Instructional design is the systematic planning of learning materials and experiences so that a defined group of people can do something specific afterwards that they could not do before. It covers everything from a one-hour product tutorial to a year-long corporate academy. The discipline exists because content alone rarely produces skill: an expert who knows a subject does not automatically know how to sequence it, test it or make it stick. Instructional designers sit between that expert and the learner, turning knowledge into objectives, activities and assessments, then checking whether the result worked.

For the marketing industry the term matters more than its academic origins suggest. Platform certifications, agency onboarding, AI literacy programmes and the steady stream of free courses from technology companies are all products of instructional design, good or bad.

How the process works

Most practice still runs on ADDIE, an acronym for five phases: analysis, design, development, implementation and evaluation.

  1. Analysis. The designer establishes the performance gap, who the learners are, what they already know and what constraints apply, such as budget, time or delivery platform. A frequent finding at this stage is that training is not the answer at all, because the problem is a tool, a process or an incentive.
  2. Design. Objectives are written, assessments are planned to match them, and content and media are chosen. Objectives are usually phrased with verbs from Bloom's taxonomy, so "understand bidding" becomes "explain how target CPA (cost per acquisition) bidding reacts to a change in conversion volume".
  3. Development. Materials are built, often in authoring tools that package lessons for a learning management system (LMS), then piloted with a small group.
  4. Implementation. Facilitators are prepared, learners enrol and the course goes live.
  5. Evaluation. Formative evaluation happens throughout; summative evaluation judges the finished product.

Within a single lesson, many designers follow Gagne's nine events of instruction: gain attention, inform learners of the objective, stimulate recall of prior learning, present the content, provide learning guidance, elicit performance, provide feedback, assess performance, and enhance retention and transfer. A module teaching agency staff a new campaign type might open with a client scenario, state that learners will be able to configure the campaign by the end, recall how the older format worked, demonstrate the setup, then ask learners to build one in a sandbox account and correct their errors.

Technical standards carry the result into software. SCORM (Sharable Content Object Reference Model) lets a course built in one tool run inside any compliant LMS and report completion and scores. Its successor, xAPI (Experience API), records learning events as simple statements, such as "learner completed simulation", including activity outside an LMS. Evaluation frequently uses Kirkpatrick's four levels: reaction, learning, behaviour and results.

The work is shared between roles. Subject matter experts supply the knowledge, instructional designers shape it, developers build it and learning and development (L&D) teams run it. In schools the nearest equivalent is the instructional coordinator; the US Bureau of Labor Statistics counted about 248,700 such jobs in 2025, with median pay of $77,440 in May 2025, and projects 2% growth from 2025 to 2035. In corporate settings, the Association for Talent Development (ATD) reported in its May 2025 State of the Industry report that instructional designers were among the most commonly added talent development roles in 2024, even as formal learning per employee fell to 13.7 hours from 17.4 hours in 2023.

Origin and evolution

The field grew out of behavioural psychology and military training. B.F. Skinner argued for programmed instruction, with small steps and immediate reinforcement, in his 1954 paper "The Science of Learning and the Art of Teaching". In 1956 a committee chaired by Benjamin Bloom published the Taxonomy of Educational Objectives, with six cognitive levels: knowledge, comprehension, application, analysis, synthesis and evaluation. Donald Kirkpatrick's evaluation levels, drawn from his 1954 doctoral dissertation, reached practitioners through a series of articles in 1959. Robert Gagne set out the nine events in The Conditions of Learning in 1965.

The systems approach arrived in 1975, when Florida State University's Center for Educational Technology, working with the US Army, produced the Interservice Procedures for Instructional Systems Development for all four armed services. Its phases were labelled analyse, design, develop, implement and control. According to Michael Molenda's 2003 review, the ADDIE label itself has no clear origin, and the iterative version most practitioners know had taken shape by the mid-1980s. Walter Dick and Lou Carey published their competing systems model in 1978.

Computers changed delivery. The US Department of Defense's Advanced Distributed Learning initiative released SCORM 1.0 in January 2000, version 1.2 in October 2001 and SCORM 2004, whose fourth edition appeared on March 31, 2009. Anderson and Krathwohl revised Bloom's taxonomy in 2001, replacing nouns with verbs and placing "create" at the top: remember, understand, apply, analyse, evaluate, create. Richard Mayer's multimedia learning principles and M. David Merrill's First Principles of Instruction, published in 2002, gave designers research-based rules for combining text, images and practice.

Speed became the next complaint. Michael Allen's Leaving ADDIE for SAM, published by ATD in September 2012, proposed the Successive Approximation Model, an iterative method built on rapid prototypes. xAPI 1.0.0 followed in April 2013, and in 2023 it became an IEEE standard, IEEE 9274.1.1-2023, according to Advanced Distributed Learning.

Why it matters to marketers

Much of what the advertising industry knows about its own tools is learned through designed courses. Google's Skillshop exams typically carry about 50 questions in 75 minutes and require 80% to pass, and Meta charges $99 for its Digital Marketing Associate exam, as set out in PPC Land's explainer on upskilling. TikTok revamped its Academy in September 2024 into five learning paths with courses of 4 to 15 minutes, a textbook microlearning structure. LinkedIn's Ads Agency Certification, announced on May 6, 2026, ties agency status to staff completing LinkedIn Marketing Academy certifications.

Regulation has added weight. Article 4 of the EU AI Act, which has applied since February 2, 2025, requires organisations deploying AI to ensure sufficient AI literacy among staff. The Dutch Data Protection Authority's implementation guide published on October 23, 2025 reads like an ADDIE outline: identify systems and knowledge gaps, set role-specific goals, implement training, evaluate and repeat.

The skills gap is measurable. In MiQ's survey of 3,169 marketers, fielded by Censuswide in September 2025, 72% planned more AI use but only 45% felt ready, and 38% cited a lack of training on AI tools. Technology companies have answered with free courses, from Microsoft's AI learning paths to Anthropic's nine-chapter prompt engineering course, each of which doubles as product marketing.

Limitations and disputes

ADDIE is criticised as a waterfall process: by the time evaluation reveals a flaw, the budget is spent. SAM and other agile approaches were built on that complaint, although defenders argue that ADDIE was never meant to be strictly linear. Bloom's taxonomy faces questions about whether its levels form a genuine hierarchy, and a 2020 study found that institutions map action verbs to levels inconsistently.

Some popular ideas have weak evidence. A 2008 review by Harold Pashler and colleagues in Psychological Science in the Public Interest found no adequate evidence base for tailoring instruction to individual "learning styles", a concept still common in corporate training. Evaluation is another weak point: completion rates and satisfaction scores, Kirkpatrick's first two levels, are easy to collect, while behaviour change and business results are hard to attribute.

Generative AI has sharpened the dispute. Tools now draft objectives, scripts, quizzes and narrated video in minutes, which lowers costs but raises the risk of plausible errors, the problem PPC Land covers under hallucination. Critics argue that automated content generation is the easy part of the job, and analysis and evaluation the hard part.

Not the same as

  • Learning experience design (LXD) borrows from user experience design and emphasises the learner's journey, emotion and context. Many practitioners treat it as a rebranding or extension of instructional design rather than a separate discipline.
  • Sales enablement equips sales teams with content, tools and training to close deals. It uses instructional design for its training component, but its scope includes playbooks, collateral and sales technology, and its success metric is revenue rather than learning.
  • Curriculum design usually refers to whole programmes of study in schools and universities, while instructional design often operates at course or module level.
  • Learning phase in advertising platforms describes the calibration period of automated bidding after a campaign change. It concerns machine learning, not human learning.

Recent developments

AI products are increasingly sold on learning design principles. Google introduced Guided Learning in Gemini, a mode that walks students through questions and steps instead of giving answers, alongside free AI Pro access for college students in August 2025. NotebookLM's Video Overviews, launched on July 29, 2025, let users set learning objectives and audience expertise before generating narrated slides. On July 16, 2026, Google renamed NotebookLM to Gemini Notebook, reporting more than 30 million users.

Usage data points the same way. Anthropic's Economic Index released on March 24, 2026 found that experienced Claude users were more likely to use a learning mode, while coursework fell from 19% to 12% of Claude.ai traffic. The pressure on traditional course providers was visible earlier: Similarweb data showed Udemy and Coursera visits falling by 12.61% and 9.98% respectively, as published in July 2025. Whether chatbots replace designed instruction or simply deliver it through a new interface remains, as of October 2026, an open question.

Timeline

  • 1954 - B.F. Skinner publishes "The Science of Learning and the Art of Teaching"; Donald Kirkpatrick completes his dissertation on training evaluation
  • 1956 - Bloom's Taxonomy of Educational Objectives, Handbook I, is published
  • 1959 - Kirkpatrick's four levels reach practitioners through a series of journal articles
  • 1965 - Robert Gagne publishes The Conditions of Learning, setting out the nine events
  • 1975 - Florida State University and the US Army produce the Interservice Procedures for Instructional Systems Development
  • 1978 - Dick and Carey publish their systems approach model
  • Mid-1980s - The iterative ADDIE form becomes established
  • January 2000 - SCORM 1.0 released by Advanced Distributed Learning
  • 2001 - Anderson and Krathwohl publish the revised Bloom's taxonomy; SCORM 1.2 follows in October
  • 2002 - Merrill publishes First Principles of Instruction
  • 2008 - Pashler and colleagues find no adequate evidence for learning-styles instruction
  • March 31, 2009 - SCORM 2004 fourth edition released
  • September 2012 - Michael Allen publishes Leaving ADDIE for SAM
  • April 2013 - xAPI 1.0.0 released
  • 2023 - xAPI becomes IEEE standard 9274.1.1-2023
  • February 2, 2025 - EU AI Act Article 4 AI literacy duty applies
  • July 29, 2025 - NotebookLM Video Overviews launch with learning-objective controls
  • October 23, 2025 - Dutch Data Protection Authority publishes its AI literacy guide
  • March 24, 2026 - Anthropic Economic Index reports learning-mode use rises with experience
  • May 6, 2026 - LinkedIn announces Ads Agency Certification
  • July 16, 2026 - NotebookLM renamed Gemini Notebook

Summary

Who: Instructional designers, subject matter experts and learning and development teams in companies, universities, schools and the military, with platforms such as Google, Meta, TikTok and LinkedIn running their own academies.

What: A systematic discipline for designing learning materials and experiences, using models such as ADDIE, Bloom's taxonomy, Gagne's nine events and Kirkpatrick's evaluation levels, delivered through standards such as SCORM and xAPI.

When: Rooted in the 1950s and 1960s, formalised as a systems model in 1975, standardised for e-learning from 2000, and reshaped by generative AI since 2023.

Where: In classrooms, corporate LMS platforms, advertising platform academies and, increasingly, inside AI assistants offering guided learning modes.

Why: Because expertise does not transfer by itself. Structured objectives, practice, feedback and evaluation are what turn content into capability, a need sharpened by AI literacy rules and the industry's AI skills gap.