And some thoughts about the question about how to decide what kind of content (facts, concepts, processes, procedures, and principles) should go into what medium of delivery….

Start your analysis by getting clear on the performance — not the subject matter — that you expect of learners. In other words, what you expect them to DO (procedures and other types of behavior) and what you expect them to PRODUCE (work outputs). In fact, if you define work outputs first, then the lists of behavior gets shorter and more focused — to avoid “paralysis analysis” — because you can get very specific about what behavior is needed to produce those outputs. And outputs, with performance criteria defined, also tell a learner when they’re “there” — when they’re performing to standard and contributing to business results. We ultimately care about behavior only when it helps us achieve organizational results.

Thomas Gilbert, in his groundbreaking book, “Human Competence,” was the first to clearly articulate the value of defining work outputs during performance/job/task analysis, although he called them “accomplishments.” That concept is core to ISPI’s Human Performance Technology (HPT) and ASTD’s Human Performance Improvement (HPI) methodologies.

After you’ve defined the performance (outputs, behavior), then define the subject needed to support the performance. That way you end up with very lean-but-effective training, and you’ll find out if you’re missing anything when you pilot it. On the other hand, if you start with subject matter, you’ll end up with overly-fat training — a burden to instructional designers and learners and a cost to the enterprise. And to make it worse, when time for training is tight, what gets cut is often practice and feedback, those factors most critical to building performance.

I find it very effective to build an analysis spreadsheet (“performance support plan”) with outputs and tasks/behavior down the first column, then populate other columns with data about those outputs and tasks (speed requirements, complexity, environment, etc.) — which provide the data to make decisions about job aids vs. training-to-recall, delivery media, etc. Those decisions get captured in additional columns. It helps me organize my thinking and make good decisions and it’s great to walk through with the client.

You don’t always have the need or the time to capture all of this data, but it’s critical to at least capture outputs, criteria and tasks. Without some level of performance analysis it’s almost impossible to develop performance-based training — training that actually contributes to performance.

I originally learned this sort of analysis through Joe Harless’s Front-End Analysis and later ASTD’s Performance DNA toolkit (taught in ASTD’s “Analyzing Human Performance” course – part of the HPI Certificate). Lately I’ve become a raving fan of faster/simpler toolkits like the Six Boxes Approach (www.sixboxes.com).