L&D is almost always your entry point. It's rarely your actual audience. The gap between the two causes more design misses than any framework failure.
Nearly every engagement starts the same way: an L&D contact reaches out, briefs you, sits on the diagnostic call, and becomes the primary relationship for the life of the program. That's completely normal, and most of the time it works fine. But it creates a structural risk that's easy to miss because the relationship feels complete even when the diagnosis isn't: your main source of truth about a business problem is someone who doesn't own that business problem.
A good L&D partner knows the organization's history with past programs, the political terrain, the budget reality, and often the competency language the business already uses. That's genuinely useful context. What they frequently don't have is direct, current, first-hand exposure to the specific decision or behavior that's actually costing the business money — because that lives with a sales leader, a plant manager, a regional head, someone L&D is representing secondhand.
The failure mode isn't that L&D is wrong. It's that a secondhand read gets treated as a firsthand diagnosis, because it arrived through the person who's supposed to know. The brief sounds authoritative. It's frequently a good-faith summary of someone else's problem, one level removed from the person actually living it.
You can't simply ask to skip L&D and go straight to the business leader — that's usually neither realistic nor fair to the relationship that got you in the room in the first place. L&D's own incentives matter here too: they're often measured on smooth vendor management and program delivery, not on the same business KPI the sponsor cares about, which means the version of "success" they're optimizing a design toward can be subtly different from the version the business actually needs. None of that is a character flaw. It's just what happens when the person diagnosing the need and the person accountable for the outcome aren't the same person.
The practitioners who consistently avoid this trap don't treat L&D as an obstacle to route around — they treat the L&D contact as the person who can get them access, and then explicitly ask for it. A direct, low-pressure request works better than an implicit one: "would it be possible to spend fifteen minutes with one or two people from the actual audience, just to validate a couple of things before we finalize the design?" Framed as validation rather than second-guessing, this is rarely refused, and it's the single highest-leverage question available once a brief has already been handed over.
When that access genuinely isn't possible — global mandates, sensitive politics, a sponsor who's simply unavailable — the honest move is to say the diagnosis is provisional, not final. Build in an explicit checkpoint after the first module or first cohort where the design gets tested against real reaction from the actual audience, rather than presenting a first draft as if it were already validated truth.
L&D is your access point, not automatically your source of truth. The two are often the same thing. When they're not, the cost of assuming they are shows up months later, in a program that was diagnosed correctly against the brief and still missed the business problem — because nobody ever checked the brief against the person actually living it.