Instructional Design Basics

By Brian Tr · Updated Oct 7, 2026

Instructional design is the practice of deliberately planning learning experiences so that people end up able to do something they couldn't do before. It sits between subject expertise and software. An expert knows the material. A developer can build the product. The instructional designer decides what the learner should be able to do, in what order they should meet the material, what they should practice and how anyone will know it worked.

For a company that builds or sells learning products, this isn't an academic side topic. It's the difference between a course library people finish and one they abandon. This article is the hub for the Content & Delivery section of this collection. It covers the core ideas, the named models, the common process frameworks and what all of it means for product quality, completion and renewals.

Key Facts

• Robert Gagné's 1965 book The Conditions of Learning set out nine events of instruction, from gaining attention through to enhancing retention and transfer (University of Florida CITT, Gagne's 9 Events of Instruction).
• Anderson and Krathwohl's 2001 revision of Bloom's taxonomy turned the original six nouns into six verbs: remember, understand, apply, analyze, evaluate and create (University of Illinois Chicago, Bloom's taxonomy).
• Merrill's 2002 paper in Educational Technology Research and Development (volume 50) distilled five first principles: task-centered, activation, demonstration, application and integration (MIT Open Learning, First Principles of Instruction).
• Wiggins and McTighe introduced backward design in their 1998 book Understanding by Design, which plans from desired results to assessment evidence to learning activities (University of Illinois Chicago, Backward design).
• ADDIE traces to a 1975 Florida State University system for US military training. Its five phases were analyze, design, develop, implement and control (Branson et al., 1975, DTIC).
• Cognitive load theory grew out of John Sweller's 1988 study of problem solving (Sweller, 1988, Cognitive Science) and holds that working memory is limited in capacity and duration (Houichi and Sarnou, 2020, ERIC).

What instructional design is

The field's usual definition is the systematic design, development and delivery of instructional materials and experiences, aimed at acquiring knowledge efficiently and effectively. The word that matters there is systematic. Good design isn't taste. It starts from a goal, picks methods because they suit that goal, and checks the result.

The discipline has military roots. Robert Reiser's history of the field starts with the efforts to develop training programs during World War II and runs through the first instructional design models of the 1960s and 1970s (Reiser, 2001, Educational Technology Research and Development). That heritage explains why so many instructional design models look like engineering processes.

It helps to separate three things people confuse:

• Content is the information itself: a video, a document, a quiz bank.
• Instruction is content arranged so that it produces learning: objectives, sequence, practice, feedback.
• Delivery is the channel and format: a platform, a live cohort, a mobile lesson.

A great video in a poor sequence with no practice is content, not instruction. Many EdTech products are strong on delivery and content and weak on instruction, which is exactly the gap this discipline closes.

Start with the objective: Bloom's taxonomy

Nearly every design process begins with learning objectives, statements of what the learner will be able to do. The most common tool for writing them is Bloom's taxonomy.

The original 1956 framework divided the cognitive domain into six levels: knowledge, comprehension, application, analysis, synthesis and evaluation (Vanderbilt Center for Teaching, Bloom's Taxonomy). In 2001, a group of experts published A Taxonomy for Learning, Teaching, and Assessing: A Revision of Bloom's Taxonomy of Educational Objectives (Iowa State CELT). The revision moved the levels from nouns to verbs: remember, understand, apply, analyze, evaluate and create. It also added a knowledge dimension that separates what kind of knowledge is involved: factual, conceptual, procedural or metacognitive (University of Illinois Chicago).

Why does this matter commercially? Because the verb in an objective decides the activity and the assessment:

If a course promises "apply" or "create" outcomes but only tests recall, buyers eventually notice. That mismatch is one of the most common reasons training gets labeled as ineffective. For the buyer's side of this question, see training needs analysis, which is where objectives should originate.

Gagné's nine events of instruction

Robert Gagné's The Conditions of Learning (1965) identified the mental conditions for learning and set out nine events of instruction that correspond to them. The events, in order, are (University of Florida CITT):

1. Gain attention
2. Inform learners of objectives
3. Stimulate recall of prior learning
4. Present the content
5. Provide learning guidance
6. Elicit performance (practice)
7. Provide feedback
8. Assess performance
9. Enhance retention and transfer to the job

Treat it as a checklist for a single lesson, not a rigid script. Plenty of EdTech lessons cover events 1, 2 and 4 (a hook, a learning-goal slide, a presentation) and skip 6 through 9. The skipped half is where learning is consolidated and where the customer eventually sees performance change.

Merrill's First Principles of Instruction

13. David Merrill reviewed many instructional design theories and in 2002 published the common ground he found in Educational Technology Research and Development (volume 50, pages 43 to 59). The five principles say learning is promoted when (MIT Open Learning):

1. Learners engage in a task-centered strategy, working on real-world problems.
2. Learners activate prior knowledge or experience.
3. Learners observe a demonstration of the new knowledge.
4. Learners apply the new knowledge.
5. Learners integrate the new knowledge into their everyday world.

The task-centered principle is the anchor. The other four describe what happens around a real problem. The principles also work as an evaluation framework: you can hold a finished product up against them and ask which of the five it fails. That makes them handy during due diligence, product review or competitive teardown.

How people process what you show them: Sweller and Mayer

The models above say what to include. Two bodies of research say how to present it so a human brain can keep up.

Cognitive load theory. John Sweller developed it in the late 1980s from studies of problem solving. The 1988 paper found that conventional problem solving takes so much processing capacity that little is left for building the knowledge structures learning depends on (Sweller, 1988). The theory starts from the finding that working memory is limited in capacity and duration, and it distinguishes (Houichi and Sarnou, 2020):

• Intrinsic load, the effort tied to the topic itself.
• Extraneous load, the effort created by the way information or tasks are presented.
• Germane load, the work that goes into building lasting knowledge.

The design lever is extraneous load. You can't make a hard topic easy, but you can stop the interface, the layout and the narration from making it harder.

Mayer's multimedia principles. Richard Mayer built a research program on how people learn from words and pictures. Among his published principles (University of Pittsburgh Teaching Center, Mayer's multimedia principles):

• Coherence: people learn better when extraneous words, pictures and sounds are excluded rather than included.
• Signaling: people learn better when cues that highlight the organization of the essential material are added.
• Redundancy: people learn better from graphics and narration than from graphics, narration and on-screen text.
• Segmenting: people learn better from a lesson presented in user-paced segments than as a continuous unit.
• Modality: people learn better from graphics and narration than from animation and on-screen text.

Each of these has direct product consequences. Decorative stock animation breaks coherence. A video that reads its slide text aloud while the slide shows the same text breaks redundancy. A 45-minute unbroken lecture breaks segmenting. The last point connects to why microlearning became popular, although short isn't automatically better. Segmenting means user-paced chunks of a coherent lesson, not a pile of disconnected clips.

One caution on evidence. Sweller's and Mayer's work comes from controlled studies, many in laboratory-style settings with short lessons. Treat the principles as strong defaults, then test them with your own learners and your own content rather than assuming a lab result transfers perfectly to a 12-week corporate program.

Backward design

Grant Wiggins and Jay McTighe introduced backward design in their 1998 book Understanding by Design. It reverses the habit of starting with textbooks and activities. The three stages are (University of Illinois Chicago):

1. Identify the desired results.
2. Determine acceptable evidence that learners achieved them.
3. Plan the learning experiences and instruction.

The practical effect is that the assessment is designed before the lessons. If you can't describe what a passing performance looks like, you aren't ready to build content. This is the single most useful habit for a product team that tends to build the lecture first and bolt a quiz on at the end. It also pairs naturally with proving training ROI, because the evidence you plan for in stage two is the evidence a buyer later asks for.

Process frameworks: ADDIE, SAM and Dick and Carey

The models above are principles. Process frameworks are the project plans that carry them out.

ADDIE traces to a 1975 Florida State University system for US military training, which had five phases: analyze, design, develop, implement and control (DTIC). Modern versions are usually drawn as a cycle, with room to revise earlier phases. Despite the jokes about it being old, it remains the reference point most teams define themselves against. A dedicated article, the ADDIE model, goes through each phase.

SAM was created by Michael Allen of Allen Interactions as an agile approach to e-learning. Its three phases are preparation, iterative design (opening with a short "Savvy Start" session where success measures are set) and iterative development, with design proof, alpha and beta versions. Allen Interactions describes it as "repeated small steps" in contrast to ADDIE's "perfectly executed giant steps" (Allen Interactions, SAM). That description comes from the method's own vendor, so read it as the case for SAM, not a neutral comparison.

Dick and Carey is a systems approach that treats instruction as a set of interconnected components (the instructor, learners, materials, activities, delivery system and learning environment), worked through in nine steps toward a defined goal (University of Florida IFAS Extension).

None of the three is universally right. Choose by risk. If requirements are stable and stakes are high, front-loaded analysis pays off. If you don't know yet what learners will respond to, prototype early and iterate. Many product teams run a hybrid: ADDIE-style analysis up front, SAM-style cycles for build.

The instructional designer role

An instructional designer turns a business or learning goal into a learning experience. In a typical EdTech or corporate-learning team, the job covers:

• Working with subject matter experts to extract what matters and cut what doesn't.
• Writing objectives and mapping them to practice and assessment.
• Choosing a format: self-paced module, live session, simulation, job aid or a blended mix.
• Storyboarding or scripting, and reviewing media against principles like coherence and segmenting.
• Defining how success will be measured, and reading the results afterwards.

The US Bureau of Labor Statistics tracks a related occupation, instructional coordinators, who "develop educational material, implement it with teachers and principals, and assess its effectiveness". It reports a median pay of $77,440 for May 2025 (BLS Occupational Outlook Handbook). That category is mostly school and curriculum roles, and a master's degree is its typical entry requirement. It isn't a measure of corporate instructional designer pay, which varies by industry and country, so use it only as a loose reference for the type of work.

Titles overlap heavily: learning experience designer, curriculum developer, learning architect, e-learning developer. What matters is whether someone on the team owns the job of connecting objectives to activities to evidence. In small EdTech startups, a founder or the subject expert often does this by default, and the gaps show up in the product.

What good design means for an EdTech company

Instructional design shows up in a company's numbers, even if it never appears on a dashboard under that name.

Product quality and trust. A buyer's L&D team judges a course by whether it can change behavior. A product built to objectives, with practice and feedback, is easier to defend internally. A product built as a content library isn't. How buyers weigh these things is covered in how companies buy training.

Completion. People drop out when lessons feel irrelevant, overloaded or pointless. Design addresses all three: relevance through task-centered problems, overload through reducing extraneous load and segmenting, and pointlessness through clear objectives and visible progress. Published completion figures vary widely by format and audience and are easy to misread, so measure your own, by cohort and by lesson, and look for the lesson where learners leave. The format choice matters too, as discussed in cohort-based vs self-paced learning.

Results, which feed renewals. Subscriptions and contracts renew when the customer can point to something that changed. If the design never defined what "changed" means, no one can show it. Backward design fixes that by forcing the evidence question early. The commercial effects on retention are explored in corporate training business models.

Cost and speed. A heavyweight process is expensive for every lesson. A light one risks shipping weak material. Matching the process to the risk, ADDIE for the high-stakes core and SAM-style iteration for fast experiments, controls cost without gutting quality.

A simple internal audit helps. For each flagship course, ask: Is there a written objective with a measurable verb? Does the assessment test that verb? Is there practice before the assessment? Does feedback arrive while it's still useful? Is anything on screen there only for decoration? Five honest answers will usually tell you where to invest design time first.

Related reading

• The ADDIE Model
• Blended Learning
• Microlearning
• What Is a Training Needs Analysis?