Bloom's Taxonomy

The six levels, original and revised

Benjamin Bloom and a group of educators published the original taxonomy in 1956 as a way to classify educational goals in the cognitive domain. It had six categories: Knowledge, Comprehension, Application, Analysis, Synthesis and Evaluation.

In 2001, Lorin Anderson and David Krathwohl published a revision that most trainers now use. The changes: categories became verbs, Knowledge became Remember, Comprehension became Understand, Synthesis became Create, and Create moved to the top, above Evaluate. The revision also added a knowledge dimension (factual, conceptual, procedural and metacognitive knowledge) so you can describe both what kind of knowledge and what level of thinking an objective targets.

The revised levels, lowest to highest:

1. Remember: recall facts and terms. Verbs: list, name, define, identify.
2. Understand: explain ideas in your own words. Verbs: describe, summarize, classify, explain.
3. Apply: use a procedure in a given situation. Verbs: process, calculate, execute, demonstrate.
4. Analyze: break information apart and see relationships. Verbs: compare, troubleshoot, diagnose, differentiate.
5. Evaluate: make and defend a judgment against criteria. Verbs: assess, decide, recommend, justify.
6. Create: put elements together into something new. Verbs: design, build, draft, plan.

Bloom's group also described affective (attitudes and values) and psychomotor (physical skills) domains, which matter for safety, equipment and customer-facing training, but the cognitive domain is what people usually mean.

How trainers use it

The taxonomy is most useful at two points in instructional design: writing objectives and writing assessments.

For objectives, pick the level the job actually demands, then use a verb from that level. "Understand the refund policy" can't be measured. "Decide whether a refund request qualifies under the 30-day policy and explain the decision to the customer" sits at Evaluate and can be observed. Our entry on learning objectives covers the rest of the format, including conditions and standards.

For assessments, match the test to the objective's level. A multiple-choice quiz tests Remember and Understand well. Apply and Analyze need scenarios, system practice or a call simulation. Evaluate and Create need case work, projects or roleplays scored with a rubric. Mismatches here are the most common design flaw in corporate training.

Worked example: a billing support new-hire course

A utility's contact center is redesigning week two of new-hire training, which covers billing disputes. The old course objectives were "know the bill layout" and "understand dispute types," and the final test was 25 multiple-choice questions. New hires passed easily, then struggled on calls.

Rewritten with Bloom's levels:

• Remember: identify the eight sections of a residential bill.
• Understand: explain to a customer why an estimated read differs from an actual read.
• Apply: process a payment arrangement in the billing system for a customer with a past-due balance.
• Analyze: find the cause of a high-bill complaint using meter history and rate changes.
• Evaluate: decide whether a dispute qualifies for an adjustment and justify the decision against policy.

The assessment changes to match: a short quiz for the first two objectives, three system exercises for Apply and Analyze, and two scored roleplays for Evaluate. Time spent on lecture drops; time spent practicing roughly doubles. That shift, more than the verbs themselves, is what usually improves performance on the floor.

Bloom's taxonomy versus other models

Bloom's describes levels of thinking. It doesn't tell you how to run a design project; that's the job of process models like the ADDIE model or SAM. And it doesn't tell you whether training worked in business terms; the Kirkpatrick model handles evaluation. In practice, teams use Bloom's inside the Design phase of ADDIE and use Kirkpatrick at the end.

Common mistakes

• Treating it as a strict ladder. Learners don't have to master every lower level before touching a higher one. Working through a realistic case often builds understanding faster than memorizing first.
• Aiming everything at the top. Not every role needs Create-level objectives. A data entry role mostly needs accurate Apply; dressing it up as Create wastes time.
• Verb lists as decoration. Swapping "know" for "identify" without changing the activity or the test changes nothing.
• Ignoring the other domains. Empathy on calls and safe forklift handling are affective and psychomotor skills. They need practice and observation, not a quiz.