Shiftto

Shiftto Usage Data: A Detailed Look at Shifts Across Multiple Events

Published: October 17, 2026

I aggregated Shiftto's production data to put numbers on what kinds of shifts are actually being made. The count covers events that have gathered a fair number of answers: scale, the share of availability marks, assignments and answer edits. So that no single event can be identified, the numbers are rounded.

What was counted and how

The target is events that were not deleted and had gathered a fair number of respondents. Events with few respondents were excluded because they may include trial uses such as checking how it works.

Event names, participant names and the contents of answers were not used in the count. So that no single event can be identified, numbers are rounded or shown as ranges, and the tables are not in order of creation. Totals are approximate.

  • Respondents: the number of people registered as participants. This includes people who were only pre-registered on a roster and have not answered yet.
  • Time slots: the number of slots excluding breaks.
  • Share of marks: the share of ○, △ and × out of the number of answers (people × slots). People who were already assigned are included.
  • Required shifts per person: the total required headcount (the required headcount of every slot × role added up) divided by the number of respondents.
  • Assignments: the number of saved assignments. “Auto” were placed by automatic assignment, and “manual” were placed by hand by the organizer.

Scale

ItemRangeMedian
RespondentsAbout 10 to 30About 10
Time slots5 to the 30sAbout 10
Roles1 to 63
Days1 to 32

Slots were mostly 30, 60 or 120 minutes, and one event mixed 60 and 90 minutes. Most events had a dozen or so respondents.

Required headcount

Required shifts per person ranged from just under 1 to about 20, with a median of about 7. The total required headcount per slot ranged from 2 to just under 20. In most events, every slot needed the same number.

Availability marks

Across the events, ○ was about 50%, △ about 15% and × about 35%.

The share varies a lot between events. ○ ranged from the 30s to the 90s of percent, and × from under 10% to the 60s. About half of the events had almost no △.

Differences by scale

So that no single event can be identified, I do not show a per-event table and summarize only the trends.

  • Every event where ○ exceeded 70% had 12 or fewer respondents. In events with 21 or more respondents, ○ was between the 30s and 50s of percent.
  • Events with a high share of × (the 60s of percent) included both small and large ones, so no link to scale was seen.
  • Events built with automatic assignment only had 20 slots or fewer. Events built manually only had 21 slots or more and three or more roles.
  • In some events, some people have not answered every slot.

Assignments

Assignments were saved in about 70% of the events. Automatic and manual assignments were roughly half and half by count. By event, there were automatic-only, manual-only and mixed ones, and automatic-only was the most common. Some events also had pinned assignments.

In the events with assignments, the average number of assignments per person ranged from just under 1 to 4 or 5. The maximum for one person ranged from a few to about 20.

The default limit on consecutive slots is 3. In the events built with automatic assignment only, the longest run of consecutive slots for one person was between 1 and 3. Some manual-only events had runs of 10 or more.

Slots fell short of the required headcount in about half of the events with assignments. The shortfall (total required headcount minus the number of assignments) was a few people at the low end and around 100 at the high end.

Answer edits and roster pre-registration

The change history records only edits from the second submission on. A person's first answer is not recorded.

KindEdited byCount
Answer editThe personFew
Answer editThe organizerAround 10 times the person's
Roster pre-registrationThe organizerMany

The times people fixed their own answers were concentrated in a few events. Per respondent, that is well under 10%. Roster pre-registration was used in more than half of the events.

Limits of these numbers

  • The sample is a limited number of events. Please read the numbers as applying to that range.
  • Events with few respondents are left out. I chose that cutoff myself.
  • The numbers are as of the time of counting. Events that are still collecting will gain more answers and assignments.
  • The numbers are rounded so that individual events cannot be identified. Small differences cannot be read from them.
  • Attributes of users, such as school, region or type of group, are not recorded, so they are not counted.

My own impressions in light of these numbers are in the article on why I built Shiftto.

Related articles

See all articles