CIPP Model of Evaluation: 4 Parts, Examples and Templates

Set up your branded LMS and deliver training with AI

Start Free

Here is a made-up but typical case. An onboarding programme for new aged-care support workers runs at three residential sites. At the end of the first quarter the report says 96% of new starters finished it. Then a site manager mentions that night-shift starters at one site never did the buddy shifts the programme depends on. The completion number was true. It arrived too late to fix anything.

The CIPP model (context, input, process, product) is Daniel Stufflebeam's framework for checking the need, the plan, the delivery and the results of a training programme or a school or university curriculum, so problems like this surface while you can still act on them.

I'm an instructional designer at Coursebox, where we make an AI LMS with learner progress reporting, so I see what training teams can check while a programme runs and what they only find out at the end. I'll use the same onboarding programme the whole way through, alongside real evaluations from five countries.

What is the CIPP model?

The CIPP model is a framework for evaluating a programme through four parts: context, input, process and product. Daniel Stufflebeam developed it in the 1960s, and it has since been used for curricula, health professions education, workplace training and community projects. Its main theme, in Stufflebeam's words, is that evaluation's most important purpose is "not to prove, but to improve".

His 2007 CIPP checklist frames the model around four questions.

• What needs to be done?
• How should it be done?
• Is it being done?
• Did it succeed?

You do not have to run all four, or run them in order. The checklist says the parts "may be employed selectively and in different sequences and often simultaneously". A team inheriting a programme that is already live might start with process, while a team asked to design something new starts with context.

Where the CIPP model came from, and what has changed

The model grew out of a practical problem. In 1965 the US Elementary and Secondary Education Act began funding programmes for disadvantaged students, and Congress attached evaluation requirements to the money. The Ohio State University Evaluation Center, working with the Columbus, Ohio public schools, developed CIPP to meet them. Stufflebeam, who died in 2017, kept revising it for the rest of his career.

For training teams, the product questions in the 2007 checklist are particularly useful: reaching the intended people, meeting their needs, sustaining gains and adapting successful practice elsewhere. They expand a simple end-of-course results check into a review of the programme's continuing contribution.

The four parts, worked through one programme

I'll use one programme throughout: onboarding for new aged-care support workers who start at any of three residential sites run by the same provider. Each site takes about 10 new starters a month. The programme is an online course, two buddy shifts with an experienced worker and a competency sign-off by a registered nurse. The example is hypothetical and every number in it is illustrative.

Context evaluation: what is needed, and why?

The 2007 checklist defines context evaluation as assessing "needs, assets, and problems within a defined environment". In practice it checks whether the programme's goals match what people need. In my experience it is the part teams skip most, because by the time someone asks for training they have usually decided what the training is.

• What to ask: what do new starters need to do safely in their first month, which mistakes cause the most harm, and does that differ between sites?
• Where the evidence lives: incident reports, exit interviews with people who left in their first 90 days, and interviews with shift supervisors and a few recent starters. A training needs analysis gives you a fuller method.
• The worked example: incident reports show most first-month incidents involve manual handling and medication prompts. Exit interviews add something the reports do not: starters at Site C feel unsupported on night shift, when fewer senior staff are on. That changes the goals. The programme now needs a night-shift buddy as well as more content.
• The decision it informs: the programme's goals and priorities.

Case study: Public hospital doctors in South Korea, where context evaluation changed the course

South Korea's National Medical Centre ran an ultrasound training programme for doctors in public hospitals and evaluated it with CIPP using data from 2017 to 2019.

• Context: a needs survey identified lack of work cover as an attendance barrier. Its percentages describe survey selections, not the proportion of all public-hospital doctors.
• Input: the course was cut from five days to two.
• Process: demand exceeded the planned places, with 45 doctors taking part against 24 planned.
• Product: self-efficacy scores rose from 3.06 to 7.9 out of 10, and the review committee kept the programme.
• The lesson: asking about barriers helped the team choose a shorter, more accessible format. Self-efficacy improved, but the authors did not measure long-term effects on patient care.

Input evaluation: which plan, with what resources?

Input evaluation "assesses competing strategies and the work plans and budgets of the selected approach". It compares the ways you could meet the need and judges which one you can deliver with the budget, people and time you have. It is where you find out that a good plan cannot be staffed before you have built it.

• What to ask: which two or three delivery options could meet the goals, what does each cost, and do we have the people to run it at all three sites?
• The worked example: the training lead costs three options: a classroom induction at head office, a fully online course, or an online course plus buddy shifts at each site. Roster data shows Site C cannot release an experienced worker for night buddy shifts without overtime. The third option is adopted with a budget line for that overtime.
• The decision it informs: which plan to adopt and fund.

In our support inbox, input evaluation shows up as a request to try before committing. Several training teams asked to pilot before they signed up, which is the right instinct: test the plan on a small scale before you fund all of it.

Process evaluation: is it running as planned?

Process evaluation is how you "monitor, document, and assess program activities". It checks whether the programme is being delivered as designed and records what is going wrong while there is still time to fix it.

• What to ask: is every site delivering the online course, the buddy shifts and the sign-off as planned? Where are people getting stuck?
• Where the evidence lives: course progress reports, buddy-shift logs, the sign-off register and a short weekly note from each site coordinator.
• The worked example: in week two of the first cohort, progress data shows Site C starters finishing the online course, but the buddy-shift log shows half of them have not done a night buddy shift. The coordinator moves two shifts in the roster that week. If the team had waited for the end-of-quarter report, a whole cohort would have started nights without the support the context evaluation said they needed.
• The decision it informs: fixes during delivery.

The questions our support team gets about live programmes are practical ones: how to nudge people who have not started or finished, and how assessors keep on top of the marking queue. Both are process evaluation, even if nobody calls it that.

These events show that creators changed courses soon after launch; they do not tell us why, whether learners prompted the changes, or whether the edits helped. Use an early process check to investigate what learners and facilitators are experiencing, and keep a change log that records the reason, action and follow-up.

Case study: Workplace language training at a factory in Morocco

Researchers evaluated a Chinese-language programme for local staff at a Chinese-managed factory in Morocco with CIPP, from needs analysis through mid-course and post-course checks.

• Process: about 35% of learners attended 30% of sessions or fewer. The supervisor said "shift schedules and production arrangements easily disrupt learning continuity".
• Product: attendance correlated strongly with the composite performance score among 40 learners (Spearman's ρ = 0.90). Attendance itself counted for 10% of that score, so part of the association is built into the measure.
• The lesson: the supervisor blamed shift schedules for the gaps in attendance, and attendance tracked closely with results. This is an association, not proof that attendance caused better learning; the study did not directly measure workplace transfer.

Product evaluation: did it work, for whom, and will it last?

Product evaluation judges the outcomes. The 2007 checklist splits it into four questions:

• The worked example: impact shows that agency workers who became permanent were never enrolled because they skipped the HR start process. Effectiveness compares day-30 sign-offs and observed practice with the goals agreed before launch. Sustainability needs later evidence that those gains continue, alongside a deputy and an ongoing budget so support does not depend on one coordinator. Transportability becomes relevant when the provider considers a fourth site.
• The decision it informs: continue, change, scale or stop.

Stufflebeam calls transportability "an optional component", so skip it when nobody plans to reuse the programme.

Case study: A research track for junior doctors in Ireland

Ireland's Academic Internship Track gives first-year doctors protected research time. Its evaluation used all four parts, and product evaluation went well past satisfaction.

• Input: the design team reviewed similar programmes overseas, such as the UK Academic Foundation Programme.
• Process: 100% of posts were filled and none of the interns dropped out.
• Product: 41 of 72 former interns (57%) had published research, and the evaluation tracked career intentions and whether they stayed in Ireland.
• The lesson: effectiveness and sustainability questions need follow-up after the programme ends. The authors run the programme and had no comparison group, which they note as limits.

Formative and summative: two uses of the same four parts

CIPP serves two purposes. Run during a programme, it is formative: it helps you improve the goals, plan and delivery. Run at the end, it is summative: it judges the programme's merit and worth and reports on it. In 1971 Stufflebeam described these as "proactive evaluation to serve decision making and retroactive evaluation to serve accountability" (Stufflebeam, 1971). The same four parts do both jobs.

How to run a CIPP evaluation in six steps

1. Agree who the evaluation is for. Name the people who asked for it, the decisions it should inform, and who gets findings and when. Stufflebeam says CIPP evaluations should rest on "explicit advance agreements with the client".
2. Choose the parts you need. A new programme starts with context and input. A live programme with problems starts with process. A programme up for renewal needs product.
3. Write two to four questions per part. Each should connect to a decision. If no decision depends on the answer, drop the question.
4. Collect evidence from more than one group, and from records and observation. Learners, supervisors, facilitators and funders see different things.
5. Report early and often. Share process findings in short updates while the programme runs, with a named owner for each action.
6. Review the evaluation itself. Ask someone outside the programme where the evidence is thin before you write the final report. Stufflebeam calls this a metaevaluation.

The plan template follows these steps.

What a learning platform can and cannot tell you

A CIPP evaluation belongs to the training team, but the course can be built to collect some of its evidence. On 6 October 2026 I tested that in Coursebox, the course builder we make, with the aged-care example. I asked for an onboarding course for new support workers across the three residential sites, told it the training team was running the CIPP evaluation separately, and asked for only two kinds of evidence inside the course: a short question at the end of each module on what was unclear or hard to apply on shift, and a final practical check against the care standard.

It kept the evaluation planning out of the learners' course, as asked, and built the onboarding content around the work: Safe Transfers, Infection Control, Care Notes, Site Help Pathways and Putting Care Together, with scenarios such as a blocked transfer route and a resident who hesitates, plus a Final Practical Check. The first outline left out the end-of-module question, so I asked for it in one message. The revision added a "Shift Reflection" page at the end of each of the five topic sections with the question word for word, and changed nothing else.

The evaluation now gets process evidence in learners' own words for each part of the programme, and product evidence from the practical check. The training team still reads it alongside the delivery plan and the site records.

The platform can supply useful process evidence, such as starts, completion and assessment results, and part of product evidence. Those records need interpretation alongside the delivery plan and intended audience. Context and input usually also need incident records, budgets, rosters and stakeholder conversations. A platform may store or integrate these records; it cannot replace the judgement required to evaluate them.

Using CIPP for curriculum evaluation

The CIPP evaluations I found in journals are mostly of curricula in universities and health professions education. A 2021 systematic review of CIPP in medical education found 41 studies, mostly in nursing and medicine. The four parts work the same way for a curriculum; the evidence changes. Context looks at graduate outcomes and what employers or accreditors expect, input at the syllabus, staffing and facilities, process at what happens in classes and placements, and product at results, portfolios and where graduates go.

Case study: An obstetrics and gynaecology residency in Laos

The University of Health Sciences in Lao PDR evaluated its OB-GYN residency, which had run since 2003 without a proper evaluation, using CIPP questionnaires with lecturers and residents (response rate 97.5%) and interviews.

• Context: scored lowest, with learning outcomes that overlapped.
• Input: a lack of classrooms, a skills lab and staff.
• Product: lecturers rated outcomes as achieved more highly than residents did (3.57 vs 3.14).
• The lesson: asking two groups showed a gap neither would have shown alone. COVID-19 ruled out face-to-face interviews, which the authors list as a limit.

The question bank has a curriculum column next to the workplace one for each part.

Common mistakes with the CIPP model

• Treating a four-section questionnaire as a CIPP evaluation. A survey with context, input, process and product questions is one data source. CIPP expects records, observation and interviews as well.
• Asking only one group. Learners, trainers and managers often disagree, and the disagreement is a finding.
• Looking only at final results. Starting with product can be sensible for a live programme up for renewal. The mistake is judging results without examining whether its goals, resources and delivery were appropriate.
• Collecting findings nobody acts on. Every finding needs an owner who can change the programme.

Case study: In-service training in Iran, where needs assessment was rated poorly

Shiraz University of Medical Sciences evaluated its employee in-service training with CIPP, surveying 250 staff. Needs assessment of the courses scored 2.47 out of 5 across 246 responses, among the lower-rated input items, and the authors found "many attend these courses with sole purpose of receiving educational certificate". A training calendar can run on schedule and still miss what people need.

Limitations of the CIPP model

• It takes time and people. Stufflebeam's own checklist suggests using on-site observers because keeping lead evaluators on site "would be too costly". A 2019 guide on executing CIPP in health education notes it is "rather complex to be fully implemented".
• It does not supply a ready-made instrument for your programme. CIPP gives you questions and a structure; you still need suitable measures, explicit judgement criteria and evidence sources. A questionnaire may contribute useful evidence, but it cannot answer every question.
• Attribution is not built in. Product evaluation can show that outcomes changed, but a comparison group or trend alone does not prove the programme caused them. You also need a credible design, consistent measures and checks for other changes that could explain the result.

For a small team, start with a proportionate plan: check the need before design and when circumstances change, compare two feasible options, and review delivery during the first cohort. Schedule outcome follow-up around when the intended change should be observable; 90 days may suit some programmes, but it is not a CIPP rule.

CIPP vs Kirkpatrick and other evaluation models

The main difference is scope. The Kirkpatrick Model evaluates the outcomes of training at four levels: reaction, learning, behaviour and results. CIPP evaluates the whole programme, including whether it was the right programme and whether it was delivered as planned. Some evaluators use the two together, with Kirkpatrick's levels inside CIPP's product evaluation.

I use CIPP when a programme is being designed or redesigned, Kirkpatrick for the results inside product evaluation, a broader checklist for how to evaluate the effectiveness of a training program when I need a quick review, and the Brinkerhoff Success Case Method when results are mixed and I need to know why. If finance asks for a percentage, the training ROI calculation sits on top.

Use CIPP to make the next programme decision

CIPP is most useful when you need to decide whether a programme addresses the right need, which approach to fund, what to change during delivery and whether to continue it. Its value comes from connecting evidence to those decisions. Completing four evaluation sections without a decision to make can leave you with a report that changes very little.

Before the next cohort, write down the decision you need to make, the criterion you will use and who will act on the findings. Choose proportionate evidence for each relevant lens and schedule it while changes are still possible. When outcomes improve, explain what the evidence supports and what other factors may have contributed. That gives the team a defensible reason to adapt, continue or stop the programme.