1. The signs
- Requests arrive as "we need a course on X", and the course gets built.
- Success is measured by attendance and smile sheets.
- Training time is squeezed into lunch hours and evenings.
- Someone quotes an ROI of several hundred percent, and nobody asks how.
2. Why it happens
The solution arrives before the problem. "Training on X" is accepted as the brief, without asking what decision the sponsor faces or whether training addresses it. Many gaps are tools, time, incentives or procedures, which no course fixes. DGN-FRM-04 · IDE-DES-02
The wrong scoreboard. Course ratings are barely related to what people learned. LSC-MSR-01
Evaluation bolted on at the end, when the data it needs was never collected. IDE-DES-07
Training time treated as free, so it's the first thing cut when the schedule slips. LSC-ORG-02
3. What good looks like
- Every request starts with a measurable goal and the behaviours that should change, before anything is built. IDE-DES-01
- The gap is diagnosed first: knowledge, skill, motivation, process, tools or capacity, with evidence for each. DGN-FRM-02
- Measures from records you already keep: rework hours, RFIs, defects, time to first independent deliverable. LSC-ORG-03
- Application checked once it's routine, often three to six months later, with quality and side effects reported next to speed. IDE-EV-02 · IDE-EV-06
- A range and a break-even point instead of one ROI figure, and very high ROI claims treated as unverified until the method is shown. IDE-EV-07 · IDE-EV-12
- Training time budgeted as a managed overhead line, at loaded labour cost. LSC-ORG-02
"Sales wants a course on the new CRM." It's built in six weeks, rated 4.6 out of 5, and the pipeline data is as patchy as before.
"Opportunities are updated late because the mobile form is slow. We fixed the form, built a ten-minute job aid, and checked update times at 90 days."
4. Your first three moves
- On the next request, write the gap with no solution in it: the outcome, the current level, the expected level, and who's affected. DGN-FRM-01
- Pick one record you already keep as the measure, and agree it before design starts. LSC-ORG-03 · IDE-DES-07
- Change one survey question from "how much did you enjoy this?" to "how will you use this on your current job?" LSC-MSR-02
5. Stress-test it
A TPC Stress Test doubles as the evaluation L&D rarely gets to run: a check before the session, a check after it, one business measure agreed in week 1, and a re-check at 90 days. It gives your learning team the evidence leaders keep asking for, about one piece of work, in a few weeks.
The short version. Diagnose before you design, set a measurable goal, measure with records you already keep, check application once it's routine, report ranges rather than one ROI, and budget training time like the real cost it is.