Give each row a clear job
A portal can look simple while hiding a maintenance problem. Today's visits, tomorrow's visits and upcoming interviews may each depend on a separate query. Add a visit type or change an export, and the same repair has to travel through several places.
In this Visiting Artisan conversation, Annie Lehwald demonstrates a more manageable approach: retrieve the visits once, export the distinctions the interface needs, and let Liquid markup organize the results. The first decision happens before the markup. Her query starts from form responses because a person can register for more than one visit. Each registration needs its own row.
That choice is worth checking in any similar build. Write down what one row represents, then test a person with two registrations. If one disappears, a polished view cannot repair the underlying grain.
In the recording · 14:07 ↗Separate the shared data from the view
The visit query combines two useful signals: when a visit happens and what kind of visit it is. Case statements turn those details into compact labels. The view assigns a group from the query results, checks whether that group has records, and loops through the records it should display.
This creates a useful division of work. The query owns shared eligibility and exports; each view owns its presentation. Adding a template still requires deliberate maintenance, but the same eligibility change no longer needs to be repeated across a collection of near-identical queries.
Tomorrow also deserves its own question. A visit can be both upcoming and tomorrow, so Annie adds a separate tomorrow-type export rather than forcing every useful view into one exclusive classification. When two categories overlap naturally, separate flags can make the intent easier to follow.
Test those classifications at the date boundary. Today's visits should not disappear because a broad filter accidentally excludes today, and a new template should not silently fall outside the type mapping. An empty-state message is helpful only after the underlying selection is correct.
In the recording · 15:25 ↗Send one message with the right relationships
The email demonstration changes the unit of communication. A parent receives one message, while a dictionary export supplies the related students inside it. Liquid loops through those student details to produce an individual event link for each student.
The distinction matters: the recipient and the person associated with a registration are different roles. The message can remain family-level while each link preserves the intended student context. Check the displayed name and the destination together rather than assuming a correctly labeled link contains the correct identifier.
Existence exports handle simpler decisions, such as whether a family needs a particular application section. Begin with one yes-or-no condition and make its meaning explicit. When the condition depends on a related student, constrain it to the intended term and group so an unrelated record does not turn on the wrong paragraph.
In the recording · 27:15 ↗Offer an event only when the offer makes sense
Annie uses an independent subquery to ask whether suitable events exist at all. That question differs from asking which events a recipient has already registered for. The message needs available programming, even when the person has no event relationship yet.
The demonstration checks the intended template, a future start date and an active or confirmed state. Annie also points out the value of checking space. A registration invitation becomes frustrating when its destination is tentative, already past or full.
Treat availability as a complete promise. Test the case where no qualifying event remains and make sure the invitation disappears cleanly. Keep the rest of the welcome message useful, so the mailing does not need to be stopped simply because the event season has ended.
In the recording · 31:30 ↗Debug the names, then read the message
Small naming mistakes can make a sound design appear broken. Annie describes confusing a loop property's dot notation with an underscore. David describes copying a loop and changing its variable name in only one place. A shared convention for loop names and yes-or-no values gives colleagues fewer moving parts to reconcile.
Run the query with the relevant exports visible before judging the template. Confirm the values and relationships, preview contrasting records, and send a test message. Inspect spacing as well as content: removing a conditional section can leave awkward gaps or push sentences together.
The practical starting point is modest. Choose one repeated view or mailing, identify the decisions that actually vary, and build test records for those differences. Consolidation earns its keep when another person can trace why a row appears, why a paragraph disappears and which record a link belongs to.
In the recording · 44:33 ↗