Why this matters
Before a board meeting, someone exports a year of gifts and spends the evening merging appeal names and building totals. A number that goes in front of the board has to hold up, and one double-counted gift is enough to make it wrong.
Better Analyst, formerly Formula Bot, does the building from a plain-English question. In our test it cleaned the appeal names unasked, and it still counted one gift twice until we told it to check. So this lesson builds in the check.
Practice
Do every step yourself in a practice copy of the real tool. It checks your work, and “Do it for me” finishes any step you get stuck on.
What you’ll build
You run fundraising at a made-up food bank. You upload 250 rows of gifts, get totals and a chart, then make Better Analyst audit the raw data before anyone sees it.
You correct what it finds, check the result, and write a three-sentence note that says what a person must still confirm. The screens are real screenshots from my own Free plan account, captured October 2026.
What you need
- Nothing to start the practiceThe lesson is a practice copy. The food bank and its donors are made up.
- A Better Analyst account, to do it for realThe pricing page lists a Free plan with 10 chat messages a month, 50 AI credits, one connected data source and one AI dashboard. Charts carry a watermark. Check the current plans.
- The made-up gift exportThe download link is in the steps for doing it for real. Use it or a test copy, not your live donor data, while you try it.
Concepts
A finished-looking chart is not a checked one
Our first answer merged Spring Appeal, spring appeal and SPRING APPEAL by itself, which is real help. It also reported a total of 46,340 when the true total was 46,290, because one gift was in the file twice.
It flagged the repeat as soon as we asked. So ask every time, by name: repeated IDs, negative amounts, and single gifts that could change the ranking.
A check only answers the question you asked
Our check asked about gifts that could change the ranking. A 5,000 gift, nearly 39% of Year-End Letter, could not, so it was not on the list.
It came up only when we asked for that appeal's largest gift directly. Look at the biggest gift in each appeal yourself, because a board may read one family's gift as a successful mailing.
Be ready to explain the choices behind a number
We kept a 500 refund in Golf Outing's net total but left it out of the gift count. Better Analyst then averaged the positive gifts only, 763.89, and said so under the table.
That is a fair choice, but it is a choice. Also, the file has no currency column, so it reported amounts as recorded. Add that context before the numbers leave your desk.
Do it for real
Now repeat it in your own account. Tick each step as you go.
Downloads
- Made-up gift export (CSV)Riverbend Community Food Bank: 250 made-up rows with messy appeal names, one repeated gift and one refund.
Quick check
3 questions to check you’ve got it.
1. The first chart looks finished. What should you do before it goes to the board?
- Ask for a specific check of the raw data, then confirm what it flags
- Nothing, the AI read every row
- Change the chart colors
Show the answer
Ask for a specific check of the raw data, then confirm what it flags. Our first answer counted one gift twice and said nothing until we asked.
2. Why did the check not list the 5,000 gift?
- It could not change the ranking, which is what we asked about
- The AI could not read it
- The file left it out
Show the answer
It could not change the ranking, which is what we asked about. A check answers the question you asked. We got the 5,000 gift by asking for the largest gift directly.
3. Golf Outing leads Year-End Letter by 335. What does that mean for your board note?
- Say how close it is, since one gift can flip the order
- Call Golf Outing the clear winner
- Leave Year-End Letter out
Show the answer
Say how close it is, since one gift can flip the order. Removing any one of the 15 listed Golf Outing gifts moves it below Year-End Letter. Our AI-written note left that out, so add it yourself.