Start with the moment someone raised a hand
A request-for-information form is a small but useful moment in a student's journey. Someone has taken the time to ask for more. A confirmation can acknowledge that action, a later message can offer another way to explore, and a further check-in may make sense if no new engagement is visible. The timing of those responses deserves as much attention as the copy.
The fourth Forgeworks Grimoire training begins with that sequence. Its practitioner, the Chronologian, keeps the focus on when a response becomes useful. David sketches a journey map before moving into Slate: life-cycle stage, touchpoint, behavior, timing, contingency, response and the reason for it. That short planning step gives the technical build a question to answer.
In the recording · 06:37 ↗Give every window a purpose
The example uses three possible messages: an immediate confirmation, a curiosity-building follow-up a day or two later, and a re-engagement message after about a week without another tracked action. These are demonstration windows, not a universal schedule. The right interval depends on the action, the next useful step and the other messages already reaching that person.
For each window, write down what starts the clock and what would stop the message. A new event registration, a reply or a completed task may make the planned follow-up unnecessary. The journey map also distinguishes an automated response from a staff conversation and leaves room for no outreach at all.
Choose a useful measure and a reason to pause the experiment. The recording raises clicks and further activity alongside opt-outs and complaints. Establish those checks before a new sequence becomes routine. More messages and more visible activity are not, by themselves, proof that the experience improved.
In the recording · 17:14 ↗Make the next step fit the available context
The confirmation build explores distance from campus and the choice between an in-person and virtual next step. Its query follows the school to an organization address and also considers a person's permanent address. A coalesce pattern chooses an available value when the preferred source is absent. Names on the joins help keep those two addresses distinct.
The demonstration uses placeholder coordinates and provisional thresholds. Replace them with the intended campus location and an explicit, tested rule before drawing conclusions from a distance export. Check the no-address case too. Missing information needs a considered fallback, rather than a label that accidentally claims to know something about the student.
The episode also explores community-level ACS data. An area's income or educational profile cannot establish an individual student's finances, goals or ability to travel. Treat that discussion as a prompt to offer accessible choices, not a basis for assuming a personal circumstance or restricting an opportunity. A student's stated preference is a better reason to tailor the next step.
In the recording · 19:22 ↗Keep content choices readable
The next example compares an RFI submission with the opening of an application cycle. An export supplies a value that selects a content block, such as an invitation to share more information or explore a virtual tour. This separates the selection rule from the reusable button or content it displays.
That separation makes maintenance easier to reason about. Someone can review what a content block says without changing the timing test, or inspect the export without rebuilding the email. Use clear names, confirm the expected keys exist and test both sides of the boundary. The recording uses a fixed demonstration date, so a live implementation also needs an owner for each new cycle's dates.
In the recording · 42:26 ↗Silence needs a careful definition
For re-engagement, the question is whether activity happened after the RFI submission. The live query reaches a useful unfinished edge: a ping timestamp has not been sorted as intended, and David identifies the need for existence logic to exclude people with later activity. He does not rebuild that branch before moving on.
Finish that logic before using it to select recipients. Test a person with no recorded ping, a ping before submission and a ping after submission. If multiple pings exist, make sure the result answers the actual question rather than depending on whichever row happens to be returned. Then consider other signals your team trusts enough to suppress the message.
No tracked activity describes the data you can see. It does not tell you that someone has lost interest, misunderstood a letter or stopped considering an institution. Keep the response modest and useful, and make waiting a legitimate outcome.
In the recording · 46:01 ↗Try one window before rebuilding the journey
The Forge Trial closes with a manageable assignment: choose one action, define its window, build the eligibility rule and decide the level of response. Use fictional cases first. Within your approved data practices, a historical check can then show who would have qualified and who would have been suppressed.
Look closely at the surprising cases. Ask whether the timing was reasonable, whether the chosen signal was sufficient and whether another outreach already covered the need. The aim is a small rule the team can explain, maintain and revise. That is a stronger foundation for adaptive engagement than a longer calendar of automatic reminders.
In the recording · 1:00:21 ↗