“Project Management for Training and L&D: A Practical Guide”

What does project management for training and L&D actually deliver?

It delivers a measurable change in workplace performance, on a date, within a budget, with evidence that the change was caused by the intervention rather than by seasonality or a concurrent process change. The deliverable is not a course. The course is one possible output of an analysis that may just as easily conclude that the real fix is a job aid, a checklist, a system change, or a supervisory conversation.

That distinction is the single most important framing decision in the discipline. A training team that accepts every request as a course request becomes a content factory with no way to demonstrate value. A team that runs a short analysis first can say, with evidence, that the problem is a missing approval step rather than a knowledge gap, and save six weeks of production.

The PM owns the lifecycle and the evidence chain, the instructional designer owns the learning design, and the subject matter expert owns the technical truth. When those roles blur, review cycles expand and version control collapses.

How do you run a needs analysis that stands up?

You run it by separating four questions and answering them in order: is this a performance gap, is it caused by a lack of knowledge or skill, is training the most efficient remedy, and what will success look like numerically?

The four-question gate

Question 1: What is the current performance, measured? Not “people are not following the process” but “only 62 percent of onboarding packs were completed within five days in Q3, against a target of 95 percent.” A number creates a baseline that evaluation can later compare against.

Question 2: Is the cause a gap in knowledge, skill, environment, or motivation? Interview six performers and two non-performers, review the last twenty error tickets, and look at the job aid people actually use (usually a photographed sticky note). If the system makes the correct action slower than the incorrect one, training will not fix it.

Question 3: What is the cheapest intervention that closes the gap? Compare a job aid, a process change, a short refresher, and a full course on cost per performer and time to competence.

Question 4: What result will the sponsor accept as proof? Get this in writing before design starts. If the answer is “fewer escalated tickets,” you now know what data to instrument in week one rather than week twelve.

Worked example: the cost of a false training request

A support organisation of 120 agents requests a two-day course on handling complex refunds. A four-day analysis finds that 71 percent of refund errors involve one policy exception that the knowledge base documents on page nine of a 34-page article.

The course was not wrong, it was premature. The decision aid shipped in two weeks and the error rate on that exception fell by 44 percent in the following quarter, which is measured evidence the team could take to the sponsor.

What are the phases of a training project?

A training project runs through five phases with explicit gates. The phases map to ADDIE, but the gates matter more than the labels, because a gate is where a sponsor can stop the project cheaply.

Two practical notes: design and development overlap in reality, and evaluation is the only phase whose duration depends on the business cycle rather than the team, which is why it must be scoped at the start.

How do you write objectives that survive review?

You write them as observable performance with conditions and criteria. “Understand the new expense policy” cannot be assessed. “Submit an expense claim with the correct cost centre and receipt attached, without escalation, in under five minutes” can be.

A workable objective has three parts:

• Performance: the observable action, using a verb you can watch or count
• Condition: the tools, references, and setting available at the time
• Criterion: the standard that counts as acceptable, expressed as accuracy, time, or frequency

Objectives then drive assessment, and assessment drives content. When a stakeholder demands an extra module late in development, the PM’s question is not “can we fit it” but “which objective does this serve, and which assessment will prove it.” If there is no answer, the module is scope creep with a friendly face.

How do you estimate build effort without guessing?

You estimate from a production ratio: development hours per finished hour of learning, calibrated against your own past projects. Industry benchmarks are a starting point, not a truth. Publicly cited figures for basic e-learning commonly sit around 40 to 80 development hours per finished hour, while branching scenarios, simulations, and high-fidelity video can exceed several hundred.

Worked example: building a six-module programme

A compliance programme needs six e-learning modules of 20 minutes each, total finished runtime two hours, plus a 90-minute virtual workshop.

At 30 productive hours per week for the assigned developer, that is 6.6 weeks of one person’s capacity, before any stakeholder delay. Add a 25 percent contingency for review lag and that becomes 8.3 weeks. If the sponsor wants delivery in six weeks, the honest answer is to cut two modules or add a second developer, not to promise overtime.

Teams that keep a small table of actual ratios by modality stop having this argument. After three projects you know whether you build at 45 or 90 hours per finished hour, and estimates become calculations.

How do you manage delivery and facilitation?

Delivery is a logistics project with a fixed date and a perishable audience. The constraints are physical and human: room capacity, shift patterns, facilitator availability, time-zone spread, and the LMS enrolment cut-off.

Three controls prevent most delivery failures:

A delivery readiness checklist reviewed ten business days out. Content approved, LMS tested in the production environment, facilitator briefed, calendar invites accepted, backfill confirmed, captions and transcript uploaded, accessibility check passed.

A no-show forecast based on history. If your voluntary sessions historically run at 68 percent attendance, plan for 68 percent and overbook accordingly, or you will report a failure that was a forecasting error.

A pilot with real learners. Ten participants from the target population, using the production environment, two weeks before launch. Pilots catch the broken link, the exercise that takes twice as long as designed, and the assessment question that everyone misreads.

Accessibility deserves its own line in the checklist. Captions, keyboard navigation, colour contrast, and screen-reader-compatible documents are legal requirements in many jurisdictions and are far cheaper to fix before build than after. The Web Content Accessibility Guidelines version 2.2, published in 2023, are the common reference point.

How do you design evaluation with the Kirkpatrick model?

You design it backward. Start at Level 4 with the business result the sponsor named during analysis, then ask what on-the-job behaviour would produce it at Level 3, then what knowledge or skill that behaviour requires at Level 2, then what reaction would sustain engagement at Level 1.

Donald Kirkpatrick introduced the four levels in the 1950s; the New World version, developed by Jim Kirkpatrick and Wendy Kayser Kirkpatrick, added the required drivers and a stronger emphasis on Level 3 and on planning from results backward.

The uncomfortable statistic in this field is that most organisations stop at Levels 1 and 2. ATD’s 2016 study of 199 organisations found that only 35 percent evaluated business results. That gap is a project management failure, not a measurement failure: nobody scoped the data collection, nobody owned the 90-day follow-up, and nobody agreed in advance what result would count.

Worked example: from behaviour to money

A field service team of 90 technicians completes 12,000 repair jobs a year. A training programme targets a specific diagnostic error that currently occurs on 7.2 percent of jobs. The cost of a repeat visit is $85 in parts, travel, and labour.

• Baseline: 12,000 × 7.2% = 864 error jobs per year
• Baseline cost: 864 × $85 = $73,440
• Target after training: 5.8% error rate, a reduction of 1.4 percentage points
• Expected benefit: 12,000 × 0.014 × $85 = $14,280 per year

Now compare to cost. Design, delivery, backfill, and evaluation total $31,000 in year one. On a one-year view the programme destroys value. On a three-year view, assuming the effect persists, the benefit is $42,840 against a $31,000 cost, a net gain of $11,840, or a benefit-cost ratio of 1.38. Add the second and third error categories that the same diagnostic method addresses and the case improves quickly.

This is why evaluation design belongs in week one. The PM needs to know whether the programme is a one-year payback or a three-year payback, because that changes the sponsor conversation entirely.

Reaction data is cheap and often misused. A cohort rating relevance 4.6 and confidence 4.1 looks excellent, but if the same cohort then scores 58 percent on the post-assessment, the session was enjoyable, not effective. Report both together, always.

What are the trade-offs in training project management?

Speed versus transfer design. A one-hour webinar can be designed in a week and forgotten in two. Spaced practice, manager reinforcement, and follow-up assignments multiply the effect and also multiply the calendar. The PM’s job is to price the difference and let the sponsor choose.

Custom versus bought content. Off-the-shelf libraries cost little per seat and rarely match your process; custom builds match exactly and carry the full production burden. Most teams buy generic topics such as security awareness and build anything touching a proprietary system.

Synchronous versus asynchronous. Live sessions give immediate feedback and cost facilitator time and scheduling friction; asynchronous modules scale and depend on self-discipline. Blending works when the module carries knowledge and the live session carries practice, and fails when the live session repeats the module.

Measuring everything versus measuring what matters. Instrumenting every level for every programme is unaffordable. Choose one or two programmes per year for full Level 3 and 4 evaluation and run the rest on Level 1 and 2 plus a business owner’s judgement.

What mistakes do L&D project managers make most often?

• Accepting the course request as the requirement. The analysis gate exists to test whether training is the right remedy.
• Estimating from optimism. Without measured production ratios, every estimate is a guess with a deadline attached.
• Skipping the design sign-off. Late SME changes cost three to five times more than changes caught at storyboard.
• Treating the LMS as the project. Enrolment and reporting are one workstream, not the delivery.
• Evaluating only reaction. Smile sheets cannot defend a budget.
• Ignoring the manager’s role. Level 3 behaviour depends on whether the learner’s supervisor notices and reinforces it, which is a stakeholder deliverable, not a training deliverable.
• No version control. Two SME edits applied in different files produce a course that contradicts itself.

Where does Doitify fit in an L&D project?

Doitify is a project management workspace, and the structure that helps most in learning projects is the stage-gate board combined with a review log. Analysis, design sign-off, beta review, pilot, and launch each become a gate with an owner and an attached artefact, so the recurring dispute about whether a module is “done” resolves against a document rather than an opinion. The review log is the second useful piece: each SME comment gets a status, a decision, and a date, which is what prevents a third review cycle from becoming a fourth.

Transparency note: this is an educational article. Doitify is one option among many; a disciplined spreadsheet with the same gates and the same review log will do the job for a small team.

Conclusion

Project management for training and L&D is the management of an evidence chain. Analysis defines the performance gap and the metric. Design converts it into objectives and assessments. Development converts those into material at a measured production ratio. Implementation puts the right people in the right room or the right module. Evaluation closes the loop with data the sponsor agreed to in advance.

Teams that do this well are not more creative than the ones that struggle. They refuse to build before the problem is measured, and refuse to call a project finished before the behaviour is observed on the job.

If this post on project management for training and L&D was helpful, you might also enjoy Project Management Software for Professional Services and Best Project Management Software for Multiple Projects in 2026.