calendar
Edited on October 7, 2026

Understanding by Design (UbD): 3 Stages of Backward Design

Understanding by Design explained: the three stages of backward design with a worked example, GRASPS performance tasks, real results and two templates.

Carolina Martin, Learning Experience Manager at Coursebox AI
Carolina Martin
Customer Success Manager
Key takeaways

Understanding by Design (UbD) is Grant Wiggins and Jay McTighe's framework for planning units and courses that build understanding and transfer, using backward design: a three-stage process that sets the desired results first, then the evidence of learning, then the learning plan.

  • Backward design is the planning process inside Understanding by Design, so the two names describe one framework and its core method.
  • The first edition of Understanding by Design came out in 1998, and the expanded second edition from 2005 is still the current one.
  • In backward design, the performance task is written before any lesson, so every activity has to prepare people for it.
  • Studies of backward design in staff training come from health care and education, and a 2025 review found better alignment but no outcome evidence.
Is your course planned backward?

Five planning prompts for one course. This is an unvalidated checklist. Review each answer separately; the total cannot establish alignment, readiness or effectiveness.

1
2
3
4
5
Can you say what people should do on the job, on their own, after the course?
Instructional Designers

Go from storyboard to an interactive course in minutes

Picture a casual-dining restaurant chain (a hypothetical example) that promotes its best server to shift supervisor. Her food safety training was a deck of temperature charts and a 20-question quiz, and she passed it. Three weeks later, on a busy Friday night, the walk-in fridge reads 9°C and two large bookings are about to arrive. Nothing in the course ever asked her what to do next.

Understanding by Design plans training from that Friday night backward. I'm an instructional designer at Coursebox, where we make an AI curriculum generator that builds courses from a designer's own plan. That gives me a stake in how courses get planned. I've used backward design on workplace courses, and the supervisor course runs through every stage here.

What is Understanding by Design (UbD)?

Understanding by Design (UbD) is Grant Wiggins and Jay McTighe's framework for planning units and courses that build understanding and transfer, using backward design: a three-stage process that sets the desired results first, then the evidence of learning, then the learning plan.

Understanding by Design diagram shows desired results, assessment evidence and the learning plan in backward planning order, with alignment and revision across all three stages.

In the authors' words, the framework "offers a planning process and structure to guide curriculum, assessment, and instruction", and its "two key ideas are contained in the title: 1) focus on teaching and assessing for understanding and learning transfer, and 2) design curriculum "backward" from those ends" (McTighe and Wiggins, ASCD white paper, 2012).

Backward design is the three-stage process. UbD is the wider framework around it, with transfer goals, essential questions, six facets of understanding, GRASPS performance tasks and the WHERETO checklist for the learning plan. Transfer is the authors' word for using what you learned, on your own, in a new situation: "The bottom-line goal of education is transfer." At work, that means doing the job well on a shift the course never showed.

Our example is a hypothetical restaurant-chain course for new shift supervisors. They need to respond to a cold-storage failure during a busy service using the chain's food-safety procedure. Backward design fits because recalling temperature limits alone will not show whether they can make and document the decision.

StageThe authors' key questionWhat you writeIn the food-safety course
1. Identify desired resultsWhat is the ultimate transfer we seek as a result of this unit?A transfer goal, understandings, essential questions, knowledge and skillsSupervisors keep food safe during a busy service and make the right call when something goes wrong
2. Determine acceptable evidenceHow will we know if students have achieved the desired results?A performance task, its criteria and other evidenceThe Friday-night fridge task, plus a short check on temperature limits and the log
3. Plan learning experiences and instructionWhat activities, sequence, and resources are best suited to accomplish our goals?A sequence of lessons and practice that builds to the taskShort lessons and practice scenarios, with the Friday-night task last
Teachers and children gathered in a classroom.
Pavel Danilyuk / Pexels

Case study: Preschool teacher development in Türkiye

A blended programme for 68 preschool teachers from 26 Turkish cities used Understanding by Design to support applying self-regulation concepts in classrooms. Its five modules covered executive functions, metacognition, emotion regulation and motivation regulation. Pre/post tests, surveys and focus groups informed the evaluation. The authors reported increases in teachers' use of metacognitive and emotional-motivational regulation practices. The available abstract gives no effect sizes and does not establish which component caused the changes. The useful planning move is to define classroom application before choosing the workshops and online activities. Read the teacher-development study for its reported findings.

The stage names come from the 2005 book. The 2012 white paper shortens them to Desired Results, Evidence and Learning Plan.

People in scrubs and white coats talking in a hospital corridor.
RDNE Stock project / Pexels

Case study: Emergency nursing education in Australia

HIRAID's education programme used backward design, constructive alignment and scaffolded learning within a wider implementation strategy. It is a related backward-design example; the source does not identify the entire programme as an implementation of UbD.

  • Setting: 1,377 nurses in 29 metropolitan, regional and rural emergency departments, June 2020 to February 2024, tested in a stepped-wedge cluster randomised trial.
  • What they did: 15 instructor courses for 162 participants between 2021 and 2023, as one part of a wider strategy of 21 behaviour change techniques (PLoS One, 2025).

Across 106,047 admissions, the trial reported lower odds of a rapid response call within 72 hours (adjusted odds ratio 0.92, 95% confidence interval 0.81 to 0.99). That result concerns the full HIRAID implementation, including its behaviour-change strategy. It cannot isolate backward design. Read the HIRAID trial alongside the education programme report.

Who created Understanding by Design, and what to read

Grant Wiggins and Jay McTighe published Understanding by Design with ASCD in 1998, and the expanded second edition followed in 2005. They never claimed the idea was new: the 1998 edition credits Ralph Tyler, who "described the logic of backward design clearly and succinctly" in 1949. Backward design also differs from "backward mapping", Richard Elmore's 1979 term for planning how a public policy gets implemented. Wiggins died on 26 May 2015, aged 64 (Education Week). McTighe still runs UbD workshops and consultations. ASCD owns the Understanding by Design and UbD trademarks.

  • Start with the 2012 white paper (ASCD). It is free, 13 pages long, and covers the three stages, the six facets and the authors' answers to common questions.
  • Read the expanded 2nd edition (2005) (ASCD) for the full argument. It is the current edition. There is no third edition, and citations dated 2004 or 2006 point to this same book.
  • Plan with the 2011 guide, The Understanding by Design Guide to Creating High-Quality Units. It introduced version 2.0 of the planning template, which McTighe compares to "a software upgrade" (ASCD).
note:Older copies use WHERE

The free "Chapter 1" PDF that circulates online comes from the 1998 first edition. Its learning plan acronym is WHERE; the 2005 edition uses WHERETO, with seven letters.

What are the three stages of backward design?

Keep that restaurant course in mind through the three stages below. Its target is an independent, documented response to a fridge failure, with food safety taking priority over continuing service. The scenario must supply the temperature history, affected stock and approved chain procedure so learners have enough information to make the decision.

The three stages are identify desired results, determine acceptable evidence, and plan learning experiences and instruction. You work through them in that order, and each stage asks one question of the course.

Stage 1: Identify desired results

Stage 1 sets what people should be able to do on their own after the course, and what they need to understand to do it. Template 2.0 splits it into three kinds of goal: transfer, meaning and acquisition. Meaning covers the understandings and the essential questions, which the authors say should "provide focus to all learning". Acquisition covers the knowledge and skills, and that is where measurable learning outcomes belong. The authors keep acquisition in its place: "Acquisition of content is a means, in the service of meaning making and transfer."

  • Key question: what should people be able to do on their own, on the job, after the course?
  • What you write: a transfer goal, one or two understandings, an essential question, and the knowledge and skills they depend on.
  • Food-safety course: supervisors keep food safe during a busy service and make the right call when something goes wrong. The understanding is that food safety checks slip at handovers and under time pressure, so they must be built into the routine. The essential question is "What would you check first if service gets out of control?"
Hands being washed at a sink beside a box of medical gloves.
https://kaboompics.com/ / Pexels

Case study: Infection prevention training across Africa

ICAP at Columbia University, working with ministries of health, used backward design to build infection prevention training in the first months of the pandemic. The team wrote that it "emphasized the need to clearly articulate learning goals beyond what health workers need to know to what they need to know how to do".

  • Setting: 8,797 health workers at 945 facilities in countries including Kenya, Malawi and Sierra Leone, June to October 2020, trained in person, online or self-guided.
  • What they did: identified the competencies first, then decided how each one should be taught and assessed.
  • Result: The infection-prevention training report found a 15-percentage-point rise in knowledge-test scores for the 2,370 trainees with paired tests. It did not assess behaviour change on the job. The programme used backward design; these results do not isolate that approach or establish an effect of UbD specifically.

Stage 2: Determine acceptable evidence

Stage 2 specifies the evidence before the learning plan. The authors describe performance tasks as culminating performances for a unit, supported by quizzes, observations and work samples. In this example the Friday-night task comes at the end, with practice and formative checks along the way. The three stages describe the designer's planning order, not three compulsory sections of a learner's course.

Wiggins and McTighe also name six facets that show someone understands: "the capacity to explain, interpret, apply, shift perspective, empathize, and self-assess". The facets were "never intended to be a hierarchy", which sets them apart from Bloom's taxonomy, and the authors recommend using only the facets that give evidence of the understanding you want. For the supervisors, that is apply (make the call) and explain (write down why in the log).

  • Key question: how will you know people can do it, in a situation like the real one?
  • What you write: a performance task, the criteria you will judge it by, and other evidence such as a quiz or an observation.
  • Food-safety course: on a Friday night the walk-in fridge reads 9°C and two large bookings are arriving. The supervisor decides what to do about the stock, the menu and the team, and records it in the food safety log. A short check on temperature limits and the log covers the facts.

In a sweep of our support threads in October 2026, "backward design" and "essential questions" returned no relevant threads. The nearest themes start from the end point: 12 threads from about 10 customers, mostly Australian training providers, about mapping courses to units of competency, and 7 threads from 7 customers about rubric criteria driving AI marking. Writing the criteria first is Stage 2 thinking, and a rubric is a good way to record it.

Stage 3: Plan learning experiences and instruction

Only now do you plan the lessons. The test for each activity is whether it prepares people for the task. The authors suggest coding every learning event T, M or A, for transfer, meaning making and acquisition, "to ensure that all three goals are addressed in instruction". WHERETO is their checklist for the sequence: where the unit is going and what is expected, hook, equip, rethink and revise, evaluate, tailor and organise.

  • Key question: what will people need to learn and practise to succeed at the task?
  • What you write: a sequence of lessons and practice, each coded T, M or A.
  • Food-safety course: the essential question opens the course (M). Short lessons on temperature limits, handover checks and the log follow (A), then practice scenarios under time pressure (M and T), and the course ends with the Friday-night task (T).
tip:Plan the whole course, then cut it into lessons

Wiggins and McTighe design at the unit or course level, because understandings, essential questions and transfer tasks are "too complex and multifaceted to be satisfactorily addressed within a single lesson". If you build microlearning, plan the whole programme backward first, then cut it into short pieces.

How do the three stages fit together?

The stages have to line up. In the authors' words, "the Stage 1 content and understanding must be what is assessed in Stage 2 and taught in Stage 3." Alignment is how backward design avoids what Wiggins and McTighe call the "twin sins" of course design: activity-oriented design, which they describe as "hands-on without being minds-on", and coverage, where learners "march through a textbook, page by page".

Both sins are familiar at work. The first is the module with a game nobody can connect to the job; the second is the policy walkthrough read out slide by slide. In the food-safety course, the understanding about handovers and time pressure is what the Friday-night task tests, and the handover lesson and the pressure scenarios are where it is taught. If a lesson cannot point to something in the task, it goes.

warning:One workshop is not enough

McTighe lists ways to kill UbD, including providing "one introductory presentation/workshop on UbD" and assuming people can now use it well, and requiring the template for "each and every lesson" (McTighe, 2016). When you hand backward design to subject matter experts, give them practice, feedback and planning time.

How to use backward design to plan a training course

Work through these steps in order, and loop back whenever Stage 2 shows that a goal is too vague to assess.

  1. Write the transfer goal: what people will do on their own, on the job.
  2. Write one or two understandings as full sentences, starting "Learners will understand that...".
  3. Write an essential question that people will come back to during the course.
  4. List the knowledge and skills the goal depends on.
  5. Write the performance task with GRASPS, and the criteria you will judge it by.
  6. Add other evidence, such as quick checks, observations or a short quiz.
  7. Plan the learning activities, code each one T, M or A, and check the sequence with WHERETO.
  8. Check alignment: everything in Stage 1 is assessed in Stage 2 and taught in Stage 3.

The authors' summary of the template closes Stages 2 and 3 with an alignment question: "Are the assessments aligned to all Stage 1 elements?" and "Are the learning events in Stage 3 aligned with Stage 1 goals and Stage 2 assessments?"

The food-safety course, planned backward

FieldFood-safety supervisor course
Transfer goalOn their own, supervisors keep food safe during a busy service and make the right call when something goes wrong
UnderstandingFood safety checks slip at handovers and under time pressure, so they must be built into the routine
Essential questionWhat would you check first if service gets out of control?
Knowledge and skillsThe chain's temperature limits, what a complete log entry contains, and how to run a handover check
Performance taskAs shift supervisor, decide what to do about the stock, the menu and the team when the walk-in fridge reads 9°C with two large bookings arriving, and record it in the food safety log
Other evidenceA short check on temperature limits and the food safety log
Learning planThe opening question, short lessons, practice scenarios, then the Friday-night task, in about 40 minutes
Backward design planning template

Plan all three stages with a question for every field, code your learning plan, and check alignment and WHERETO, with the food-safety supervisor course filled in. Editable Word file, no email required.

Write the performance task with GRASPS

GRASPS is McTighe and Wiggins's frame for writing a performance task as a realistic scenario: goal, role, audience, situation, product or performance, and standards. Their own example is already a workplace one: "You are an engineer in the packaging department of the M&M Candy Company", pitching a container design to "non-engineer company executives" (McTighe and Wiggins, GRASPS design sheets).

GRASPSThe Friday-night task
GoalProtect food safety and decide whether and how service can continue
RoleYou are the shift supervisor on a Friday night
AudienceYour kitchen and floor team now, and the venue manager who reads the log in the morning
SituationThe fridge reads 9°C before two bookings arrive. The task provides temperature logs, affected stock and the chain's approved response procedure.
Product or performanceDecisions on the stock, the menu and the team, and an entry in the food safety log
StandardsFood at risk is handled under the chain's rules, the menu changes before orders are taken, the team knows the plan, and the log entry is complete
GRASPS performance task builder

Write a workplace performance task with all six GRASPS elements, set the criteria you will judge it by, and check it against Stage 1, with the Friday-night food-safety task filled in. Editable Word file, no email required.

Which stage of backward design gets skipped most?

In course requests, it is Stage 2. I looked at 103 free-written course requests typed into Coursebox in the 90 days to 7 October 2026, one per person, sorted by an AI classifier and spot-checked by hand. About 4 in 10 (42) said what learners should be able to do by the end. At most about 1 in 5 (20) described how learners would show it, with a named task, criteria or product.

1 in 8
course requests that asked for a real-world performance task, at most 13 of 103 free-written requests typed into Coursebox in the 90 days to 7 October 2026
Source: Coursebox course requests, PostHog, October 2026

The 20 and the 13 are upper bounds, because the classifier counted generously. Goals show up far more often than the evidence that would prove them, and that is the gap backward design closes. If you take one habit from the framework, write the performance task before the first lesson.

Understanding by Design vs ADDIE and Dick and Carey

UbD is a design logic for a unit or a course. ADDIE is a five-phase pattern for running a training project, and the Dick and Carey model is a ten-step systems model, first published in 1978, built on instructional analysis and tryouts with learners. They work together: run the project with ADDIE, use backward design in its design phase, and bring in Dick and Carey when you need formal task analysis and formative evaluation.

FeatureUbD (backward design)ADDIEDick and Carey
What it isA framework for planning units and courses for understanding and transferA five-phase pattern for running a training projectA ten-step systems model from one textbook
Starts fromThe transfer goal and the evidenceThe needs analysisThe instructional goal and a skills analysis
Best forCourses where people must use judgement in new situationsA shared project structure for any trainingStable, high-stakes procedures
Weak spotLittle guidance on needs analysis or project managementLittle guidance on setting goals and evidenceSlow, with detailed analysis and three tryout rounds
Use it whenYou know the job outcome and need the course to prove itYou need phases, roles and sign-offsErrors are costly and you can test drafts with learners

UbD is also confused with UDL. Universal Design for Learning is a framework for making learning accessible to people with different needs. It works alongside UbD at Stage 3, when you plan how people will learn.

Does backward design work? Limitations

Backward design improves alignment, and the evidence on outcomes is thinner. A 2025 rapid review of its use in health professional education found that it "better assured the alignment of course learning outcomes, assessments, and learning activities", and also that "There was no empirical evidence of the efficacy of BWD in achieving competency-based outcomes" (Rinaldi et al., Journal of Professional Nursing, 2025). Programmes designed backward have reported results, from knowledge gains in African health facilities to fewer rapid response calls in Australian emergency departments, but in each one backward design was part of a larger effort.

  • Time. The authors acknowledge that "it takes time to develop high-quality curriculum using the UbD framework".
  • Old habits. Many teachers have told the authors that the process "feels awkward because it requires a break from comfortable planning habits."
  • Built for whole courses. The authors built it for units and courses and advise against using it to plan single lessons, so a short module planned on its own is a poor fit.
  • Open-ended learning. When there is no defined end goal, such as ongoing reading in a field, there is less to plan backward from.
  • Workplace evidence. I found no private-sector company case with measured results. Published evidence comes from health workers, hospital staff and educators, and much of it measures knowledge, which is level 2 of the Kirkpatrick model.

A US university ran a seven-week course design institute for 12 health science educators, built on backward course design, and scored the course blueprint each one produced (Speer et al., BMC Medical Education, 2022). The study assessed the blueprints participants produced. It does not establish how quickly subject-matter experts generally become competent designers.

16.57 / 20
average rubric score on the course design blueprints of 12 US health science educators after a seven-week backward course design institute
Source: Speer et al., BMC Medical Education, 2022

The study's authors call the small sample "one of the major limitations", and participants named time as a common challenge, the same warning Wiggins and McTighe give.

Building a backward design course in Coursebox

On 7 October 2026 I gave Coursebox my three stages for the food-safety supervisor course. It is our product, so weigh my read with that in mind. Stage 1 had the transfer goal, the understanding about handovers and time pressure, and the essential question "What would you check first if service gets out of control?" Stage 2 had the Friday-night task and a short check on temperature limits and the log. Stage 3 asked for short lessons and practice scenarios that build to the task, with the task as the final assessment, in about 40 minutes.

Coursebox said it would build the course around temperature control, handover checks, food-safety logging, practice scenarios and the Friday-night performance task. The first outline included those elements and placed the task at the end, as requested. I checked the outline against the desired results, planned evidence and learning activities.

It left out the essential question. The question had no page of its own and did not come back in the final task, so the idea meant to tie the course together was missing.

Two short follow-up messages fixed it. The first asked for the question to open the course and come back in the final task, with Australian spelling throughout. The second asked Coursebox to trim the course to the 40 minutes I had asked for, keeping the essential question page, the temperature and log check, and the final task. The finished outline ran about 40 minutes, with 4 sections, 10 pages, 1 quiz and 1 assignment. It had an Essential question page straight after the welcome, kept pages such as Temperature limits in practice and Handover and log routine, and made the Friday-night performance task the last assessment, followed by a short Summary section. Those changes brought the outline closer to the brief. I checked the page structure and estimated time, not the finished task, its marking criteria or learners' ability to transfer the skill.

Coursebox course outline for a food-safety supervisor course, about 40 minutes over four sections, opening with an essential question page and ending on the Friday-night performance task

Putting the model to work

Write the transfer goal and performance criteria before you add another lesson. Give learners the information and practice they need to meet those criteria, then try the task with a small group and revise what does not work. The value of backward design comes from the alignment you can show between the goal, evidence and learning plan. Completing a template or getting an outline in the right order is only the start.

Frequently Asked Questions

Understanding by Design (UbD) is Grant Wiggins and Jay McTighe's framework for planning units and courses that build understanding and transfer, using backward design: a three-stage process that sets the desired results first, then the evidence of learning, then the learning plan.

Identify desired results, determine acceptable evidence, and plan learning experiences and instruction. The 2012 ASCD white paper shortens them to Desired Results, Evidence and Learning Plan.

Not quite. Backward design is the three-stage planning process at the centre of Understanding by Design. UbD is the wider framework, which adds transfer goals, essential questions, the six facets of understanding, GRASPS performance tasks and the WHERETO checklist.

UbD is a framework for planning units and courses backward from the results you want. UDL, or Universal Design for Learning, is a framework for making learning accessible to people with different needs. The two work together.

Yes. The expanded 2nd edition (2005) is still the current edition, the 2011 guide updated the planning template, and Jay McTighe still trains designers in the framework. A 2025 review found it improves alignment, while evidence on learner outcomes remains limited.

Carolina Martin, Learning Experience Manager at Coursebox AI

Carolina Martin

Customer Success Manager

Carolina is a highly experienced learning designer with tech start-up experience. She is based in Sydney and previously designed courses alongside Coursebox founder Travis in a previous job. She is passionate about delivering engaging learning experiences and brings the same perspective to customer experience too. She manages our customer support documentation and product testing processes.