Salesforce Flow Loops: Create a Record for Every Date or Shift

Published October 29, 2026 · MultiDatePick

A recruiter needs to add twelve shifts to a job. A manager approves a week off. An event planner books a three-day workshop and a follow-up day. Each one is a single decision on a single screen, and each one should become several records.

Screen Flow has no element that says "create one record per date". You build it from three standard pieces: a loop, an assignment and one save. This post walks through that pattern and gives you three working flows built on it.

The pattern comes from a recruiting build on TalentRover, where recruiters had to add a run of shifts to a job. Entering them one record at a time took longer than the scheduling itself. Picking the days and hours once and letting a loop write the shifts fixed that, and it works the same for any object.

The pattern in five elements

Why the save goes after the loop

It is tempting to drop a Create Records inside the loop: one pass, one record. It works for three dates and breaks for thirty. Every Create Records is a separate database save, and Salesforce allows 150 of them in a transaction, shared with every trigger and flow those saves set off. A save inside the loop spends one per date.

Building a collection and saving it once spends one save in total, whatever the user picked. It is also faster, and it either saves everything or nothing, so a failure never leaves half the shifts behind.

	Create Records inside the loop	Collection, saved after the loop
Saves for 30 dates	30	1
Limit risk	Grows with every date picked	None from the loop
On an error	Some records saved, some not	All or nothing

Use case: add shifts to a job

The recruiter opens the flow from the job, picks the shift days on the calendar and the hours on the time grid, and clicks Next. The flow writes one shift per day, linked to the job.

The calendar hands the picks to the flow as JSON in its selectedDatesWithTimesJson output. Flow cannot read JSON, which is where the parser comes in. Each entry it returns has these fields:

Field	Type	Holds
fromDate	Date	The shift's day
toDate	Date	The same day for a shift; the last day for a range
startTime	Time	The first picked slot
endTime	Time	The end of the last picked slot

Adjacent slots make one shift, so 13:00 and 13:30 become 13:00 to 14:00. Picking two separate blocks on a day, say a morning and an afternoon, gives two entries and two shifts for that day.

Use case: one record for every day off

Time off is usually tracked per day: payroll counts days, and a half-approved request has to be split by day anyway. Here you do not even need the parser. The Date component also outputs selectedDates, a plain text list with every picked day, ranges spelled out day by day.

Loop through that list directly. The only catch is that each item is text such as 2026-11-09, so a one-line formula, DATEVALUE({!Loop_Days}), turns it into a date before the assignment copies it onto the record.

Use case: one record per date range

Some things span days but are one thing: a three-day workshop, an equipment rental, a site visit that runs Monday to Wednesday. Those want one record with a start date and an end date, not three.

That is the selectedDateRangesJson output. The calendar joins adjacent days into a single range, so picking the 9th, 10th and 11th gives one entry from the 9th to the 11th. Parse it, loop, and copy fromDate and toDate onto the start and end fields.

Use a starter flow in your org

The starter flows save to the Schedule and Event objects that come with MultiDatePick, so they deploy as they are and run straight after install. Swap in your own objects once you have seen them work.

1. Copy or download the flow's XML above.
2. Deploy it from VS Code. Deploy a Flow with VS Code walks through it step by step.
3. Open it in Flow Builder and click the calendar on the first screen. Type the Config Name of a config you saved in the Setup Wizard, or one you imported from the Config Library, to give it your settings.
4. Point it at your objects, if you are not using Schedule and Event: change the record variable's object and the fields in the assignment.
5. Activate it and add it to the record page with the Flow component or as an action. It reads the record's Id from its recordId variable.

Mistakes that cost an afternoon

• Saving inside the loop. Fine in a test with three dates, a limit error in production with forty.
• Forgetting to add the record to the collection. The assignment needs two parts: set the fields, then Add the record variable to the collection. Without the second, the save at the end has nothing to save.
• Assigning text to a date field. The selectedDates list is text. Wrap each item in DATEVALUE() first.
• Leaving Output As JSON off. The JSON outputs are only filled when the calendar's Output As JSON setting is on. The starter flows turn it on.
• Launching from the wrong place. The flow links each record to recordId. Run it from the job, event or person the records belong to, or the link stays empty.

Questions people ask