← All Kits · Excel Kit · Article 0: the build order

Write the Sentence Your Dashboard Is Arguing

Article 17 of Excel for Business Analysts · The finding line: the last step, and the test of all the others · Part of the Analyst Prep Kit

This article gives you the last step of the build order, and the test that tells you whether the other sixteen articles added up to anything: write the page's claim as one sentence with a number in it, and put that sentence on the page where a reader meets it first.

The build's sentence: 175 games are loved just as much as the famous ones and reach a twentieth of the audience. Everything on that dashboard exists to back those words. Anything that does not back them came off.

A claim, not a list

Try your own current project first: say what the page shows, in one sentence, without the word "and" doing heavy lifting. If what comes out is "it shows sales by region, and trends over time, and the top products," that is a list. A list is what you have before you have found anything.

A claim is a sentence somebody could disagree with. "West region's growth is entirely two customers" is a claim: it could be wrong, it could be argued, and a decision hangs on it. Nobody can disagree with a list, which is exactly why lists feel safe to ship and why they change nothing when they land. The disagreeability is the value.

What the sentence must contain

Length check, the same one this series uses for everything: say it aloud in one breath. The sentence will be quoted, forwarded and misremembered, and short sentences survive all three.

The claim is a fork: what would change someone's mind?

The strongest test of a claim is whether it settles a real either-or. The build's dashboard was built to answer one question with two possible answers, and it is worth seeing the fork plainly, because your page has one too.

The question: is there a reason these games stayed unknown?

Answer one: they are worse. Cheaper, shorter, lower rated. If true, the market did its job, being unknown makes sense, and the page's finding is "nothing to see."

Answer two: they are the same. Same ratings, same prices, same playtime as the found games. If true, quality does not explain anything, and being unknown is mostly luck.

What decides it: the comparison table, hidden against found, side by side on the same measures. If the rows look alike, answer two.

The rows looked alike, so the sentence claims answer two, with the 21.2x audience gap as the stakes. Now say your own page's fork: what are the two answers your data could have given, and which did it give? A page that cannot name its fork is usually a list with formatting.

Where the sentence goes on the page

At the top, as the title or directly under it, before the KPI row. Not the file name, not "Sales Dashboard Q3," not saved for a closing text box nobody scrolls to. The reading order of the finished page is the argument in miniature:

The sentence        the claim, in words
The KPI row         the claim, in numbers
The hero chart      the claim, drawn
Everything else     support, for whoever wants to check

A reader who leaves after five seconds leaves with the finding. That is the payoff-first rule this entire series is written by, applied to your own work: the reader gets the result before the method, because the result is what they came for.

The sentence as the editor of everything else

Written early, a draft claim becomes the cheapest editorial tool you own. Hold each element of the page against it and ask one question: does this back the sentence?

Charts that back it stay. The comparison table backs it. The audience-gap chart backs it. Interesting-but-unrelated work comes off, however long it took to make. A genre breakdown that neither supports nor challenges the claim is decoration wearing analysis's clothes. And anything that challenges the sentence gets dealt with honestly: either the claim narrows, or the challenge earns a spot on the page as a stated limitation, per documenting what your data cannot tell you. A claim that survived its own counter-evidence is the only kind worth shipping.

This is also article 14's audit from the constructive side: the same sentence you test for circularity is the one you build the page around. Write it, audit it, then let it edit.

When you cannot write it

Sometimes the sentence will not come, and that is diagnostic, not failure. Three honest cases:

The analysis is not done. You have distributions and cuts, no fork settled. The fix is more step 3 through 5 of the build order, not more formatting. The page is not late. It is early.

The finding is "no difference." Write that sentence, with its number: "the two groups differ by under 2% on every measure." Null findings kill wrong plans, which makes them findings. The build's whole claim is half a null: same quality, so quality is not the explanation.

The page is genuinely a monitoring tool, built to watch numbers rather than argue one. Then say so in its one line: "daily volumes by region, updated each morning." That is a purpose sentence instead of a claim sentence, and choosing which kind your page is, on purpose, is the decision this article exists to force.

Run it on your own page

1. Write the sentence now, before the page is done. One sentence, one number, comparator named. Draft is fine.
2. Say the fork it settles. Two possible answers, which one the data gave.
3. Audit it against article 14: could it have come out differently?
4. Put it at the top and order the page as claim, numbers, hero, support.
5. Hold every element against it. Backs it, stays. Neither backs nor challenges, comes off. Challenges it, narrows the claim or ships as a limitation.
6. Read it aloud once more before sending. One breath, plain words, a number. That sentence is what your work becomes in other people's mouths.

A cheat sheet

So: the sentence you tried to write at the top of this page. Was it a claim, or was it a list? Write the claim version now, number and all, and pin it above your dashboard before you touch another cell.