A table tells you what; a chart tells you how it’s going. The trouble is scale: forty products, each with twelve months of sales, wants forty trend lines — and forty real charts is a wall nobody reads. The tool built for exactly this gap is the sparkline: a chart the size of a word, living inside a cell, one per row.

The name and idea come from Edward Tufte, who defined sparklines as “small, intense, word-sized graphics” — meant to sit in the flow of a table or sentence, giving numbers their shape without interrupting the reading. Excel has had them since 2010, three flavours: Line (trend over time), Column (magnitudes), Win/Loss (above/below zero, and only that).

Making them

Insert → Sparklines → Line: data range = the twelve month columns, location = the empty column at the row’s end. The efficient habit — insert one, then fill-handle it down; sparklines copy like formulas, each row charting itself. Better again, live inside a Table so new rows sprout their own.

Now every product carries its own twelve-month story in a centimetre: steady, collapsing, seasonal, recovering — visible at reading speed, forty rows scanned in the time one chart takes to decode. That scan is the product: sparklines are for triage, finding which rows deserve a real chart.

Product · 12 months · trend Widgets …monthly cells… Gaskets Flanges ← Sprockets forty rows scanned at reading speed — Flanges is the afternoon's question
One word-sized chart per row. Nobody reads forty charts; everybody notices the one line falling — that's triage, and it's what sparklines are for.

The three settings that make them honest

Sparklines inherit chart ethics with one twist of their own.

Same axis, or no comparison. By default each sparkline scales to its own min and max — so a product wobbling between 98 and 102 looks as dramatic as one swinging 0 to 1,000. Fine for “what’s this row’s shape?”, a lie for “which row is moving most?”. For comparisons: select the group → Sparkline tab → Axis → Same for All Sparklines (both min and max). This is the truncated-axis rule in miniature — per-row axes are forty tiny zoomed charts.

Mark the points that matter. Tick High Point and Low Point (and Negative Points for anything crossing zero). A line with its extremes dotted answers “when?” as well as “how’s it going” — still without a single axis label.

Handle the gaps deliberately. Sparkline tab → Edit Data → Hidden & Empty Cells: blanks can be gaps, zeros, or bridged with a line. The right answer is the same one AVERAGE forced on you: a missing month and a zero month are different facts. Choose, per dataset, on purpose.

Where they sit — and where they don’t

Sparklines earn their keep in dashboards (a KPI row with a twelve-month sparkline under each number is the densest honest information in Excel), pivot-adjacent summaries, and any Table where rows have a time dimension. They coexist happily with conditional formatting — data bars answer “how big?”, sparklines answer “which way?” — and the pair covers most of what row-level graphics can honestly do.

Their limits, honestly: no axes means no values — a sparkline says “falling”, never “by how much” (keep the numbers beside them; Tufte’s point was always words and graphics together). One series per cell. Invisible in some older viewers. And they’re a view, not analysis — a falling line is a question, and the real chart is where you take it.

A chart the size of a word, one per row, axes shared, extremes dotted. The big table finally shows its shapes — and the one line heading the wrong way finds you.