Unit 02.00: Starting from the decision, not the data

A chart cannot be right or wrong until there is a decision it serves.

Decisions name the comparison for you

"Should we keep the Tuesday promotion?" already tells you what to compare. "Make a dashboard of everything" tells you nothing.

The code sorts four openings by what they actually are.

STARTS = [
    ("we have the sales table, what should we plot?",
     "exploration -- legitimate, but not a deliverable"),
    ("should we keep the Tuesday promotion?",
     "a decision -- it names the comparison"),
    ("make a dashboard of everything",
     "no decision, so no chart can be judged right or wrong"),
    ("which supplier should we drop?",
     "a decision -- it names the ranking and the unit"),
]
print(f"{'the ask':44} what it is")
for ask, kind in STARTS:
    print(f"{ask:44} {kind}")

print("""
A decision names the comparison for you. "Should we keep the Tuesday
promotion?" implies Tuesday against other days, over a period long enough to
matter, in a measure the promotion could plausibly move.

Without a decision, a chart cannot be wrong -- which sounds comfortable and
means it also cannot be right.
""")

The first is exploration, which is legitimate and is not a deliverable. The third is the one that produces work nobody uses: with no decision, any chart is defensible and none is checkable.

The two decisions each imply a comparison, a unit and a period without anyone specifying them separately.

The mistake this prevents

The mistake is accepting "a dashboard of the sales data" as a brief. It is a data source, not a question, and the resulting dashboard will be judged by whoever opens it against a decision they never told you about.

Takeaway

Start from the decision. It names the comparison, and without one a chart cannot be wrong - which means it also cannot be right.

Mark complete