How do you report to several funders without collecting everything twice?

In short: collect once, into one set of records with one definition per metric, then report many times. Each funder gets its own format; none of them gets its own data collection.

Teams that report to four funders often end up running four small data systems, one per template. The same trainee is counted four ways, and nobody can say which count is right.

This chapter follows Partner C from the funder road, now from its own side. It is a fictional workforce training program that reports to several funders. Step 4, mapping every ask to your evidence, decides whether the rest of its road is calm or frantic.

Step 3 is not taught here. Deciding what to collect, and building it into the forms and records you already run, belongs to the Foundations course. This chapter assumes those records exist and asks a narrower question: which of them answers each funder's ask?

What goes in a funder requirements map?

In short: one row per funder request: the funder, what they ask in their own words, the metric in your dictionary that answers it, and a note on any difference in definition.

Keep the funder's wording exactly as written, with the page or clause it came from. Your metric column points to a row in your data dictionary, not to a new field. If no row fits, that is a gap to discuss, not a form to add in a hurry.

Requirements map for Partner C · fictional example. Funder C's code is from the IRIS+ catalog.

Build the map the week an agreement is signed, not the week a report is due. The onboarding chapter shows how to get the definitions into the agreement in the first place.

What if a funder defines a metric differently from you?

In short: report their number, labelled with their definition, and add one sentence on how it differs from yours. Do not change your data or your dictionary to match.

Funder A's template asks for "participants served" and counts every visit. Partner C's dictionary counts unique trainees. Attendance gives visits, the enrolment record gives people. Neither is wrong; they answer different questions.

The same rule covers Funder B's six-month placement window. Give the six-month count in their narrative, say that your agreed measure elsewhere is 90 days, and keep the two apart. This is where mistakes cross over. On the funder road, the regional workforce fund found that Partner C reported placements "within 6 months" where the agreement said 90 days, and sent the question back. A map row with a note would have caught it before the report left.

Why does one ID per person matter for every wave?

In short: change is measured by comparing the same person over time. If each survey wave creates a new record, you cannot connect intake to follow-up, and no funder's outcome question can be answered.

Girls Inc. of Metropolitan Dallas runs a coding program with surveys before, during and after the program, then follow-ups at 6 months and 1 year. Its reporting cycle follows a July 1 fiscal year. Five waves for one participant only tell a story of change if all five land on the same participant record.

Partner C faces the same test. "Retained at 12 months" needs the exit record and the 12-month follow-up to belong to one person. When a funder's year and your program year differ, the ID lets you place each follow-up in the right reporting period. The measuring change chapter in Foundations covers the survey design side.

In Sopact Sense today, a persistent unique ID is issued at the first form, and each later workflow is added to the same record. With any other tool, the practice is the same: one ID, created once, carried into every form by a link or a lookup, never retyped.

How can AI map a new funder's template to your dictionary?

In short: give it the template and your dictionary, and ask for a draft map with a gap list. Then a person confirms each match against the funder's own guidance.

AI does the first pass well: reading a long template and matching labels to dictionary rows. It cannot know what the funder meant. A match on the word "participants" is a guess until someone checks it.

You are helping a funded program map a new funder's reporting template to its existing data dictionary.

Inputs:
1. The funder's reporting template and guidance: [PASTE OR ATTACH]
2. Our data dictionary (metric, definition, dimension, standard code, collection point, disaggregation, source): [PASTE OR ATTACH]
3. Our signed agreement with this funder, if available: [PASTE OR ATTACH]

Tasks:
- List every item the funder asks us to report. Quote their exact wording and give the page or section.
- For each item, name the dictionary metric that answers it, or write "no match".
- Mark each match as SAME, DIFFERENT or NO MATCH. For DIFFERENT, say exactly how the definitions differ (population, period, time window, unit such as visits versus people, disaggregation).
- For each DIFFERENT item, draft one sentence we could put in the report to explain the difference.
- List the gaps: items we cannot answer from current records, and what we would need to collect or ask.

Rules: do not invent definitions the documents do not state. If the funder's meaning is unclear, write it as a question for their program officer. Return a table with columns: Funder asks | Our metric | Match | Note.

An honest limit: the map shows where the gaps are, it does not fill them. If Funder B wants 12-month retention before a full year has passed since exit, the right line in the report is "not yet available" with the date it will be. Templates also change at renewal, so reread each funder's guidance every cycle.