Pick a depth. Each prompt opens in your AI pre-loaded with the lesson. Click a row to preview the prompt.
The most dangerous analyst is a confident one with a wrong number, because people act on it. It is shockingly easy to produce a wrong answer that looks right: a filter that silently dropped half your rows, a SUM that included refunds, a currency in cents read as dollars making everything 100x too big. The habit that protects you is a 60-second reflex - before you share any number, ask 'is this even plausible?' - and it is the single trait that makes people trust your work over time. This task plants that reflex; Module 2 builds it into a full data-trust discipline, but you start it here, on your very first answer.
Run three quick checks on any number before you trust it. Check one, magnitude: does the size make sense against something you already know? If your sheet says your coffee shop made 380 dollars average per day but you know from the bank it is closer to 700, something is wrong - maybe your filter excluded the busiest weekdays, or you averaged only cash sales. A number that is 2x or 100x off a known figure is a red flag, not a discovery. Check two, does the total reconcile? Sum your Revenue column with no filter at all and compare to a known real total - if your annual revenue in the sheet is 95,000 dollars and your tax return said 140,000, rows are missing or a category is excluded. Check three, did a filter silently drop rows? This is the sneakiest one. After filtering, glance at the row count - Google Sheets shows how many rows a filter is displaying, or you can put =COUNTA(A2:A200) before and after. If filtering to 'Saturday' left you 4 rows but three months should have about 13 Saturdays, your Weekday column has inconsistent values ('Sat' versus 'Saturday') and you are missing most of them. Worked example of catching an error: a learner computed average order value at 4,500 dollars, paused, thought 'no one spends 4,500 on coffee,' and discovered the Amount column was in cents - the real answer was 45 dollars. That pause saved a very embarrassing report. Write down, in one line, the check you ran and one thing that could still be wrong - because being honest about residual uncertainty is what makes people trust you more, not less.