A capacity-building grantee builds five connected reports across one training cohort: enrollment baseline, pre/post skill, score-confidence correlation, 90-day behavior change, and funder-ready narrative synthesis. Each report rests on a persistent learner ID assigned at intake.
A capacity-building grantee builds five connected reports across one cohort. Cohort baseline at intake. Pre/post skill change at program close. Score-confidence correlation that distinguishes capability from confidence. Ninety-day behavior change in the workplace. A funder-ready narrative synthesis that ties all four together. Each report below shows how the build is done — what the raw input is, what the dictionary rule extracts, what the funder ends up reading.
The worked example threading through every section: Pathways Forward, an Oakland nonprofit, received a $200K two-year capacity-building grant from the Bay Area Workforce Foundation. The grant funds Career Bridge — a 12-week digital-skills and career-readiness program. Cohort 3 has 47 enrolled participants across six mastery skills. The five reports below are what Pathways Forward delivers to its funder.
Who enrolled, where they started, the segments every later report will disaggregate by.
Jump to build →Delta per skill dimension, paired by persistent learner ID, segmented by demographic.
Jump to build →The shape only possible when qualitative and quantitative live on one learner record.
Jump to build →What percentage of the cohort is applying skills on the job at 30, 60, and 90 days.
Jump to build →One live URL combining the four prior reports plus participant voice with citation chain.
Jump to build →Most training grant reports fail not at writing but at joining. Pre-program data sits in one form; post-program in another; ninety-day follow-up in a third. When the funder report is due, an analyst tries to reconnect them and finds the IDs do not match. The five-report architecture below works because every instrument writes to the same learner record from the moment of enrollment.
Before the first cohort enrolled, Pathways Forward set up five instruments inside Sopact Sense and assigned them all to one schema: intake form, baseline rubric, weekly check-in, post-program rubric, and 90-day follow-up. The intake form generates a persistent learner ID at first contact. Every later instrument inherits that ID automatically — no name matching, no email reconciliation, no analyst time spent stitching records together when the report is due.
The architectural choice is upstream of the reports. Once the persistent ID is wired in, the five reports below become filtered views of one dataset rather than five separate authoring projects. The first cohort pays for the architecture. Every cohort after that produces the same five reports automatically.
A capacity-building grant is an investment in the grantee's ability to keep producing evidence — not only in this cohort's outcomes. When the Bay Area Workforce Foundation reads Pathways Forward's report, two layers matter: the cohort outcomes themselves, and the proof that the next three cohorts will produce comparable evidence without a separate evaluation project each time.
This is what Kirkpatrick Level 3 actually requires. Most programs report at Level 1 (satisfaction) or Level 2 (skill change) not because they choose to but because the infrastructure for Level 3 was never built. Connecting a 90-day follow-up to the original intake record is a data architecture problem, not a survey writing problem. Persistent IDs solve it; nothing else does.
The next five sections walk through each report in the order Pathways Forward produces them. Each section has the same shape: a three-stage build (raw input → dictionary rule → report fragment), then three callouts that name why the build works, what decision it enables for the program team, and what the funder reads in it.
The first report the funder ever sees from a capacity-building grant is the one most grantees skip — the baseline. It answers who enrolled and where they started. Without it, every later number has no reference point, and the funder cannot tell whether the program reached its target population. The baseline is also where the persistent learner ID gets assigned, so every later report in this series depends on it.
The persistent learner ID is generated at the moment of intake — before the participant has filled out anything else. Every later form in this article inherits that ID automatically. No name matching, no email collision, no analyst reconciliation step at week 12 or day 90. The baseline rubric also uses the same 1–5 scale on the same six skills as the post-program rubric, so the pre/post join in the next report is structural, not interpretive.
Which skill the curriculum should weight most heavily in Cohort 3. The lowest baseline bar — project lifecycle at a mean of 1.25 — is the one that should get the most contact hours. The team adjusts the syllabus in week 1, not after the cohort closes.
Proof of reach to the target population (68% first-gen confirms the demographic target). A baseline that is honest about how far participants have to travel, so the post-program delta is measured against where they actually started — not against an aspirational benchmark. Demographic disaggregation at intake means the segmented analysis in later reports does not require a retrofit.
The Kirkpatrick Level 2 report most funders expect. The reader has to see what changed in participants over the program — by skill, by demographic, and in their own words. The build is straightforward once one architectural choice was made at intake: the persistent learner ID that joined the pre-rubric is the same ID joining the post-rubric. The join is automatic; the analyst no longer reconciles by name and email.
The learner ID assigned at the first form carries through every later form. When the post-survey arrives, the join is automatic and the delta calculates without anyone reconciling exports by name and email at year-end. The same architecture lets the open-text reflection sit next to the number it describes — quote and chart, same person, one record. Every score on the bar chart traces back to one participant; every quote traces back to one response.
Which skill dimensions need curriculum adjustment in the next cohort. The bar that did not move is the one to redesign. The bar that moved most is the one to defend at renewal. Project lifecycle went from a baseline of 1.25 to a post-program 3.1 — that's the largest delta in the cohort and the syllabus weighting decision from §04 paying off.
A skill delta with sample size disclosed (n=47, 100% completion, 4.3/5 confidence) and a segment-level breakdown that does not collapse the cohort into a single average. The funder reads this section first to confirm Kirkpatrick Level 2 is supported by evidence rather than asserted from completion alone. The methodology paragraph that says "pre and post paired by persistent learner ID assigned at intake" is the proof that the join was structural — the funder's audit committee reads exactly that line.
The shape only possible when qualitative and quantitative live on one participant record. Standardized rubric scores measure capability. Open-ended reflections — when an AI-extracted confidence rubric is run against them at the moment of collection — measure something different: how the participant feels about applying what they learned. Most programs collect both and analyze them separately. The build below joins them on persistent learner ID and produces one chart that answers a question neither side could answer alone.
Coding the confidence rubric at the moment the reflection arrives — not at year-end — is the architectural move. The rubric attaches to the persistent learner ID immediately, so the join with the test score is already done by the time anyone opens the analysis. The outliers are visible right away — the coral dots above the trend line are participants who scored well but reported low confidence, a pattern worth investigating.
Which participants need coaching versus re-training. The three outliers — high score, low confidence — need follow-up support to apply what they learned. Re-teaching the curriculum would be the wrong intervention. The cohort with low score and matching low confidence is a different cohort with a different need. Pathways Forward routes the two groups to different post-program supports.
Evidence the program looks at capability and confidence as distinct dimensions rather than collapsing them into a single satisfaction number. The Pearson r = 0.71 with a 95% confidence interval is the kind of statistical depth that distinguishes a Level 2 report from a Level 1 report. The outlier callout shows the program team is using the data to act, not only to report — which is what a capacity-building grant is funding.
Kirkpatrick Level 3 is the level most programs claim and almost no programs prove. Did participants apply what they learned on the job? Three months after the cohort closed, who is still using the skills, who has fallen back, and who has been promoted into a role that uses them every day. The build below sends three follow-up touchpoints — at day 30, 60, and 90 — as personalized links tied to the original learner ID, then runs AI rubric scoring against the open-ended manager observations that come back.
Personalized follow-up links tied to the persistent learner ID produce three times the response rate of bulk survey emails sent to whoever opens them. Every response from day 30, 60, and 90 lands back on the same learner record as the intake form and the post-program rubric — no name-matching, no email reconciliation. AI rubric scoring on the open-text manager observations runs at the moment the response arrives, so the report below updates in real time as more responses come in. The status track does not need to wait for all 47 participants to respond.
Which graduates need a refresher touchpoint and which are ready to mentor the next cohort. The 12% in the red band at day 90 get a coaching call. The 68% in the green band become candidates for the alumni mentor track that scales the program — and proves the capacity-building investment to the funder by showing the cohort is producing its own teaching capacity.
This is the section that decides renewal. Most training grants stop reporting at Level 2 because Level 3 follow-up was never wired in. The 68% applying at day 90 is the headline number that justifies the next two years of funding. The 38/47 manager response rate proves the follow-up architecture is working at scale. The Green/Yellow/Red distribution with a clean methodology paragraph below it is what funders increasingly require — and what most grantees cannot produce.
The four reports above are the underlying detail. The funder reads a fifth report — the synthesis — that combines the headline outcomes, the segment evidence, the behavior transfer, the baseline context, and one participant voice quote that traces back to a specific source response. The build below is the only report in this article that does not introduce new data. It is one live link, updated as the underlying reports update, with a defined narrative structure underneath.
The synthesis introduces no new data and requires no separate authoring project. It is a defined template that pulls from the four prior reports automatically. The funder opens one live URL and sees the full picture. If a number raises a question, every figure in the synthesis clicks through to its underlying report — and from there, to the individual learner record. The audit trail is the property of how the data was collected, not a separate document attached at the end.
Whether to apply for the next two-year capacity-building grant cycle now or to wait one more cohort. The synthesis above is what Pathways Forward sends with the renewal application six weeks before the current grant ends. Because the link is live, the program officer can keep checking it as Cohort 4 enrolls — moving the conversation from compliance to partnership.
Metrics and narrative in one place, not two. The four signals that decide renewal — statistical movement with sample size, behavior transfer at 90 days, methodology in plain language, and participant voice with citation — are all on the synthesis page. Every claim traces back to source. The capacity-building investment is visible structurally: the synthesis above will look the same for Cohort 4, Cohort 5, and Cohort 6 without a separate evaluation project each time. The architecture is what is being funded; the report is what proves it.
Foundation capacity-building grants typically require the live synthesis above and a separate compliance package — narrative on the funder's template, financial reconciliation against budget, and demographic disclosure aligned to the funder's categories. Federal grants add prescribed forms on top. The four cards below name what still gets submitted and which underlying report it pulls from.
The funder's own template, typically 8–15 pages of prose with embedded charts. Pulls from the synthesis (§08) for headline outcomes, the pre/post report (§05) for skill movement, the behavior report (§07) for Level 3 evidence, and the baseline report (§04) for demographic reach.
Two-year capacity-building grant means twenty-four months of budget-to-actual. This sits outside Sopact Sense — financial actuals come from the accounting system (QuickBooks, NetSuite, Sage) via API integration. The reconciliation document still gets filed; the underlying program reports above prove the dollars produced the outcomes claimed.
Whether the funder uses foundation categories or federal categories, the disaggregation was structured at intake — see §04 Cohort Baseline. The disclosure form is auto-populated from the demographic fields captured at the persistent learner ID. No retrofitting categories at deadline.
Funders increasingly require that every beneficiary count, every outcome claim, and every quote trace back to a source record. The audit trail is the property of how data was collected — not a separate document. From any number in the synthesis, the funder clicks through to the underlying report, then to the individual learner record. No appendix; the chain is the artifact.
Where Sopact Sense ends and operational tools continue. Sopact Sense is the system of record for the program evidence — every form, survey, rubric, and follow-up flows through it so the persistent learner ID is preserved end to end. The accounting system stays in place for financial reconciliation. The grants management platform (Submittable, Fluxx, Foundant) stays in place for application intake and award workflow. The donor CRM (Salesforce NPSP, Bloomerang) stays in place for gift records. Reporting is what moves to a tool designed for the orchestration.
Plain answers to the questions training-program grantees send us most often. The structured versions of these answers also appear in this page's schema, so the same content shows up in search-result rich snippets and AI Overview answers.
A capacity-building grant report is the document a grantee submits to a funder showing how a capacity-building grant was used to strengthen organizational ability to deliver outcomes — often by building evaluation infrastructure, scaling cohort programs, or developing systems that produce continuous evidence. For training-focused grantees, the report shows both program outcomes for the cohort served and the architectural investment that makes future cohorts measurable. The five-report series in this article is the structure most foundation capacity-building grants now expect.
A training program grant report follows five connected reports rather than one summary document: a cohort baseline report covering who enrolled and where they started; a pre/post skill report covering Kirkpatrick Level 2 learning outcomes; a score-confidence correlation report distinguishing capability from confidence; a 90-day behavior change report covering Kirkpatrick Level 3 on-the-job application; and a funder-ready narrative synthesis tying them together. Each report rests on a persistent learner ID assigned at intake. See the five builds above.
A persistent learner ID is a unique identifier assigned to each participant at the moment of enrollment and carried automatically through every later form, survey, rubric, and follow-up. It matters because grant reporting at Kirkpatrick Level 3 and above requires connecting a 90-day follow-up response to the same person's original intake record. Without a persistent ID, those records sit in separate tools under separate identifiers and reconnection becomes a manual analyst project that consumes 80% of evaluation time and usually fails to complete before deadline.
Kirkpatrick has four levels. Level 1 measures participant satisfaction. Level 2 measures knowledge and skill acquisition through pre/post assessments. Level 3 measures whether participants applied skills on the job, tracked through manager observations and 30–90 day follow-ups. Level 4 measures organizational results. Most training grants now require Level 3 evidence at minimum. The pre/post skill report covers Level 2; the 90-day behavior change report covers Level 3; the synthesis report ties both to Level 4 outcomes where the program is at that scale.
A funder-ready capacity-building grant report typically runs 8–15 pages of synthesis backed by underlying detail reports the funder can drill into. The five-report architecture in this article produces both layers from the same dataset — the synthesis is one live link the funder opens directly, and the four underlying reports are accessible from it. Length is no longer the constraint; depth of evidence is. Funders read the synthesis and click through to the underlying detail when a number raises a question.
Funders read training grant reports for four signals in this order. First, statistical movement with the sample size disclosed — not a single average but distribution and segment-level breakdown. Second, depth beyond completion — Level 3 evidence that participants applied skills on the job, not just that they finished the program. Third, methodology in plain language — how the pre and post were paired, what the response rate was, what was excluded. Fourth, participant voice with citation chain — quotes that trace back to the source response, not anonymous testimonials. The five-report architecture supplies all four by default.
The Learner Identity Break is the structural moment when a persistent learner record fragments across disconnected tools. The LMS assigns one ID at enrollment, the survey platform creates a new submission record, the 90-day follow-up goes out as a bulk email to whoever opens it. When the report is due, an analyst tries to reconnect these three records and finds the matching is unreliable. Preventing it requires assigning the persistent learner ID before any data is collected and carrying that same ID through every later instrument — which is what Sopact Sense does at the point of first contact. More detail on the Learner Identity Break framework.
Behavior change is measured by sending structured rubric-based observation surveys to managers and participants at 30, 60, and 90 days after the program ends — each survey linked to the same persistent learner ID assigned at intake. The rubric should specify four to six observable behaviors identified during program design, not generic questions about whether the training helped. Personalized links tied to the original record produce three times higher response rates than bulk survey emails. AI rubric scoring extracts behavior evidence from open-ended manager notes without manual coding. See the 90-day behavior change report above.
Yes — when the persistent learner ID is in place and every instrument writes to the same learner record. The five reports in this article are filtered views of the same underlying dataset, not five separate data-collection cycles. The baseline report uses intake fields; the pre/post report joins intake to post-program rubric; the correlation report adds the AI-extracted confidence rubric; the behavior report adds the 90-day follow-up; the synthesis combines all four. No reconciliation, no separate exports.
A program impact report covers what the program produced for participants. A capacity-building grant report covers both what the program produced and what the organization built — the evaluation infrastructure, the data architecture, the systems that make future cohorts measurable. Capacity-building funders are investing in the grantee's ability to keep producing evidence, not only in this cohort's outcomes. The five-report architecture in this article demonstrates both layers simultaneously: the reports prove this cohort's outcomes; the architecture proves the grantee can produce the same evidence for every future cohort.
Building the architecture takes one to two days at the start of the first cohort: designing the intake form, baseline rubric, weekly check-in, post-program rubric, and 90-day follow-up inside Sopact Sense. After that, the reports themselves take minutes per cohort because they are live views of the data, not assembled documents. The first cohort runs once and pays for the configuration; subsequent cohorts produce the same five reports automatically as data arrives.
Sopact Sense is the system of record for the program evidence — every form, survey, rubric, and follow-up is delivered through it so the persistent learner ID is preserved end to end. It connects via API or webhook to accounting systems (QuickBooks, NetSuite, Sage) for budget-to-actual reconciliation, to LMS platforms for completion data, and to CRM and grants management platforms (Salesforce NPSP, Submittable, Fluxx) for grant lifecycle metadata. The reports stay in Sopact; the operational tools stay where they are.
The five builds above are the deliverable. The pages below cover the training evaluation framework that decides which Kirkpatrick levels matter, the broader impact reporting cycle the grant report feeds into, and adjacent practices.
Kirkpatrick, Phillips ROI, CIRO, Brinkerhoff — seven methods and how to choose. The parent framework underneath every report on this page.
Read the framework → Solution · architectureThe full data architecture documented end to end. Persistent learner IDs, AI rubric scoring, live funder reports. The product layer underneath Pathways Forward.
See the solution → Sister page · adjacentSame five-report architecture broadened to multi-cohort evaluation, including longitudinal comparison across years. The next step after one capacity-building cycle.
See the sister page → Framework · upstreamThe broader six-step architecture that turns clean collection into evidence of change across all programs, not only training. The framework grant reporting sits inside.
Read the framework → Sister page · audienceThe annual organization-wide impact report that the grant report contributes to. Same evidence architecture, broader scope, different audience.
See the sister page → Sister page · audienceThe same evidence shaped for individual and major donors instead of foundation funders. Useful when development teams report on the same program.
See the sister page →A 60-minute working session. Bring a recent cohort export, a Google Form intake you want to convert, or a funder template you fill out today. We will build the persistent learner ID architecture live against your data and walk through what would change to produce the five reports above for your next cohort.
A working call, not a sales call. Camera optional, screen-share required.
A cohort intake form, a sample funder template you fill out today, or a one-paragraph description of the program you need to report on next.
A persistent-learner-ID architecture sketched against your cohort and a clear next step for the next reporting cycle.
