Ask for the thinking, not just the artefact
A finished course can look excellent and teach poorly. The most reliable way to judge instructional design is to ask for the reasoning: what behaviour is this module trying to change, what did the performance analysis show, what learning objectives were set, and how does each screen or section map to an objective. A designer who can answer these questions is designing; one who cannot is decorating.
Ask to see a sample lesson built to your subject area, and review it against the objectives rather than against the visuals. Then watch the designer work through the sample with you, explaining what they would change for your audience. The walkthrough reveals far more than the sample alone.
A productive framing is to describe a real learner: their experience level, the time they have, the environment they are in, and what they already believe about the topic. Ask what the design assumes about that learner, and where the design deliberately reduces load.
Check the alignment chain
Strong learning design maintains a visible chain from business need to behaviour to assessment. Ask the supplier to walk one objective through the whole chain: the reason the behaviour matters, the practice the learner performs, the feedback they receive, and the evidence that the behaviour has transferred.
Where the chain breaks, you will usually find either an activity that is fun but purposeless, or an assessment that checks recall without checking judgement. Both are common and both are fixable, but only if you can see them, which is why the chain is such a useful diagnostic.
Ask how the design handles the difference between knowledge, skill, and judgement objectives. A module that requires a learner to make a judgement under uncertainty needs scenario practice with realistic distractors and consequences.
A module that requires only recall can use simpler interaction. Suppliers who treat these identically are pattern-matching rather than designing.
Look for realistic practice and useful failure
The quality of practice is the clearest separator between good and mediocre custom work. Look for activities where the learner must retrieve information, make a decision, and see a consequence. Ask what happens when the learner chooses the wrong option, and whether the feedback explains the reasoning rather than simply marking the answer wrong.
Ask about cognitive load management: how the design chunks content, how it uses dual channels, how it reduces extraneous material, and whether the pacing suits a busy professional. Screens that are dense with text, or narration that repeats the on-screen text word for word, are reliable indicators of design that has not been reviewed critically.
Ask how the design supports application after the course, such as job aids, checklists, or prompts in the workflow. A module that ends at the screen is a weaker intervention than one that plans for the transfer problem. See our guide on evaluating AI-assisted content production for how these same design questions apply to machine-assisted drafting.
Assess whether the design is fit for the delivery constraints
Good design serves the operating environment. Ask how the design changes for mobile, for a short window of time during a shift, for a cohort with limited bandwidth, or for a screen that is not a laptop. If a course was designed for a desktop and then shrunk, the interaction patterns often break.
Also consider the maintenance horizon. A design that depends on a single complex branching tree is harder to keep accurate as policy changes than a modular design with clearly separated components. Ask the designer to explain what they would do differently if you needed to update one policy statement in six months without rebuilding the module.
Where the audience includes accessibility needs, ask how the design accounts for them from the start, including alternative routes for learners who cannot complete a timed or interactive element.
Use the proposal conversation as a diagnostic
You can learn a great deal from how a supplier handles your own questions during pitching. A designer who becomes visibly interested in the actual learning problem, pushes back constructively on an unrealistic request, and is candid about a weakness in their proposed approach is giving you a preview of how they will behave when the project is difficult.
Be careful with designers who answer quickly and generally. Fluency in design vocabulary is easy; it is less common than it appears, and it is not the same as competence. The distinction shows up when you ask a specific, applied question, such as how they would introduce a new policy to a sceptical audience, or what they would cut first to fit a shorter time budget.
Finally, ask for one thing they would do differently if they were starting this project again. The answer reveals the limits of their self-awareness, which is a meaningful indicator of professional maturity. See our guide on how to select a custom eLearning development company for how this assessment fits into a wider selection process.
Commonly overlooked pitfalls
- Judging design quality from a polished portfolio piece with no access to the objectives or the performance problem behind it.
- Accepting novelty and polish as evidence of learning effectiveness, when the two are only loosely connected.
- Skipping the alignment check, so purposeless activity and shallow assessment pass unnoticed.
- Failing to test the design against real delivery constraints such as mobile access and short study windows.
- Overlooking the maintenance implication of a design that cannot be updated without a rebuild.
Next steps
- Ask each finalist to walk one objective through the full chain from business need to assessment evidence.
- Request a sample lesson and critique it against your own learning objectives rather than against visual polish.
- Include one scenario-based or judgement-based objective in your requirement, so design quality is visible in the response.
- Ask every candidate to state their maintenance and update strategy for policy change after launch.
- Score design responses against written criteria with at least two independent reviewers per proposal.
Related guides
Put this into practice
The vendor evaluation checklist and the RFP template turn a guide into a process you can run this week.