UserVane
Menu
Product Pricing Agents Docs Calculator Compare Blog Sign in Start free
Start free

Blog

2026-10-20

← Back to blog

Blog

The Monday bucket is not This period.

Tuesday 09:12 the row labeled 2026-10-12 to 2026-10-18 is NPS 33 on 12. This period is NPS 38 on 13, and it is not that row.

Tuesday, 20 October 2026, 09:12 UTC. Mina owns customer experience at Sable, an 18-person product for clinic front desks. The plan is Scale. Survey visit-nps, version 2, source Measured by UserVane only. She opens Advanced analytics. Time bucket is Weekly. From is 2026-07-22 and To is 2026-10-20, the default last 90 days when the page was opened at 09:12. The check-in rewrite shipped Monday, 12 October, at 10:30 UTC. The widget asks when a desk saves the morning list, and it stays dark for seven days after a measured answer.

The row labeled 2026-10-12 to 2026-10-18 reads NPS 33, plus/minus 55, N=12, and the line still stabilizing. The source split says measured 12. The next row, labeled 2026-10-19 to 2026-10-25, reads Insufficient (N=4). The denominator column repeats N=4. When at least two buckets in the range clear ten, the chart draws a bar and the tick under that Monday is 10/12. A bucket under ten is hatched, the word Insufficient, and a short fixed height. It is not drawn as a low score.

The same minute, basic results on version 2 and measured answers shows This period at NPS 38, plus/minus 53, N=13. That card does not say still stabilizing. The cut of that 168-hour window is the headline post. This post is the other clock.

Free and Pro do not get these rows. On those plans the analytics URL is the heading Advanced analytics is a Scale feature, with a link back to basic results. Basic results keeps the headline and the two 7-day cards. It does not keep a Monday table. If that is the plan, use the 7-day cards. The rest of this page is the Scale table.

The weak version, and what it costs at 09:20

The read that produced the slide looks like this:

The week of 12 October is NPS 33. This period is 38. The check-in rewrite held.

The row through 25 October is four answers. A quiet open. Leave it.

33 and 38 are the same week told twice. Park the no-show follow-up.

The read that keeps Tuesday honest looks like this:

2026-10-12 to 2026-10-18 is Monday 00:00 UTC on 12 October through the end of Sunday 18 October. NPS 33, N=12, plus/minus 55, still stabilizing. Jordan's 6 at 08:10 is in it, and the ship was at 10:30.

This period is 13 October 09:12 through 20 October 09:12. NPS 38, N=13, plus/minus 53. Jordan's 3 on Monday afternoon is in it. The Monday row does not have that 3.

The row labeled through 25 October is Insufficient (N=4). One of the four is Jordan saying the no-shows came back. The label names Sunday. Tuesday morning has not reached Sunday.

38 does not replace 33, and 33 does not mean the rewrite held.

The weak version costs the follow-up. The room hears that the week held, because 33 and 38 look close. Jordan's Monday afternoon line never gets an owner. The chart did not draw those four answers as a score, and the slide did not read a hatched column.

Count the sixteen rows

These are the measured version 2 answers from Monday 12 October 00:00 UTC through Tuesday 20 October 09:12. No link answer and no email answer is stored in that span, so the weekly row happens to be in-product. The weekly table has no channel control. A link answer in the same Monday window would sit inside the 33. The results page can still be set to In-product only. That filter is not this table. The channel post is that filter.

Three answers are in the Monday row and outside This period. This period starts at 13 October 09:12, so all three sit in Prior period, which runs back to 6 October 09:12. An answer from before 12 October 00:00 is not in this row.

  • Monday 12 October, 08:10 UTC. Jordan, 6. "Sunday's no-shows are still on Monday's list." This is before the 10:30 ship.
  • Monday 12 October, 11:41 UTC. Sam, 9. "Check-in stuck on the right patient."
  • Tuesday 13 October, 07:55 UTC. Casey, 8. "The list is right. The printed labels are yesterday."

In the Monday row and in This period:

  • Tuesday 13 October, 09:40 UTC. Drew, 9. "Saved the morning without retyping the list."
  • Wednesday 14 October, 13:05 UTC. Noel, 7. "Check-in works. The color legend is last week's."
  • Thursday 15 October, 08:18 UTC. Parker, 10. "Monday's no-shows are gone."
  • Thursday 15 October, 18:02 UTC. Morgan, 9. "Two desks stopped double-booking room 2."
  • Friday 16 October, 09:02 UTC. Reese, 4. "I checked in the 9:00 patient on the 9:20 chart."
  • Friday 16 October, 16:05 UTC. Riley, 9. "The afternoon list matched the door."
  • Saturday 17 October, 11:20 UTC. Avery, 2. "Saturday clinic still shows Friday's no-shows."
  • Sunday 18 October, 07:30 UTC. Skyler, 9. "Printed labels match the screen."
  • Sunday 18 October, 21:44 UTC. Quinn, 10. "Closed Sunday without a refresh."

That is the 12. Seven promoters (Sam, Drew, Parker, Morgan, Riley, Skyler, Quinn), two passives (Casey, Noel), three detractors (Jordan, Reese, Avery). NPS is 7 minus 3, over 12, which rounds to 33. The cell adds still stabilizing because 12 is under 100, and it prints plus/minus 55.

In This period, and only in the row labeled 2026-10-19 to 2026-10-25:

  • Monday 19 October, 10:15 UTC. Taylor, 9. "Check-in held for the walk-in."
  • Monday 19 October, 15:48 UTC. Jordan, 3. "Sunday's no-shows came back Monday night." The earlier 6 was 12 October 08:10, and the widget stays dark for seven days, so Monday afternoon is the next ask.
  • Tuesday 20 October, 08:05 UTC. Sam, 10. "Tuesday labels matched the list."
  • Tuesday 20 October, 08:50 UTC. Casey, 8. "Labels caught up. The legend did not."

This period is those four plus the nine that sit in both clocks. Eight promoters, two passives (Noel, Casey's Tuesday 8), three detractors (Reese, Avery, Jordan's 3). N=13. NPS is 8 minus 3, over 13, which rounds to 38, plus/minus 53. The four alone are under ten, so their row says Insufficient (N=4) and does not print a score. Under ten is the same floor as the post on why a naked score lies.

Monday starts at 00:00 UTC

Each answer is filed under the Monday 00:00 UTC of its own timestamp. The label prints that Monday and the Sunday six days later, so 2026-10-12 to 2026-10-18 means from 12 October 00:00 UTC up to, and not including, 19 October 00:00 UTC. Sunday night is in. Monday morning of the next week is the next row.

This period does not use that cut. Opened at 09:12 on Tuesday 20 October, it is the 168 hours that ended at that instant: 13 October 09:12 inclusive, through 20 October 09:12 exclusive. It takes Casey's Tuesday 8 at 08:50 and leaves Casey's earlier 8 at 07:55 on the 13th in the Monday row only. Refresh the results page at 15:40 and both 7-day windows slide. The Monday label does not slide. To is still 20 October, so a new answer at 14:00 on Tuesday joins the row labeled through 25 October, and the label stays.

The segment trait on the same page scores the breakdown above the chart. It does not cut the weekly row. The 33 is every measured version 2 answer in that Monday window, on every plan trait that was stored.

The comment list under the chart is a second form. Its date fields start empty, so the list is not these sixteen rows. Match a sentence to a bucket by the timestamp on the row.

The label runs through Sunday. The count may not.

From and To are calendar dates. The default range is the last 90 days: 22 July through 20 October. The To day is included, and the next midnight is not, so Tuesday 20 October counts through the end of that date. The row labeled 2026-10-19 to 2026-10-25 therefore holds Monday and Tuesday only. Wednesday through Sunday are in the label because the label is the whole Monday week. They are not in the count, because they are after To, and because they have not happened.

The first row is the same clip at the other end. It is labeled 2026-07-20 to 2026-07-26. From is 22 July 00:00 UTC, a Wednesday, so Monday 20 July and Tuesday 21 July are in the label and not in the count.

Check N=12 and N=4 on the sample size calculator before either one becomes the climate. The calculator wants the people inside that bucket, not every answer Sable has ever stored. A Monday bucket that clears ten prints a score and a margin. It does not prove the 10:30 ship caused the 33. Twelve answers move when the next desk answers, and the four newest rows are still a hatched column. That is the limit.

Bottom line: at 09:12 the row labeled 2026-10-12 to 2026-10-18 is NPS 33 on 12, starting Monday 00:00 UTC, including a 6 from before the ship. This period is NPS 38 on 13. The row labeled through 25 October is Insufficient on 4, and one of those four is Jordan saying the no-shows came back. Do not tell the room the rewrite held because 33 and 38 look close.