does not coincide with a rise in unsubscribes or complaints. Statistical tools can estimate confidence, but the practical question remains: would the expected increase justify changing the standard template? A modest gain may be worthwhile across a large statewide campaign yet too small to justify extra work for a small branch list.

Treat open rates as supporting information rather than a precise measure of attention. Image previews, security software and privacy features in major email platforms can create recorded opens that no longer represent a person reading the subject line. Unique clicks generally provide stronger evidence because they require an intentional interaction, although tracking pixels, antivirus scanners and repeated automated requests can distort the figures. Check click patterns for unusual bursts or links that receive implausibly high traffic.

Follow the journey after the click. If the message promotes a digital audiobook service, count logins or loans as well as button clicks. If it advertises an exhibition, track bookings, referrals and attendance where privacy and operational systems permit. This deeper view helps distinguish an attractive subject line that attracts the wrong people from a useful message that reaches patrons ready to act.

Build a repeatable testing habit

Turn the strongest result into a documented pattern rather than a one-off success. Save the subject line, artwork, preview text, audience definition and sending conditions in a simple test log. Note unusual events such as a citywide public holiday, a major book release or a temporary closure. Over time, this history will show which combinations work for different library audiences and will prevent the team from repeating tests that have already been answered.

Create a regular cycle that suits the library’s capacity. A small team might test one subject line every second month, while a larger metropolitan service could run separate tests for branches, member groups or campaign types. Review results at a scheduled meeting and assign one action from each experiment: adopt the variation, repeat it with a different audience, or reject it and document why. This approach turns email marketing into an iterative service rather than a series of disconnected campaigns.

Protect trust throughout the process. For promotional sends, follow the Spam Act 2003, including appropriate consent, clear sender information and a simple unsubscribe route. Do not purchase or scrape patron lists. The Privacy Act 1988 and applicable state, council or organisational policies also shape how personal data may be handled, with requirements differing between government and private organisations. Campaigns built with local bookshops, community groups, makerspaces or other Australian partners should preserve the same standards, regardless of who distributes the email.

Choose one recurring library email and test a single element today. Create two clearly defined versions, identify the action you want to improve, and schedule a fair comparison using your existing mailing platform. Review the result after the planned period, record the learning and apply the stronger version to the next campaign.