Sopact is a technology based social enterprise committed to helping organizations measure impact by directly involving their stakeholders.
Copyright 2015-2026 © sopact. All rights reserved.
The Craft is the reliability layer under every Sopact course — the standards that make impact analysis come out the same way twice, with every number traceable to the response it came from. The Loop is the method; the Craft is how you run it so it holds up in front of a funder.
The Craft is the reliability layer under every Sopact course — the standards that make impact analysis come out the same way twice, with every number traceable to the response it came from. The Loop is the method: collect clean at the source, analyze the moment data arrives, improve while you can still act. The Craft is how you run that method so it holds up in front of a funder.
It sits under all four Intelligence courses — Case, Grant, Portfolio, and Feedback — as a shared library, not a course you take start to finish. Define a field once and it means the same thing across every program, site, and year. Write a rubric once and it grades every response the same way. The method becomes the asset, instead of living in one analyst’s head.
Key takeaways
You can. That is exactly the point worth being honest about. Any capable model will summarize open-ended responses, draft a narrative, or sketch an SROI. The difference is not the answer on the first try — it is whether you get the same answer on the second.
“If you write the report, you can ask the report twice, it gets the same result. If you put it in ChatGPT twice, it’s going to give you two different results. That’s not what Sopact is. Sopact’s job is to give you reliable, traceable results.” — Unmesh Sheth
That is the line between a demo and a decision. A number you can’t reproduce is a number you can’t defend, and a funder who asks “where did this come from?” needs an answer on the page, not a re-run that returns something new. Reliability, traceability, and consistency are the whole product — and they are the things a general chat tool does not give you.
The Craft rests on a simple pipeline, and it helps to see it plainly. You collect rich data from many sources — surveys, open text, interviews, uploaded proposals and documents. The moment it lands, it is analyzed against a rubric you defined, not read by an analyst weeks later. That analysis is written to a structured record you can query in plain language.
“It immediately analyzes that — that’s Intelligent Cell and Intelligent Row. It applies the rules you have defined; we call it a rubric. It puts it in a database that now you can start using the Assistant, or Claude, to ask any question.” — Unmesh Sheth
The reason this beats a chat window is not intelligence — it is persistence and consistency. The rubric is fixed, so the grade is mechanical. The record is durable, so the same question returns the same number. The trail is kept, so every figure points back to a source.
Take one open-ended question two of your programs both ask, and grade it with a rubric that is deterministic by design — a fixed scale, a mechanical rule, a quote required for every score.
Grade each response to “What was the biggest barrier you faced?” against this FIXED scale: ACCESS, COST, TIME, CONFIDENCE, OTHER. Rules: assign exactly one label; quote the exact words that justify it; if the response does not state a barrier, label it NOT STATED and do not infer. Return: response id · label · the quoted words. The same response set must return the same table every run.
Run it twice. Because the scale is fixed and the rule is mechanical, you get the same table both times — and every label carries the words that earned it. That reproducibility is the thing you cannot get from a free-form prompt, and it is what a customer noticed the moment they switched.
“The reports I’m getting are much more consistent and accurate. I can more easily gauge where the Assistant got its numbers from, and they’re consistent from report to report.” — Marco, Open Play Foundation
| Strong fit | Why |
|---|---|
| You run more than one program | One shared definition makes every program, site, and year comparable |
| A number will be quoted or audited | Deterministic grading returns the same figure, cited to source, every run |
| You’re tired of AI drift | Same input, same graded output — not a fresh answer each time |
| Not this | Why |
|---|---|
| A data warehouse | Sopact is the collection and intelligence layer; connect out for warehousing |
| A standalone product path | It’s the standard under the four courses, not a course on its own |
| A free-form chatbot | Deterministic and cited, not a model that drifts run to run |
“Sopact is not a data warehouse. It’s a data collection and intelligence platform. We don’t want to load the platform as a warehouse — there’s a better solution: integrate with Claude or ChatGPT.” — Unmesh Sheth
The honest boundary is the credibility. The Craft is what makes the broad best practices you already know — clean data, clear indicators, defensible reporting — actually hold when someone checks your work.
Frequently asked questions
The reusable reliability layer under Sopact’s four courses — the data dictionary, rubrics, frameworks, the SROI method, and the deterministic-output standard, defined once and applied everywhere.
General models give a fluent answer, but a different one each run. The Craft is deterministic and cited: the same input returns the same graded output, and every figure traces to its source.
No. It’s a library every course reaches into — the graded framework in Case, the cited scoring in Grant, the standards mapping in Portfolio, the themed open text in Feedback.
Take one term two of your programs both use and reconcile it to a single definition. That one line is the first entry in your data dictionary — and the reason your programs finally become comparable.
Collection works anywhere; analysis runs the moment data syncs. The point is timing — themes, scores, and a draft report exist within minutes of arrival, not weeks later.
Data leads, analysts, and anyone responsible for keeping a whole organization’s measurement consistent across programs, funders, and years.
Next: Reliability: The Same Answer, Every Run → · or Try Sopact Sense →
Open Sopact Sense, paste your program description, and put it to work.
Try in Sopact