The Green Pill Problem
An analyst has [Order Date] as a proper Date field. She drags it to Columns and Tableau hands her a continuous green MONTH(Order Date) axis spanning 36 months; the tick labels only print at Jan, Apr, Jul and Oct, so a stakeholder insists "the chart skips months." She switches the pill to Discrete, gets 36 blue column headers - and loses the axis entirely. On a second sheet she puts the same continuous month pill on Color to shade a bar chart by recency and gets a smooth multi-step gradient legend instead of the four quarters she pictured. Meanwhile the first sheet, with 36 months and a measure on Rows, renders fine but her eyeballed "months are evenly spaced" reading is wrong in a way nobody can articulate.
1. Is the continuous axis skipping months? Explain what continuous means here and give the fix for the stakeholder's complaint.
Read the worked answer
Nothing is missing. A continuous field is treated as an unbroken number line: every month has a mark, and Tableau decides how many tick labels fit at the current width and prints them at a regular interval. Quarterly ticks at that width is a rendering decision, not a data filter - hover any point and the value is there.
So answer the complaint at the axis, not the pill: right-click the axis → Edit Axis → Tick Marks → Fixed, major tick every 1, and set a custom date format like mmm yy so 36 labels stay legible. If they still collide, give the sheet more width, or split by year into columns.
Switching to Discrete is the wrong fix if she also wants a real time scale - that is the next question. And note the tell: blue pill with headers means categorical; green pill means an axis. Same field, different contract with the reader.
2. Why did Color produce a gradient instead of four quarters, and how do you get four?
Read the worked answer
The pill's discrete/continuous state drives the encoding. Drop a continuous field on Color and Tableau builds a quantitative colour range - a continuous legend with min and max and a smooth ramp. Drop a discrete field on the same shelf and you get a categorical palette with one swatch per member.
So: replace the continuous month with a discrete quarter - QUARTER([Order Date]), or DATETRUNC('quarter',[Order Date]) set to Discrete - and edit the colours so the four quarters are deliberately ordered and distinguishable. A secondary option is to keep the continuous field and use Edit Colors → Stepped Color with four steps, which gives a stepped ramp rather than a smooth one; it works, but it implies an ordered scale where a categorical palette says "four groups," which is what she means.
Pick the encoding to match the question being asked: a gradient answers "how does this vary over time," four swatches answer "which quarter is this." Mixing them makes the reader work.
3. Discrete gives readable labels; continuous gives a true time scale. Which do you ship, and what is the trap?
Read the worked answer
Ship both, on two sheets, and stop toggling the pill on one sheet - because the PDF a stakeholder opens is whatever state was last saved.
- Sheet A (continuous): the true time scale. Horizontal distance is proportional to elapsed time, so a gap in trading shows as a gap, trend lines and reference bands are fitted on real time, and zoom actually works. Fixed ticks every month solve the labelling complaint.
- Sheet B (discrete): equal-width columns, every member labelled, easiest to read values off. The trap is that discrete spacing is categorical, not temporal - 28-day February and 31-day July occupy identical widths, and any visual trend judgement across members is now about ordering, not time.
Say that out loud in the answer: the two sheets can look identical and imply different things, so the titles must state which is which. If only one sheet can live, pick continuous for anything trend-shaped and put the numbers in a discrete crosstab below it.