How ImpactBI works
A feedback loop, not a dashboard.
Bring in the numbers and the words you already collect, however messy. Ask what you want to know, keep what holds up, share it, and ask the people themselves. Their answers come back in.
Files
Start where your data already lives.
A shared drive full of spreadsheets, survey exports, intake forms, case notes saved as PDFs, and the same person spelled three ways. That’s normal. Bring it as it is.
- Connect Google Drive. Pick files straight from your Drive, with no exporting or re-uploading. Or upload files directly.
- Written answers and PDFs count. Open-ended survey answers and case notes are read alongside the spreadsheets.
- Names and contact details are set aside first. Data Assistant never sees them, in a table or in a sentence.
- Every file gets a Trust Artifact. A plain record of what changed and why.
From messy to ready
Messy data is the starting point, not the problem.
Most teams track the same people in different places, with different column names and different habits. ImpactBI sorts that out before you ask a single question, and shows you exactly what it did.
- It reads what each column means“Client ID”, “participant_id” and “PID” are recognized as the same thing. Anything it isn’t sure about, you confirm in a click.
- It cleans what’s inconsistentDates written four ways, ZIP codes with extra digits, yes-ish answers and repeated rows are tidied, and every change is written down.
- It connects your filesOnce the columns line up, a question can run across the roster, placements and surveys at once, without anyone’s name.
Data Assistant
Ask the way you’d ask a colleague.
Data Assistant works across all your files, shows its steps, and builds a canvas beside the conversation as you follow up.
- Facts and inferences are labeled. You always know what the data says and what Data Assistant thinks it means.
- Every number traces back. “How I got here” shows the files, versions and steps behind it.
- Follow-ups build a canvas. A line of questioning turns into a picture you can show someone.
The words matter as much as the numbers
Numbers show what happened. People tell you why.
- PDFs are synthesizedCase notes and reports are read page by page and summarized into what they say about your question.
- Written answers are groupedOpen-ended survey answers are sorted into themes and counted, so you see how many people said something.
- Quotes sit next to the chartsThe people behind a number are heard in their own words, anonymized first.
Insights
Keep what holds up, and say what doesn’t yet.
Save an answer as an insight. It keeps the charts, the quotes and how it was reached, and a program lead can verify it before it travels further.
- Hunches are labeled. When the evidence is thin, Data Assistant calls it a hunch, not a finding, and suggests how to find out.
- Hunches can grow into findings. As survey answers, updated files and next quarter’s data come in, Data Assistant checks the hunch again and tells you when it holds up.
- Recommendations rest on real evidence. Data Assistant tests whether a difference is statistically significant before it recommends anything, and says how confident to be.
- Sharing follows access. People who can’t open every file behind an insight never see it.
Reports
Turn insights into reports for your board and funders.
Pick the insights and ImpactBI drafts the report. Every number links back to the insight and the file it came from.
- Draft, review, send. A lead reviews before it goes out.
- No one can be identified. Every chart and quote is checked before it leaves.
Surveys
When your files can’t settle it, ask the people.
Data Assistant can turn a hunch into a short survey. You review every question before anything is sent, and you can follow responses as they come in.
- Track responses as they arrive. See how many have answered and what they’re saying.
- Answers come back into Files. Ask again, and the hunch becomes a finding, or doesn’t.
- Answers stay anonymous. The same protections apply from the first file to the last answer.
Built in, all the way round
The people in your data are protected at every step.
- Outside data comes only from trusted sourcesWhen a question needs public data, like transit maps or census figures, Data Assistant uses only sources on your organization’s approved list, and asks you first.
- Some questions it won’t answerQuestions that could single out or endanger people are refused, whoever asks.
- Access follows your organization and teamsEach person sees the files, insights and reports for their own teams. Admins see across the organization, and anything shared follows the same rules.
Built with nonprofits first.
We’re preparing for a closed beta in 2027 with a small group of program-based nonprofits.
