Design backward from the decision
Name what you’ll actually decide, then the evidence that would move it — and the instrument designs itself.
Pick the instrument that reveals most
Data, document, survey (mixed), and interview — and when each gives deeper insight than a survey alone.
Write the instruction & the rubric
Plain-language instructions that read every document on arrival, cite the source sentence, and never guess.
Seven deep, practical chapters
Case, Application, Grant, ESG, Impact, Learning, Program — each the hands-on next step from its eBook.
What is AI data design?
AI data design is the practitioner skill of choosing the right instrument — data, document, survey, or interview — and writing the plain-language instruction and rubric that read it, so AI turns the 95% of evidence a survey misses into cited, comparable results. No code, no statistics required.
What this guide is about
A survey only answers a question you already knew to ask. The richest signal in any program — why a participant dropped out, where a grantee’s narrative contradicts their numbers, the blocker no checkbox offered — lives in the documents, interviews, and open-ended replies a survey never captures. The AI Data Design Guide is the practical, no-code companion to the whole Sopact Intelligence Library: it teaches you to design for that signal and to write the instructions that turn it into evidence you can defend.
Who should read it
Program managers, M&E and impact leads, foundation officers, and review committees — anyone who runs a program and has to prove it worked, without a data team. If you’ve only ever reached for a survey, this is the guide that opens the other 95%.
What’s inside
Part 1 — the craft: design backward from the decision; choose the instrument (data, document, survey, interview); write the question and the instruction; build the rubric; and pick the reporting scope (cell → row → column → grid).
Part 2 — seven deep chapters, one per use case: case, application, grant, ESG, impact, learning, and program intelligence. Each works one program manager’s real situation end to end — the data to design, the instructions to write, the rubric, the four scopes with worked numbers, and the reports to build.
How to use it
It’s the practical next step from each eBook: the eBook gives you the why; the matching chapter gives you the how. Start at sopact.com/ebooks, grab the guide for your work, then design it here.
Frequently asked questions
How is AI data design different from writing a survey?
A survey is a fixed list of questions, so it can only return answers you anticipated. AI data design picks the instrument that reveals the most — often a document or an interview — and writes an instruction that reads it on arrival against a rubric, so the evidence a survey can’t capture becomes a cited, comparable score.
Do I need to know how to code or do statistics?
No. The guide is written for non-technical program staff. You design the question, choose the instrument, and write the instruction in plain language, like briefing a careful new analyst; the system reads the documents and cites the evidence.
What does the guide cover?
Part one teaches the craft — design backward from the decision, choose the instrument, write the question and instruction, build the rubric, and pick the reporting scope. Part two applies it in seven deep chapters, one per Sopact use case, each with worked input, transform, and report examples.
Is the guide free?
Yes. It is a free download from Sopact and the practical companion to every guide in the Sopact Intelligence Library.
