Data-Driven Program Improvement: Using Outcomes to Change What You Do

Data sitting in a spreadsheet does not improve anything. Data that sparks action improves everything.

Here is how to build a culture of continuous learning using outcomes data.

The monthly ops review (30 minutes)

Every month, a small team meets to review outcomes data.

Who attends

• Program director.
• Data/admin person (has the latest numbers).
• 1-2 front-line staff.
• Optional: funder or board member.

Agenda

1. <strong>Participation (5 min):</strong> How many people participated this month? On track?
2. <strong>Outcome metrics (10 min):</strong> Key metrics (confidence pre/post, job placements, engagement rate). Trend up, down, or flat?
3. <strong>Qualitative feedback (10 min):</strong> What are participants saying? Any themes?
4. <strong>Quick fixes (5 min):</strong> Any small improvements to try this month?

Example metrics to review

• Attendance rate: 85% of enrolled participants attending (drop-off?)
• Confidence increase: average +1.5 points (better than last month?)
• Job placement rate: 50% placed within 3 months (on target?)
• Participant satisfaction: 4.2/5 (strong feedback).
• Feedback themes: 12 mentions of "great mentoring", 3 of "logistics confusing".

The quarterly strategic review (90 minutes)

Every quarter, a deeper look at patterns and bigger decisions.

Participants

• Leadership team.
• Program director.
• Finance/data lead.
• Board representative (optional but useful).

Agenda

1. <strong>Quarterly summary (10 min):</strong> How did the quarter go? Overall progress toward annual goals?
2. <strong>Outcomes deep dive (20 min):</strong> Break down data by cohort, demographic, or program type. Are some groups doing better?
3. <strong>Cost per outcome (10 min):</strong> Money spent ÷ outcomes achieved. Is this efficient? Could we reallocate?
4. <strong>Feedback analysis (15 min):</strong> Themes from participant feedback. What will we change?
5. <strong>Decisions (20 min):</strong> What will we stop, start, or scale based on this data?
6. <strong>Action plans (15 min):</strong> Who owns each change? Timeline?

The data dashboard (simple version)

A one-page dashboard summarizes key metrics for quick review.

Essential columns

• Metric name.
• Q1 actual.
• Q1 target.
• Q2 actual.
• Q2 target.
• Trend (arrow up, flat, down).
• Owner.

Example metrics

• Participation: 200/250 target = 80%. Trend: stable.
• Confidence increase: +1.8/+1.5 target = 120%. Trend: up.
• Job placement rate: 45/50 target = 90%. Trend: up.
• Participant satisfaction: 4.2/4.0 target = 105%. Trend: stable.

Decision framework: what to change

Rule 1: Stop what is not working

If an activity is not driving outcomes (e.g., group coaching has no measurable impact but costs £5k/quarter), stop it.

Reallocate to high-impact activities (mentoring, 1-on-1 coaching).

Rule 2: Scale what is working

If job interviews are the bottleneck (80% of participants struggle with interviews), double down.

Add more interview practice. Hire interview specialists. Measure impact of the increase.

Rule 3: Test before scaling

Do not roll out a big change to everyone. Test with 20-30 people first. Measure. Then decide to scale.

Rule 4: Keep what is hard to change but working

Some things cost a lot but drive massive value. Mentoring is time-intensive but moves the needle. Keep it.

The learning cycle: monthly iterations

Month 1

• Observation: participants struggle with job interviews (feedback theme).
• Hypothesis: more interview practice will increase placement rate.

Month 2

• Action: add 2 mock interviews per participant (small cost).
• Measurement: track interview prep hours and placement rate.

Month 3

• Result: placement rate up from 40% to 55%.
• Decision: keep mock interviews; expand to all participants.

Months 4-12

• Repeat: identify next bottleneck, test, measure, iterate.

Communicating data-driven improvements

To funders

"We measured outcomes monthly. We found job interviews were the bottleneck. We added interview coaching. Placement rate improved from 40% to 55% in one quarter. We are scaling this model to all 10 cohorts."

To team

"Our data shows mentoring is the most valued element (9/10 satisfaction vs 6/10 for group sessions). We are investing more here and scaling back group sessions."

To board

"Cost per job placement dropped from £400 to £220 through data-driven improvements. We are now the most cost-effective provider in the region."

Common pitfalls

• <strong>Measuring but not acting.</strong> If data shows a problem and you do not act, trust erodes.
• <strong>Paralysis by analysis.</strong> You do not need perfect data to decide. 60% confidence with action beats perfect data with inaction.
• <strong>Change fatigue.</strong> Do not change everything at once. One or two improvements per quarter.
• <strong>No baseline.</strong> If you did not measure before your change, you cannot claim credit. Always measure before.

Good organizations measure. Great organizations measure and act on what they learn.

This guide is part of our Charity Impact and Strategy hub, where you can explore every practical guide in this area.